diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,78318 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 11184,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.532864086188304,
+      "learning_rate": 1.1904761904761906e-07,
+      "loss": 1.1655,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.944273669298862,
+      "learning_rate": 2.3809523809523811e-07,
+      "loss": 1.2205,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.469496186016453,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": 1.2554,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.867958079502744,
+      "learning_rate": 4.7619047619047623e-07,
+      "loss": 1.2875,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.140723090413525,
+      "learning_rate": 5.952380952380953e-07,
+      "loss": 1.1761,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.615748101162847,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 1.2246,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.212336122935811,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.2101,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.464571140448921,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 1.1208,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.609388505322048,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 1.1655,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.756385590531069,
+      "learning_rate": 1.1904761904761906e-06,
+      "loss": 1.0375,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.605522517525138,
+      "learning_rate": 1.3095238095238096e-06,
+      "loss": 1.064,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.741460192238878,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 1.113,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.09116378972063,
+      "learning_rate": 1.5476190476190479e-06,
+      "loss": 0.9726,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8756625649492962,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 0.9575,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6829109778539475,
+      "learning_rate": 1.7857142857142859e-06,
+      "loss": 0.9765,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6530624134866179,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 0.9803,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7426516018791256,
+      "learning_rate": 2.023809523809524e-06,
+      "loss": 1.0163,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2855121752972152,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 0.9308,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5905687459869657,
+      "learning_rate": 2.261904761904762e-06,
+      "loss": 1.0064,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3781637014306847,
+      "learning_rate": 2.380952380952381e-06,
+      "loss": 0.8882,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3071730544614337,
+      "learning_rate": 2.5e-06,
+      "loss": 0.9055,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.15710890702201,
+      "learning_rate": 2.6190476190476192e-06,
+      "loss": 0.9527,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9351287247809734,
+      "learning_rate": 2.7380952380952387e-06,
+      "loss": 0.8737,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9022707513126496,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.926,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9128865245709114,
+      "learning_rate": 2.9761904761904763e-06,
+      "loss": 0.9071,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0783896028603837,
+      "learning_rate": 3.0952380952380957e-06,
+      "loss": 0.9362,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1835560401686442,
+      "learning_rate": 3.2142857142857147e-06,
+      "loss": 0.9289,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.264410526743412,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.9207,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.16697145634602,
+      "learning_rate": 3.4523809523809528e-06,
+      "loss": 0.9051,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.936458759692709,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 0.9034,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7445199758612073,
+      "learning_rate": 3.690476190476191e-06,
+      "loss": 0.8734,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9696956808071493,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 0.8751,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9909057129982866,
+      "learning_rate": 3.928571428571429e-06,
+      "loss": 0.9189,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1988867861319965,
+      "learning_rate": 4.047619047619048e-06,
+      "loss": 0.9043,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1086870540671567,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.8924,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8652296384254909,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.8893,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.754050863917242,
+      "learning_rate": 4.404761904761905e-06,
+      "loss": 0.862,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7005911541865115,
+      "learning_rate": 4.523809523809524e-06,
+      "loss": 0.864,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7467940302836517,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 0.8604,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7065865304351598,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 0.8234,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8617594093337063,
+      "learning_rate": 4.880952380952381e-06,
+      "loss": 0.8885,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8354683099430769,
+      "learning_rate": 5e-06,
+      "loss": 0.8662,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6750476944874323,
+      "learning_rate": 5.119047619047619e-06,
+      "loss": 0.8222,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7194394013061935,
+      "learning_rate": 5.2380952380952384e-06,
+      "loss": 0.8525,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8569459385640346,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.8898,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9408159417744165,
+      "learning_rate": 5.476190476190477e-06,
+      "loss": 0.7984,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7992705092496207,
+      "learning_rate": 5.595238095238096e-06,
+      "loss": 0.8194,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7129259371323274,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.8225,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7395977161363956,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 0.813,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7512348458605638,
+      "learning_rate": 5.9523809523809525e-06,
+      "loss": 0.8681,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6726463446368257,
+      "learning_rate": 6.071428571428571e-06,
+      "loss": 0.8234,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6921492758439253,
+      "learning_rate": 6.1904761904761914e-06,
+      "loss": 0.789,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6921981884240925,
+      "learning_rate": 6.30952380952381e-06,
+      "loss": 0.794,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.6119753861905523,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 0.841,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.692532657696066,
+      "learning_rate": 6.547619047619048e-06,
+      "loss": 0.8588,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7779259286901284,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.8028,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6607734022405941,
+      "learning_rate": 6.785714285714287e-06,
+      "loss": 0.8042,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.739263019772448,
+      "learning_rate": 6.9047619047619055e-06,
+      "loss": 0.8727,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6739565722205683,
+      "learning_rate": 7.023809523809524e-06,
+      "loss": 0.8095,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7379642111851756,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.8318,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7552279225325318,
+      "learning_rate": 7.261904761904762e-06,
+      "loss": 0.8103,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6723721211303265,
+      "learning_rate": 7.380952380952382e-06,
+      "loss": 0.8309,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7046798143689864,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.8057,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7327105746053033,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 0.653,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.667453066331941,
+      "learning_rate": 7.738095238095238e-06,
+      "loss": 0.8469,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6889865051153221,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.8265,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6056406269305498,
+      "learning_rate": 7.976190476190477e-06,
+      "loss": 0.7639,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.647141510133748,
+      "learning_rate": 8.095238095238097e-06,
+      "loss": 0.7883,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.686875733580455,
+      "learning_rate": 8.214285714285714e-06,
+      "loss": 0.8148,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6340441941351969,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.7864,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6006918678742399,
+      "learning_rate": 8.452380952380953e-06,
+      "loss": 0.7544,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6753138634670008,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.7843,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6660042967074176,
+      "learning_rate": 8.690476190476192e-06,
+      "loss": 0.803,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6480197150024184,
+      "learning_rate": 8.80952380952381e-06,
+      "loss": 0.7857,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6187024650522432,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.7762,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5695367471898216,
+      "learning_rate": 9.047619047619049e-06,
+      "loss": 0.7578,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7123295291125451,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 0.7582,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6796943699025415,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 0.7442,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7927345724915081,
+      "learning_rate": 9.404761904761905e-06,
+      "loss": 0.8098,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7473189891282259,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 0.8383,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.768656259955626,
+      "learning_rate": 9.642857142857144e-06,
+      "loss": 0.7792,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.62381545802169,
+      "learning_rate": 9.761904761904762e-06,
+      "loss": 0.7642,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8550297248163874,
+      "learning_rate": 9.880952380952381e-06,
+      "loss": 0.7925,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6636016972038153,
+      "learning_rate": 1e-05,
+      "loss": 0.7711,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6144322945650744,
+      "learning_rate": 1.011904761904762e-05,
+      "loss": 0.7562,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8138477216672986,
+      "learning_rate": 1.0238095238095238e-05,
+      "loss": 0.7817,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9466586596834066,
+      "learning_rate": 1.0357142857142859e-05,
+      "loss": 0.5968,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7685049458288253,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 0.7858,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7170809472318564,
+      "learning_rate": 1.0595238095238096e-05,
+      "loss": 0.7604,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7730513650298215,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.7933,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8641993847450423,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 0.8207,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7074469379473086,
+      "learning_rate": 1.0952380952380955e-05,
+      "loss": 0.7799,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8617233708268109,
+      "learning_rate": 1.1071428571428572e-05,
+      "loss": 0.7741,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7093923805988073,
+      "learning_rate": 1.1190476190476192e-05,
+      "loss": 0.7477,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7304264612007426,
+      "learning_rate": 1.130952380952381e-05,
+      "loss": 0.8068,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8385807744947523,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.7804,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7356751503890707,
+      "learning_rate": 1.1547619047619047e-05,
+      "loss": 0.7921,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6940669865113623,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.7798,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6832400727754339,
+      "learning_rate": 1.1785714285714287e-05,
+      "loss": 0.7534,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7166241035201556,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 0.7957,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5895311714507747,
+      "learning_rate": 1.2023809523809525e-05,
+      "loss": 0.756,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6161045706185114,
+      "learning_rate": 1.2142857142857142e-05,
+      "loss": 0.7422,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5536085456687476,
+      "learning_rate": 1.2261904761904763e-05,
+      "loss": 0.7948,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6065550488333646,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 0.7805,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6404449468802775,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7749,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6802193780586949,
+      "learning_rate": 1.261904761904762e-05,
+      "loss": 0.8054,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6353340336407852,
+      "learning_rate": 1.2738095238095238e-05,
+      "loss": 0.7587,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6209059464872395,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.7735,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6215270192983944,
+      "learning_rate": 1.2976190476190478e-05,
+      "loss": 0.7517,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5691723083382634,
+      "learning_rate": 1.3095238095238096e-05,
+      "loss": 0.7641,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6174397868288832,
+      "learning_rate": 1.3214285714285716e-05,
+      "loss": 0.7578,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.63588799440003,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7549,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5960229431996272,
+      "learning_rate": 1.3452380952380954e-05,
+      "loss": 0.7724,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.677604721098971,
+      "learning_rate": 1.3571428571428574e-05,
+      "loss": 0.8258,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6425623895664552,
+      "learning_rate": 1.3690476190476192e-05,
+      "loss": 0.7324,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6312288619876921,
+      "learning_rate": 1.3809523809523811e-05,
+      "loss": 0.7504,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7102748517430989,
+      "learning_rate": 1.3928571428571429e-05,
+      "loss": 0.7497,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3845554201554648,
+      "learning_rate": 1.4047619047619048e-05,
+      "loss": 0.7895,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1758127777405933,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 0.7875,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.025663350213421,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.7638,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0082204187958907,
+      "learning_rate": 1.4404761904761907e-05,
+      "loss": 0.7704,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9451505559078778,
+      "learning_rate": 1.4523809523809524e-05,
+      "loss": 0.7787,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6494362338702195,
+      "learning_rate": 1.4642857142857144e-05,
+      "loss": 0.7606,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7987353749969532,
+      "learning_rate": 1.4761904761904763e-05,
+      "loss": 0.7886,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7152454418584201,
+      "learning_rate": 1.4880952380952383e-05,
+      "loss": 0.7996,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7316600301128386,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7957,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.756133677040311,
+      "learning_rate": 1.511904761904762e-05,
+      "loss": 0.7456,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6691241112745875,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 0.7733,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6895385597068372,
+      "learning_rate": 1.535714285714286e-05,
+      "loss": 0.7644,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5208383704639331,
+      "learning_rate": 1.5476190476190476e-05,
+      "loss": 0.7329,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5732260367633706,
+      "learning_rate": 1.5595238095238098e-05,
+      "loss": 0.7566,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.698404069331571,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.7312,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6435899459661656,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 0.7724,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5777785237261925,
+      "learning_rate": 1.5952380952380954e-05,
+      "loss": 0.7287,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6687409925607796,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.7916,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6665169644826535,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 0.8025,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5575158965296348,
+      "learning_rate": 1.630952380952381e-05,
+      "loss": 0.7024,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7229230364790548,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 0.6353,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6802655404657898,
+      "learning_rate": 1.6547619047619046e-05,
+      "loss": 0.7644,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6198844150508233,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.7322,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5895433611538993,
+      "learning_rate": 1.678571428571429e-05,
+      "loss": 0.7853,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5688371941281741,
+      "learning_rate": 1.6904761904761906e-05,
+      "loss": 0.7541,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6505721176960555,
+      "learning_rate": 1.7023809523809524e-05,
+      "loss": 0.7498,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6546931685057096,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.7908,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5660449841981595,
+      "learning_rate": 1.7261904761904763e-05,
+      "loss": 0.7628,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5937038934609243,
+      "learning_rate": 1.7380952380952384e-05,
+      "loss": 0.8009,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.615635167314993,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.7903,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5474136840938296,
+      "learning_rate": 1.761904761904762e-05,
+      "loss": 0.7288,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5837686062102428,
+      "learning_rate": 1.7738095238095237e-05,
+      "loss": 0.7755,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5930838259360622,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.732,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5678351818668746,
+      "learning_rate": 1.797619047619048e-05,
+      "loss": 0.7349,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5812882004708979,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 0.7814,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6365998482621859,
+      "learning_rate": 1.8214285714285715e-05,
+      "loss": 0.7966,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5728612508328779,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.761,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6089886372518973,
+      "learning_rate": 1.8452380952380954e-05,
+      "loss": 0.7428,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5714291428534173,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.7678,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5992182867176703,
+      "learning_rate": 1.8690476190476193e-05,
+      "loss": 0.7839,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5707452081190949,
+      "learning_rate": 1.880952380952381e-05,
+      "loss": 0.759,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6268774783430948,
+      "learning_rate": 1.892857142857143e-05,
+      "loss": 0.7253,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6659135314322924,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 0.7158,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5931888917669059,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 0.7429,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5390127385883621,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.765,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5068343934644247,
+      "learning_rate": 1.9404761904761906e-05,
+      "loss": 0.7124,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.632741517455677,
+      "learning_rate": 1.9523809523809524e-05,
+      "loss": 0.7563,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6466107179171507,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.5895,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6094108727118128,
+      "learning_rate": 1.9761904761904763e-05,
+      "loss": 0.7657,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5955814539307702,
+      "learning_rate": 1.9880952380952384e-05,
+      "loss": 0.7063,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6582451769774909,
+      "learning_rate": 2e-05,
+      "loss": 0.7754,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6268424952094938,
+      "learning_rate": 2.011904761904762e-05,
+      "loss": 0.7544,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6122265545718524,
+      "learning_rate": 2.023809523809524e-05,
+      "loss": 0.7788,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5750883170916328,
+      "learning_rate": 2.0357142857142858e-05,
+      "loss": 0.7874,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6400700779733298,
+      "learning_rate": 2.0476190476190476e-05,
+      "loss": 0.7435,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6450467740545872,
+      "learning_rate": 2.0595238095238094e-05,
+      "loss": 0.7527,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6071685498510763,
+      "learning_rate": 2.0714285714285718e-05,
+      "loss": 0.7813,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6458903105087137,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.7447,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5649367709509586,
+      "learning_rate": 2.0952380952380954e-05,
+      "loss": 0.7815,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6159274503763696,
+      "learning_rate": 2.1071428571428575e-05,
+      "loss": 0.7296,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5664313664795827,
+      "learning_rate": 2.1190476190476193e-05,
+      "loss": 0.7623,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7305912380111351,
+      "learning_rate": 2.130952380952381e-05,
+      "loss": 0.8262,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5853478461727479,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.7629,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5614958337140147,
+      "learning_rate": 2.154761904761905e-05,
+      "loss": 0.7077,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6612063746505116,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 0.726,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5831533619176446,
+      "learning_rate": 2.1785714285714285e-05,
+      "loss": 0.75,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5863723260580967,
+      "learning_rate": 2.190476190476191e-05,
+      "loss": 0.7465,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6030115370197878,
+      "learning_rate": 2.2023809523809527e-05,
+      "loss": 0.7843,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5102965780529193,
+      "learning_rate": 2.2142857142857145e-05,
+      "loss": 0.7339,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5943667408571197,
+      "learning_rate": 2.2261904761904766e-05,
+      "loss": 0.7592,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5802312874531076,
+      "learning_rate": 2.2380952380952384e-05,
+      "loss": 0.7697,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.741240882409294,
+      "learning_rate": 2.25e-05,
+      "loss": 0.759,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6285771872370397,
+      "learning_rate": 2.261904761904762e-05,
+      "loss": 0.7556,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5651821814981262,
+      "learning_rate": 2.273809523809524e-05,
+      "loss": 0.7297,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6319931590611245,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 0.7438,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6470415625001963,
+      "learning_rate": 2.2976190476190476e-05,
+      "loss": 0.7587,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5931742899128494,
+      "learning_rate": 2.3095238095238094e-05,
+      "loss": 0.7674,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6483393073549113,
+      "learning_rate": 2.3214285714285718e-05,
+      "loss": 0.7879,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5620149266526098,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.7567,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5464530890000893,
+      "learning_rate": 2.3452380952380957e-05,
+      "loss": 0.774,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6423392792888704,
+      "learning_rate": 2.3571428571428575e-05,
+      "loss": 0.7665,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6075496680659802,
+      "learning_rate": 2.3690476190476192e-05,
+      "loss": 0.7472,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6594663740473954,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 0.7989,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5725946728158854,
+      "learning_rate": 2.392857142857143e-05,
+      "loss": 0.7982,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.584271274108707,
+      "learning_rate": 2.404761904761905e-05,
+      "loss": 0.7568,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6809363267648646,
+      "learning_rate": 2.4166666666666667e-05,
+      "loss": 0.7391,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7098218845986971,
+      "learning_rate": 2.4285714285714285e-05,
+      "loss": 0.7584,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5427770097890682,
+      "learning_rate": 2.440476190476191e-05,
+      "loss": 0.7323,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6510645333743412,
+      "learning_rate": 2.4523809523809527e-05,
+      "loss": 0.7396,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6251204092229138,
+      "learning_rate": 2.4642857142857145e-05,
+      "loss": 0.7465,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6441944613693724,
+      "learning_rate": 2.4761904761904766e-05,
+      "loss": 0.7724,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9326146884290419,
+      "learning_rate": 2.4880952380952383e-05,
+      "loss": 0.7681,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5919289417362279,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7268,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6046174119484848,
+      "learning_rate": 2.5119047619047622e-05,
+      "loss": 0.7813,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6883320402450621,
+      "learning_rate": 2.523809523809524e-05,
+      "loss": 0.7453,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5953516596120827,
+      "learning_rate": 2.5357142857142858e-05,
+      "loss": 0.7503,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6680260936471486,
+      "learning_rate": 2.5476190476190476e-05,
+      "loss": 0.7952,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5957925922282522,
+      "learning_rate": 2.5595238095238097e-05,
+      "loss": 0.7278,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5622382027003635,
+      "learning_rate": 2.5714285714285718e-05,
+      "loss": 0.7361,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6463759911275861,
+      "learning_rate": 2.5833333333333336e-05,
+      "loss": 0.7846,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5041392643828205,
+      "learning_rate": 2.5952380952380957e-05,
+      "loss": 0.7514,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5516427189205614,
+      "learning_rate": 2.6071428571428574e-05,
+      "loss": 0.7399,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5733511259743098,
+      "learning_rate": 2.6190476190476192e-05,
+      "loss": 0.7535,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.530207288905465,
+      "learning_rate": 2.630952380952381e-05,
+      "loss": 0.767,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5671482522011778,
+      "learning_rate": 2.642857142857143e-05,
+      "loss": 0.771,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5328465086225305,
+      "learning_rate": 2.654761904761905e-05,
+      "loss": 0.7425,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5943130115819673,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.7101,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5369044704940678,
+      "learning_rate": 2.6785714285714284e-05,
+      "loss": 0.7226,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6097601859882652,
+      "learning_rate": 2.690476190476191e-05,
+      "loss": 0.7257,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5711916281986175,
+      "learning_rate": 2.7023809523809527e-05,
+      "loss": 0.7598,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5422918037466427,
+      "learning_rate": 2.7142857142857148e-05,
+      "loss": 0.7294,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5007248341615239,
+      "learning_rate": 2.7261904761904765e-05,
+      "loss": 0.7148,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5575997237382395,
+      "learning_rate": 2.7380952380952383e-05,
+      "loss": 0.7633,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5186275684869287,
+      "learning_rate": 2.75e-05,
+      "loss": 0.682,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5534312870375429,
+      "learning_rate": 2.7619047619047622e-05,
+      "loss": 0.7751,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5534181012476704,
+      "learning_rate": 2.773809523809524e-05,
+      "loss": 0.7502,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5365957870585824,
+      "learning_rate": 2.7857142857142858e-05,
+      "loss": 0.7525,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5539416497573924,
+      "learning_rate": 2.7976190476190475e-05,
+      "loss": 0.7445,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5599380911897965,
+      "learning_rate": 2.8095238095238096e-05,
+      "loss": 0.7579,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5916056820639968,
+      "learning_rate": 2.8214285714285718e-05,
+      "loss": 0.7848,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5820871966674261,
+      "learning_rate": 2.833333333333334e-05,
+      "loss": 0.7717,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5969368184849709,
+      "learning_rate": 2.8452380952380957e-05,
+      "loss": 0.7469,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5417299036574983,
+      "learning_rate": 2.8571428571428574e-05,
+      "loss": 0.7737,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5671042607745205,
+      "learning_rate": 2.8690476190476192e-05,
+      "loss": 0.7236,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.545161817261602,
+      "learning_rate": 2.8809523809523813e-05,
+      "loss": 0.735,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.45438614241272357,
+      "learning_rate": 2.892857142857143e-05,
+      "loss": 0.7335,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5843024555570739,
+      "learning_rate": 2.904761904761905e-05,
+      "loss": 0.7614,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5511853312859214,
+      "learning_rate": 2.9166666666666666e-05,
+      "loss": 0.7489,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5943775319443511,
+      "learning_rate": 2.9285714285714288e-05,
+      "loss": 0.7555,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.542968325218305,
+      "learning_rate": 2.940476190476191e-05,
+      "loss": 0.7631,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6040353087270796,
+      "learning_rate": 2.9523809523809526e-05,
+      "loss": 0.7528,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5786082474634588,
+      "learning_rate": 2.9642857142857148e-05,
+      "loss": 0.7602,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6534829152432731,
+      "learning_rate": 2.9761904761904765e-05,
+      "loss": 0.7878,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5730421628464214,
+      "learning_rate": 2.9880952380952383e-05,
+      "loss": 0.7532,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5888201131738682,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.7182,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5810499086232593,
+      "learning_rate": 3.0119047619047622e-05,
+      "loss": 0.7196,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5766410579762363,
+      "learning_rate": 3.023809523809524e-05,
+      "loss": 0.7264,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.638141996731187,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.7256,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.611474366760603,
+      "learning_rate": 3.047619047619048e-05,
+      "loss": 0.7255,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5727669042636377,
+      "learning_rate": 3.059523809523809e-05,
+      "loss": 0.7421,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6398999811149553,
+      "learning_rate": 3.071428571428572e-05,
+      "loss": 0.7935,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5476864506951684,
+      "learning_rate": 3.0833333333333335e-05,
+      "loss": 0.7779,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7268466302997726,
+      "learning_rate": 3.095238095238095e-05,
+      "loss": 0.7946,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5033306207312507,
+      "learning_rate": 3.107142857142858e-05,
+      "loss": 0.6805,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5846092614146331,
+      "learning_rate": 3.1190476190476195e-05,
+      "loss": 0.5779,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.556776372397354,
+      "learning_rate": 3.130952380952381e-05,
+      "loss": 0.7582,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.589672400159479,
+      "learning_rate": 3.142857142857143e-05,
+      "loss": 0.7976,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5502264269359317,
+      "learning_rate": 3.154761904761905e-05,
+      "loss": 0.738,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5723571557225681,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.7575,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5379290876049996,
+      "learning_rate": 3.1785714285714284e-05,
+      "loss": 0.7382,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5731835092983499,
+      "learning_rate": 3.190476190476191e-05,
+      "loss": 0.7452,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5733561950027842,
+      "learning_rate": 3.2023809523809526e-05,
+      "loss": 0.7924,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5120812821987489,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.7327,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5120437113546159,
+      "learning_rate": 3.226190476190477e-05,
+      "loss": 0.7306,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5340870113113855,
+      "learning_rate": 3.2380952380952386e-05,
+      "loss": 0.7405,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5704653306569644,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.7349,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.538714926406352,
+      "learning_rate": 3.261904761904762e-05,
+      "loss": 0.7445,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5564353304281775,
+      "learning_rate": 3.273809523809524e-05,
+      "loss": 0.7832,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5129678974601757,
+      "learning_rate": 3.285714285714286e-05,
+      "loss": 0.7498,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5406164772387339,
+      "learning_rate": 3.2976190476190475e-05,
+      "loss": 0.611,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6647357088071517,
+      "learning_rate": 3.309523809523809e-05,
+      "loss": 0.7314,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5174607263548913,
+      "learning_rate": 3.321428571428572e-05,
+      "loss": 0.7246,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6188301181417957,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.733,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5417287914527605,
+      "learning_rate": 3.345238095238096e-05,
+      "loss": 0.731,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6409352098375853,
+      "learning_rate": 3.357142857142858e-05,
+      "loss": 0.7881,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45863714275053236,
+      "learning_rate": 3.3690476190476195e-05,
+      "loss": 0.5653,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1673157670602157,
+      "learning_rate": 3.380952380952381e-05,
+      "loss": 0.7654,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6416320917778324,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.7423,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7393370036611505,
+      "learning_rate": 3.404761904761905e-05,
+      "loss": 0.7512,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6311203256024884,
+      "learning_rate": 3.4166666666666666e-05,
+      "loss": 0.7315,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7337089080981514,
+      "learning_rate": 3.4285714285714284e-05,
+      "loss": 0.7158,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6631517088125097,
+      "learning_rate": 3.440476190476191e-05,
+      "loss": 0.7904,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7930605195743535,
+      "learning_rate": 3.4523809523809526e-05,
+      "loss": 0.7377,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5449220045910447,
+      "learning_rate": 3.4642857142857144e-05,
+      "loss": 0.7809,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5206181872547706,
+      "learning_rate": 3.476190476190477e-05,
+      "loss": 0.7573,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.509827139322677,
+      "learning_rate": 3.4880952380952386e-05,
+      "loss": 0.7669,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.522233514284411,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 0.7691,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5167562648093539,
+      "learning_rate": 3.511904761904762e-05,
+      "loss": 0.7445,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5386247397602718,
+      "learning_rate": 3.523809523809524e-05,
+      "loss": 0.776,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5490724905383472,
+      "learning_rate": 3.535714285714286e-05,
+      "loss": 0.7786,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5628479709671907,
+      "learning_rate": 3.5476190476190475e-05,
+      "loss": 0.7585,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5415014142115584,
+      "learning_rate": 3.55952380952381e-05,
+      "loss": 0.7546,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.563080232964059,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.7457,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5459907627175489,
+      "learning_rate": 3.5833333333333335e-05,
+      "loss": 0.7682,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4869429024389171,
+      "learning_rate": 3.595238095238096e-05,
+      "loss": 0.7305,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5069110544680069,
+      "learning_rate": 3.607142857142858e-05,
+      "loss": 0.7456,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5796898482018323,
+      "learning_rate": 3.6190476190476195e-05,
+      "loss": 0.725,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.560095911880239,
+      "learning_rate": 3.630952380952381e-05,
+      "loss": 0.8047,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5054370492248484,
+      "learning_rate": 3.642857142857143e-05,
+      "loss": 0.7417,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5647380697280561,
+      "learning_rate": 3.654761904761905e-05,
+      "loss": 0.753,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5389262085927387,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.5915,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5187666702490271,
+      "learning_rate": 3.6785714285714283e-05,
+      "loss": 0.7178,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6381298996049699,
+      "learning_rate": 3.690476190476191e-05,
+      "loss": 0.7509,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5081929158680163,
+      "learning_rate": 3.7023809523809526e-05,
+      "loss": 0.7664,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4879552185064864,
+      "learning_rate": 3.714285714285715e-05,
+      "loss": 0.7155,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47473507454292474,
+      "learning_rate": 3.726190476190477e-05,
+      "loss": 0.7624,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5069938050918781,
+      "learning_rate": 3.7380952380952386e-05,
+      "loss": 0.7341,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5402492990834894,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.7561,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5172345168516695,
+      "learning_rate": 3.761904761904762e-05,
+      "loss": 0.7257,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4978229904294143,
+      "learning_rate": 3.773809523809524e-05,
+      "loss": 0.706,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.662456760187022,
+      "learning_rate": 3.785714285714286e-05,
+      "loss": 0.7099,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4631301414155739,
+      "learning_rate": 3.7976190476190474e-05,
+      "loss": 0.7149,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5141306004886866,
+      "learning_rate": 3.80952380952381e-05,
+      "loss": 0.7429,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4992728497434938,
+      "learning_rate": 3.821428571428572e-05,
+      "loss": 0.744,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5862039924538878,
+      "learning_rate": 3.833333333333334e-05,
+      "loss": 0.7714,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5537774732029892,
+      "learning_rate": 3.845238095238096e-05,
+      "loss": 0.7334,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.51371631555111,
+      "learning_rate": 3.857142857142858e-05,
+      "loss": 0.7287,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5224113591725343,
+      "learning_rate": 3.8690476190476195e-05,
+      "loss": 0.7306,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5771131545590683,
+      "learning_rate": 3.880952380952381e-05,
+      "loss": 0.7463,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5637690563868413,
+      "learning_rate": 3.892857142857143e-05,
+      "loss": 0.758,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5372980887897281,
+      "learning_rate": 3.904761904761905e-05,
+      "loss": 0.7567,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5726113759150327,
+      "learning_rate": 3.9166666666666665e-05,
+      "loss": 0.742,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5625813961098711,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.7644,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5301049446568558,
+      "learning_rate": 3.940476190476191e-05,
+      "loss": 0.7528,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49739863320816274,
+      "learning_rate": 3.9523809523809526e-05,
+      "loss": 0.7422,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4914234489670035,
+      "learning_rate": 3.964285714285715e-05,
+      "loss": 0.7565,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5903246345221899,
+      "learning_rate": 3.976190476190477e-05,
+      "loss": 0.752,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6933853420565607,
+      "learning_rate": 3.9880952380952386e-05,
+      "loss": 0.7633,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5823532332358063,
+      "learning_rate": 4e-05,
+      "loss": 0.7429,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5314087287880989,
+      "learning_rate": 3.999999916131208e-05,
+      "loss": 0.7405,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6688647704094244,
+      "learning_rate": 3.999999664524838e-05,
+      "loss": 0.7134,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5873417470645471,
+      "learning_rate": 3.999999245180912e-05,
+      "loss": 0.7797,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.589278355337866,
+      "learning_rate": 3.999998658099464e-05,
+      "loss": 0.7738,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5432008633917671,
+      "learning_rate": 3.999997903280544e-05,
+      "loss": 0.723,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4547934828813513,
+      "learning_rate": 3.999996980724216e-05,
+      "loss": 0.74,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6143289555043561,
+      "learning_rate": 3.999995890430556e-05,
+      "loss": 0.7623,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5208995295525581,
+      "learning_rate": 3.999994632399656e-05,
+      "loss": 0.7619,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.538810980189716,
+      "learning_rate": 3.999993206631622e-05,
+      "loss": 0.743,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5802304278378508,
+      "learning_rate": 3.9999916131265726e-05,
+      "loss": 0.7839,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5306884968022325,
+      "learning_rate": 3.9999898518846424e-05,
+      "loss": 0.7791,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5785394942287962,
+      "learning_rate": 3.9999879229059785e-05,
+      "loss": 0.7693,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5335561184859662,
+      "learning_rate": 3.999985826190743e-05,
+      "loss": 0.7491,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5399889668347386,
+      "learning_rate": 3.999983561739112e-05,
+      "loss": 0.7236,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5154434004774399,
+      "learning_rate": 3.999981129551274e-05,
+      "loss": 0.7644,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6201262836765419,
+      "learning_rate": 3.999978529627435e-05,
+      "loss": 0.7768,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5820655321770196,
+      "learning_rate": 3.9999757619678115e-05,
+      "loss": 0.7193,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5257559770330158,
+      "learning_rate": 3.9999728265726364e-05,
+      "loss": 0.7316,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.542293515303263,
+      "learning_rate": 3.999969723442155e-05,
+      "loss": 0.7882,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.564584340592624,
+      "learning_rate": 3.99996645257663e-05,
+      "loss": 0.746,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5199657229490499,
+      "learning_rate": 3.999963013976333e-05,
+      "loss": 0.7431,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46071494864752566,
+      "learning_rate": 3.999959407641553e-05,
+      "loss": 0.7812,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49559500932447687,
+      "learning_rate": 3.999955633572593e-05,
+      "loss": 0.7868,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.43911185834075045,
+      "learning_rate": 3.999951691769769e-05,
+      "loss": 0.7342,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5457658239962435,
+      "learning_rate": 3.999947582233413e-05,
+      "loss": 0.7893,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48662410099809084,
+      "learning_rate": 3.999943304963868e-05,
+      "loss": 0.736,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5195562013377248,
+      "learning_rate": 3.999938859961493e-05,
+      "loss": 0.762,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46189905622260136,
+      "learning_rate": 3.999934247226662e-05,
+      "loss": 0.7534,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5001915993358932,
+      "learning_rate": 3.99992946675976e-05,
+      "loss": 0.7498,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48567975543512937,
+      "learning_rate": 3.99992451856119e-05,
+      "loss": 0.7562,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5293496915509556,
+      "learning_rate": 3.999919402631365e-05,
+      "loss": 0.7251,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4820271523571588,
+      "learning_rate": 3.999914118970715e-05,
+      "loss": 0.7115,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5162547235696885,
+      "learning_rate": 3.999908667579684e-05,
+      "loss": 0.7232,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49220379573536477,
+      "learning_rate": 3.9999030484587274e-05,
+      "loss": 0.7665,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5174390889717441,
+      "learning_rate": 3.999897261608318e-05,
+      "loss": 0.7936,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5241671924746567,
+      "learning_rate": 3.99989130702894e-05,
+      "loss": 0.7663,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5664819683887483,
+      "learning_rate": 3.999885184721094e-05,
+      "loss": 0.7359,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5414043669621265,
+      "learning_rate": 3.999878894685292e-05,
+      "loss": 0.735,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4819119080862045,
+      "learning_rate": 3.999872436922063e-05,
+      "loss": 0.71,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4742963851958092,
+      "learning_rate": 3.9998658114319475e-05,
+      "loss": 0.7619,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5225364786421732,
+      "learning_rate": 3.9998590182155024e-05,
+      "loss": 0.7779,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5230852732890423,
+      "learning_rate": 3.999852057273296e-05,
+      "loss": 0.7304,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5227145211461718,
+      "learning_rate": 3.999844928605913e-05,
+      "loss": 0.7497,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5051387524973007,
+      "learning_rate": 3.9998376322139506e-05,
+      "loss": 0.7295,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.488686948127181,
+      "learning_rate": 3.999830168098021e-05,
+      "loss": 0.7302,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5600050146157606,
+      "learning_rate": 3.999822536258752e-05,
+      "loss": 0.736,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49358520731972294,
+      "learning_rate": 3.99981473669678e-05,
+      "loss": 0.7608,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49619703943567767,
+      "learning_rate": 3.9998067694127626e-05,
+      "loss": 0.7606,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48695179268297734,
+      "learning_rate": 3.999798634407366e-05,
+      "loss": 0.7209,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4886419068857495,
+      "learning_rate": 3.999790331681273e-05,
+      "loss": 0.7583,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4826092008354928,
+      "learning_rate": 3.9997818612351804e-05,
+      "loss": 0.8089,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5184753232172953,
+      "learning_rate": 3.999773223069798e-05,
+      "loss": 0.7581,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7127319597430397,
+      "learning_rate": 3.9997644171858504e-05,
+      "loss": 0.6284,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5860508213910678,
+      "learning_rate": 3.999755443584077e-05,
+      "loss": 0.7929,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5032029131479192,
+      "learning_rate": 3.999746302265228e-05,
+      "loss": 0.7286,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49271940007396975,
+      "learning_rate": 3.999736993230073e-05,
+      "loss": 0.7414,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49433565079726693,
+      "learning_rate": 3.9997275164793915e-05,
+      "loss": 0.7486,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6342479002610476,
+      "learning_rate": 3.9997178720139785e-05,
+      "loss": 0.7411,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.48580595455005543,
+      "learning_rate": 3.9997080598346414e-05,
+      "loss": 0.7578,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6281660323853705,
+      "learning_rate": 3.9996980799422054e-05,
+      "loss": 0.5836,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5373522035703616,
+      "learning_rate": 3.999687932337506e-05,
+      "loss": 0.7615,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4776483680599048,
+      "learning_rate": 3.9996776170213945e-05,
+      "loss": 0.7598,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5356124558077655,
+      "learning_rate": 3.9996671339947366e-05,
+      "loss": 0.7369,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49053066187648053,
+      "learning_rate": 3.9996564832584114e-05,
+      "loss": 0.7442,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4905810216054048,
+      "learning_rate": 3.999645664813311e-05,
+      "loss": 0.7106,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5110355160153056,
+      "learning_rate": 3.999634678660344e-05,
+      "loss": 0.7491,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.449191222115474,
+      "learning_rate": 3.9996235248004326e-05,
+      "loss": 0.7465,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.486215988476232,
+      "learning_rate": 3.999612203234511e-05,
+      "loss": 0.7829,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4573956537542962,
+      "learning_rate": 3.999600713963527e-05,
+      "loss": 0.7444,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5391940027105249,
+      "learning_rate": 3.999589056988447e-05,
+      "loss": 0.763,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5766458175579481,
+      "learning_rate": 3.9995772323102484e-05,
+      "loss": 0.7901,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.485125317802673,
+      "learning_rate": 3.9995652399299215e-05,
+      "loss": 0.7259,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4908417028516081,
+      "learning_rate": 3.999553079848473e-05,
+      "loss": 0.7026,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4806078758008247,
+      "learning_rate": 3.999540752066923e-05,
+      "loss": 0.7475,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4960805024860488,
+      "learning_rate": 3.9995282565863034e-05,
+      "loss": 0.7431,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4268189805426852,
+      "learning_rate": 3.999515593407665e-05,
+      "loss": 0.7162,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.460399327223398,
+      "learning_rate": 3.9995027625320684e-05,
+      "loss": 0.789,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5428838724597804,
+      "learning_rate": 3.99948976396059e-05,
+      "loss": 0.7653,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4494055560443093,
+      "learning_rate": 3.999476597694319e-05,
+      "loss": 0.7554,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5553718294874996,
+      "learning_rate": 3.999463263734361e-05,
+      "loss": 0.7308,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5982754094682857,
+      "learning_rate": 3.999449762081835e-05,
+      "loss": 0.7238,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5862287074835592,
+      "learning_rate": 3.9994360927378706e-05,
+      "loss": 0.8153,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45167126730861723,
+      "learning_rate": 3.9994222557036165e-05,
+      "loss": 0.7124,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5124684394090839,
+      "learning_rate": 3.9994082509802325e-05,
+      "loss": 0.763,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42771846128691526,
+      "learning_rate": 3.999394078568893e-05,
+      "loss": 0.737,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5053175348584915,
+      "learning_rate": 3.9993797384707865e-05,
+      "loss": 0.7555,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4523975999901704,
+      "learning_rate": 3.9993652306871165e-05,
+      "loss": 0.7163,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49906597196996155,
+      "learning_rate": 3.999350555219099e-05,
+      "loss": 0.7189,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.487990421238943,
+      "learning_rate": 3.9993357120679655e-05,
+      "loss": 0.7674,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.461944686599269,
+      "learning_rate": 3.999320701234959e-05,
+      "loss": 0.698,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4557716951947821,
+      "learning_rate": 3.9993055227213415e-05,
+      "loss": 0.7172,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.463976095703579,
+      "learning_rate": 3.9992901765283836e-05,
+      "loss": 0.7407,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5251122194592373,
+      "learning_rate": 3.999274662657374e-05,
+      "loss": 0.7595,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6875016230801866,
+      "learning_rate": 3.9992589811096115e-05,
+      "loss": 0.7719,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5148375746018641,
+      "learning_rate": 3.999243131886413e-05,
+      "loss": 0.7605,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.487356106387141,
+      "learning_rate": 3.9992271149891086e-05,
+      "loss": 0.7359,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4520600185097253,
+      "learning_rate": 3.999210930419039e-05,
+      "loss": 0.7163,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49778108190732934,
+      "learning_rate": 3.999194578177565e-05,
+      "loss": 0.7311,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4754697605012462,
+      "learning_rate": 3.9991780582660545e-05,
+      "loss": 0.752,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46566986332980537,
+      "learning_rate": 3.9991613706858956e-05,
+      "loss": 0.7074,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46966922832823366,
+      "learning_rate": 3.999144515438486e-05,
+      "loss": 0.7415,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4855294520615354,
+      "learning_rate": 3.999127492525241e-05,
+      "loss": 0.7373,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4690439082816576,
+      "learning_rate": 3.9991103019475875e-05,
+      "loss": 0.7461,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44537461366057335,
+      "learning_rate": 3.9990929437069675e-05,
+      "loss": 0.7333,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49183258386884837,
+      "learning_rate": 3.9990754178048356e-05,
+      "loss": 0.7629,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.514119169789668,
+      "learning_rate": 3.999057724242664e-05,
+      "loss": 0.7072,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5564116223036176,
+      "learning_rate": 3.999039863021934e-05,
+      "loss": 0.7954,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4734350255646209,
+      "learning_rate": 3.999021834144146e-05,
+      "loss": 0.7611,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5275432382927435,
+      "learning_rate": 3.99900363761081e-05,
+      "loss": 0.7934,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5328301469372142,
+      "learning_rate": 3.998985273423454e-05,
+      "loss": 0.7778,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5395694384419383,
+      "learning_rate": 3.9989667415836166e-05,
+      "loss": 0.6915,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5029603899138787,
+      "learning_rate": 3.9989480420928536e-05,
+      "loss": 0.7632,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4617668017967549,
+      "learning_rate": 3.9989291749527314e-05,
+      "loss": 0.7357,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.47088226127604277,
+      "learning_rate": 3.998910140164834e-05,
+      "loss": 0.7443,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.499630458957727,
+      "learning_rate": 3.998890937730757e-05,
+      "loss": 0.7708,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5054652842617218,
+      "learning_rate": 3.9988715676521115e-05,
+      "loss": 0.7596,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5367292566611183,
+      "learning_rate": 3.998852029930521e-05,
+      "loss": 0.7933,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5054848152796124,
+      "learning_rate": 3.9988323245676256e-05,
+      "loss": 0.7348,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.463950414847341,
+      "learning_rate": 3.9988124515650765e-05,
+      "loss": 0.7592,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5044164982611214,
+      "learning_rate": 3.9987924109245406e-05,
+      "loss": 0.7438,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4999060856375772,
+      "learning_rate": 3.998772202647699e-05,
+      "loss": 0.7323,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5627428217094849,
+      "learning_rate": 3.998751826736248e-05,
+      "loss": 0.7453,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5008312417133703,
+      "learning_rate": 3.998731283191895e-05,
+      "loss": 0.7673,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5303407165469599,
+      "learning_rate": 3.998710572016362e-05,
+      "loss": 0.7173,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5156074647602665,
+      "learning_rate": 3.998689693211388e-05,
+      "loss": 0.7544,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49073488503881985,
+      "learning_rate": 3.998668646778723e-05,
+      "loss": 0.7443,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5653172707136737,
+      "learning_rate": 3.998647432720132e-05,
+      "loss": 0.818,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4835953646814199,
+      "learning_rate": 3.998626051037396e-05,
+      "loss": 0.7646,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4833066477056972,
+      "learning_rate": 3.998604501732305e-05,
+      "loss": 0.7455,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5016043704415218,
+      "learning_rate": 3.998582784806669e-05,
+      "loss": 0.7499,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5146015944412317,
+      "learning_rate": 3.998560900262309e-05,
+      "loss": 0.6816,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4762394154438001,
+      "learning_rate": 3.998538848101059e-05,
+      "loss": 0.7298,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4832773400609639,
+      "learning_rate": 3.99851662832477e-05,
+      "loss": 0.7391,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.40357735770115066,
+      "learning_rate": 3.998494240935305e-05,
+      "loss": 0.6977,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7775999136286902,
+      "learning_rate": 3.998471685934541e-05,
+      "loss": 0.6587,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5110450390993293,
+      "learning_rate": 3.998448963324371e-05,
+      "loss": 0.7187,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5688434039816684,
+      "learning_rate": 3.9984260731066995e-05,
+      "loss": 0.7377,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4593094441782497,
+      "learning_rate": 3.998403015283447e-05,
+      "loss": 0.7001,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4773582662917976,
+      "learning_rate": 3.998379789856547e-05,
+      "loss": 0.7162,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.608718200010222,
+      "learning_rate": 3.998356396827947e-05,
+      "loss": 0.8292,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5139364238109617,
+      "learning_rate": 3.99833283619961e-05,
+      "loss": 0.7448,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45218235608653295,
+      "learning_rate": 3.998309107973511e-05,
+      "loss": 0.7011,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4843460532711641,
+      "learning_rate": 3.9982852121516405e-05,
+      "loss": 0.7324,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.503612714439431,
+      "learning_rate": 3.9982611487360026e-05,
+      "loss": 0.7572,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5134955246470321,
+      "learning_rate": 3.998236917728616e-05,
+      "loss": 0.7828,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4423966601587893,
+      "learning_rate": 3.998212519131512e-05,
+      "loss": 0.7317,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4428867483224958,
+      "learning_rate": 3.9981879529467366e-05,
+      "loss": 0.7324,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46094934379322944,
+      "learning_rate": 3.998163219176352e-05,
+      "loss": 0.7271,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4948788209856045,
+      "learning_rate": 3.998138317822431e-05,
+      "loss": 0.7233,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4697223690728793,
+      "learning_rate": 3.998113248887062e-05,
+      "loss": 0.7579,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46863135537999306,
+      "learning_rate": 3.998088012372348e-05,
+      "loss": 0.7338,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49051048925977436,
+      "learning_rate": 3.998062608280405e-05,
+      "loss": 0.8152,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5137668060175888,
+      "learning_rate": 3.998037036613365e-05,
+      "loss": 0.7364,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.506809486445524,
+      "learning_rate": 3.9980112973733714e-05,
+      "loss": 0.7362,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5493311241079986,
+      "learning_rate": 3.997985390562583e-05,
+      "loss": 0.7215,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.705687269389673,
+      "learning_rate": 3.997959316183173e-05,
+      "loss": 0.6412,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6009143582950591,
+      "learning_rate": 3.9979330742373286e-05,
+      "loss": 0.6966,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.516418274325998,
+      "learning_rate": 3.99790666472725e-05,
+      "loss": 0.763,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7117593688805837,
+      "learning_rate": 3.997880087655152e-05,
+      "loss": 0.7216,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5292223724376587,
+      "learning_rate": 3.997853343023263e-05,
+      "loss": 0.7395,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45106813722883526,
+      "learning_rate": 3.9978264308338287e-05,
+      "loss": 0.7363,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5112027854880274,
+      "learning_rate": 3.9977993510891036e-05,
+      "loss": 0.7361,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5065479809904685,
+      "learning_rate": 3.9977721037913594e-05,
+      "loss": 0.7417,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.548034401558303,
+      "learning_rate": 3.997744688942882e-05,
+      "loss": 0.7635,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.555254480951112,
+      "learning_rate": 3.99771710654597e-05,
+      "loss": 0.7747,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4568627961681327,
+      "learning_rate": 3.997689356602937e-05,
+      "loss": 0.6933,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4172601256823878,
+      "learning_rate": 3.99766143911611e-05,
+      "loss": 0.7619,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.502581597604271,
+      "learning_rate": 3.997633354087831e-05,
+      "loss": 0.7669,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.573362943787691,
+      "learning_rate": 3.9976051015204544e-05,
+      "loss": 0.7572,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.47547628879231846,
+      "learning_rate": 3.9975766814163506e-05,
+      "loss": 0.7691,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4579997520408388,
+      "learning_rate": 3.997548093777904e-05,
+      "loss": 0.7216,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5110363122188222,
+      "learning_rate": 3.997519338607511e-05,
+      "loss": 0.7972,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4754940976784406,
+      "learning_rate": 3.997490415907582e-05,
+      "loss": 0.7609,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5181983892112857,
+      "learning_rate": 3.997461325680545e-05,
+      "loss": 0.7486,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46615563868220794,
+      "learning_rate": 3.9974320679288395e-05,
+      "loss": 0.7034,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44062048368069706,
+      "learning_rate": 3.997402642654919e-05,
+      "loss": 0.7605,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4357162723829196,
+      "learning_rate": 3.99737304986125e-05,
+      "loss": 0.7354,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4674788622174979,
+      "learning_rate": 3.997343289550316e-05,
+      "loss": 0.7402,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4559095715870351,
+      "learning_rate": 3.9973133617246125e-05,
+      "loss": 0.7487,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49189474671377603,
+      "learning_rate": 3.9972832663866496e-05,
+      "loss": 0.7868,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4402143881624007,
+      "learning_rate": 3.997253003538952e-05,
+      "loss": 0.7555,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4637485369929408,
+      "learning_rate": 3.997222573184056e-05,
+      "loss": 0.7566,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4882733384689876,
+      "learning_rate": 3.997191975324515e-05,
+      "loss": 0.7848,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46303021891554025,
+      "learning_rate": 3.9971612099628945e-05,
+      "loss": 0.7357,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49286269564427126,
+      "learning_rate": 3.997130277101776e-05,
+      "loss": 0.7739,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5350027590117901,
+      "learning_rate": 3.997099176743753e-05,
+      "loss": 0.7695,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5086785370244832,
+      "learning_rate": 3.997067908891434e-05,
+      "loss": 0.7215,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4972091216470832,
+      "learning_rate": 3.9970364735474406e-05,
+      "loss": 0.7597,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4542807836538181,
+      "learning_rate": 3.9970048707144104e-05,
+      "loss": 0.7438,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4525914051313871,
+      "learning_rate": 3.996973100394994e-05,
+      "loss": 0.7533,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.47081896726606776,
+      "learning_rate": 3.996941162591855e-05,
+      "loss": 0.7598,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.518697720808655,
+      "learning_rate": 3.9969090573076725e-05,
+      "loss": 0.7733,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5098420914352816,
+      "learning_rate": 3.996876784545139e-05,
+      "loss": 0.7481,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4900198152464831,
+      "learning_rate": 3.996844344306961e-05,
+      "loss": 0.7595,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43438279678263364,
+      "learning_rate": 3.9968117365958594e-05,
+      "loss": 0.7191,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5085730160168709,
+      "learning_rate": 3.9967789614145694e-05,
+      "loss": 0.7808,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3666255558473451,
+      "learning_rate": 3.996746018765839e-05,
+      "loss": 0.7058,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4822794120037121,
+      "learning_rate": 3.996712908652431e-05,
+      "loss": 0.7341,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44439607908117196,
+      "learning_rate": 3.996679631077124e-05,
+      "loss": 0.7361,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4736669687444654,
+      "learning_rate": 3.996646186042707e-05,
+      "loss": 0.7374,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49410885481363276,
+      "learning_rate": 3.996612573551986e-05,
+      "loss": 0.7888,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45502135372218483,
+      "learning_rate": 3.9965787936077795e-05,
+      "loss": 0.7743,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46731409326076667,
+      "learning_rate": 3.996544846212921e-05,
+      "loss": 0.7384,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5425808602805752,
+      "learning_rate": 3.996510731370257e-05,
+      "loss": 0.7934,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4499074599860976,
+      "learning_rate": 3.99647644908265e-05,
+      "loss": 0.7536,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44745051186118084,
+      "learning_rate": 3.996441999352973e-05,
+      "loss": 0.7373,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4650436951930732,
+      "learning_rate": 3.996407382184118e-05,
+      "loss": 0.7302,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4659760800595109,
+      "learning_rate": 3.996372597578987e-05,
+      "loss": 0.7581,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4914638738856295,
+      "learning_rate": 3.996337645540496e-05,
+      "loss": 0.7453,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.48450093068131883,
+      "learning_rate": 3.9963025260715784e-05,
+      "loss": 0.7164,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4850269856682674,
+      "learning_rate": 3.996267239175179e-05,
+      "loss": 0.733,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.48442592462946815,
+      "learning_rate": 3.9962317848542565e-05,
+      "loss": 0.7137,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4745376692735461,
+      "learning_rate": 3.9961961631117853e-05,
+      "loss": 0.7483,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44523215575647823,
+      "learning_rate": 3.996160373950753e-05,
+      "loss": 0.758,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44719074086191785,
+      "learning_rate": 3.99612441737416e-05,
+      "loss": 0.7324,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42660002130618996,
+      "learning_rate": 3.9960882933850236e-05,
+      "loss": 0.7328,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4388195654080585,
+      "learning_rate": 3.996052001986373e-05,
+      "loss": 0.7588,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46454227100251205,
+      "learning_rate": 3.996015543181251e-05,
+      "loss": 0.7804,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5218042617850587,
+      "learning_rate": 3.995978916972716e-05,
+      "loss": 0.7339,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5077729860679063,
+      "learning_rate": 3.9959421233638405e-05,
+      "loss": 0.7634,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4695935150588257,
+      "learning_rate": 3.9959051623577085e-05,
+      "loss": 0.6909,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.47212728735520015,
+      "learning_rate": 3.995868033957422e-05,
+      "loss": 0.7473,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5041198730232135,
+      "learning_rate": 3.995830738166093e-05,
+      "loss": 0.7584,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5626244382780249,
+      "learning_rate": 3.995793274986851e-05,
+      "loss": 0.8016,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3986517981554386,
+      "learning_rate": 3.995755644422837e-05,
+      "loss": 0.6708,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4821895054650317,
+      "learning_rate": 3.995717846477207e-05,
+      "loss": 0.7225,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4022324644000024,
+      "learning_rate": 3.995679881153132e-05,
+      "loss": 0.7294,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40050503486078864,
+      "learning_rate": 3.9956417484537956e-05,
+      "loss": 0.7028,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45896161164635413,
+      "learning_rate": 3.9956034483823954e-05,
+      "loss": 0.7426,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43583022188703097,
+      "learning_rate": 3.995564980942145e-05,
+      "loss": 0.7348,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41905247320742617,
+      "learning_rate": 3.995526346136269e-05,
+      "loss": 0.7096,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4774978386431727,
+      "learning_rate": 3.995487543968008e-05,
+      "loss": 0.7601,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49000538991323433,
+      "learning_rate": 3.995448574440617e-05,
+      "loss": 0.7117,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.428993837873373,
+      "learning_rate": 3.995409437557364e-05,
+      "loss": 0.7779,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45967836497720205,
+      "learning_rate": 3.9953701333215314e-05,
+      "loss": 0.7576,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4767651602302258,
+      "learning_rate": 3.995330661736416e-05,
+      "loss": 0.7014,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4092451323542536,
+      "learning_rate": 3.995291022805326e-05,
+      "loss": 0.7355,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.499839941098459,
+      "learning_rate": 3.995251216531589e-05,
+      "loss": 0.7889,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.441522639223317,
+      "learning_rate": 3.995211242918542e-05,
+      "loss": 0.7691,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5306510379455828,
+      "learning_rate": 3.9951711019695375e-05,
+      "loss": 0.729,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46803123377346373,
+      "learning_rate": 3.995130793687942e-05,
+      "loss": 0.7183,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41761593975070116,
+      "learning_rate": 3.9950903180771365e-05,
+      "loss": 0.7282,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45751247442225373,
+      "learning_rate": 3.995049675140516e-05,
+      "loss": 0.6901,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5423798931289802,
+      "learning_rate": 3.995008864881488e-05,
+      "loss": 0.7401,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.48614506050475736,
+      "learning_rate": 3.9949678873034766e-05,
+      "loss": 0.7329,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.47929650321763134,
+      "learning_rate": 3.9949267424099175e-05,
+      "loss": 0.7212,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4866502758373775,
+      "learning_rate": 3.994885430204261e-05,
+      "loss": 0.7756,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49015035904417786,
+      "learning_rate": 3.994843950689974e-05,
+      "loss": 0.7698,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4564079015045863,
+      "learning_rate": 3.994802303870533e-05,
+      "loss": 0.7364,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45808968954550616,
+      "learning_rate": 3.994760489749432e-05,
+      "loss": 0.7627,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3719042662753131,
+      "learning_rate": 3.994718508330178e-05,
+      "loss": 0.7127,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4234109373037785,
+      "learning_rate": 3.994676359616292e-05,
+      "loss": 0.7312,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4183585747422093,
+      "learning_rate": 3.9946340436113084e-05,
+      "loss": 0.7415,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43350609066901163,
+      "learning_rate": 3.994591560318776e-05,
+      "loss": 0.7299,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46977403689332414,
+      "learning_rate": 3.994548909742259e-05,
+      "loss": 0.8002,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49385177624897775,
+      "learning_rate": 3.994506091885333e-05,
+      "loss": 0.7549,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46458520120318764,
+      "learning_rate": 3.994463106751591e-05,
+      "loss": 0.7135,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43815553749522457,
+      "learning_rate": 3.994419954344635e-05,
+      "loss": 0.7158,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4892271759524825,
+      "learning_rate": 3.9943766346680884e-05,
+      "loss": 0.7549,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5036278307920189,
+      "learning_rate": 3.994333147725581e-05,
+      "loss": 0.764,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4243066085350861,
+      "learning_rate": 3.994289493520761e-05,
+      "loss": 0.724,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5287222412159699,
+      "learning_rate": 3.994245672057289e-05,
+      "loss": 0.7907,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4175193930789452,
+      "learning_rate": 3.994201683338842e-05,
+      "loss": 0.7128,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43904272468745076,
+      "learning_rate": 3.994157527369107e-05,
+      "loss": 0.7386,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46285653286800393,
+      "learning_rate": 3.994113204151789e-05,
+      "loss": 0.7212,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.515188646810342,
+      "learning_rate": 3.994068713690606e-05,
+      "loss": 0.7596,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.429906653134929,
+      "learning_rate": 3.9940240559892875e-05,
+      "loss": 0.6941,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.442164660882956,
+      "learning_rate": 3.993979231051579e-05,
+      "loss": 0.7232,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4534540773910036,
+      "learning_rate": 3.993934238881241e-05,
+      "loss": 0.7427,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4495892054360521,
+      "learning_rate": 3.9938890794820465e-05,
+      "loss": 0.7015,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5023373894031626,
+      "learning_rate": 3.9938437528577824e-05,
+      "loss": 0.7838,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43495458072075716,
+      "learning_rate": 3.993798259012252e-05,
+      "loss": 0.7381,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.47040317798553366,
+      "learning_rate": 3.993752597949268e-05,
+      "loss": 0.7285,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44468793519768696,
+      "learning_rate": 3.993706769672662e-05,
+      "loss": 0.7422,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42649780516957064,
+      "learning_rate": 3.9936607741862774e-05,
+      "loss": 0.6936,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4217826384282669,
+      "learning_rate": 3.993614611493971e-05,
+      "loss": 0.699,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42470267670744277,
+      "learning_rate": 3.993568281599615e-05,
+      "loss": 0.7241,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4509763740751476,
+      "learning_rate": 3.993521784507095e-05,
+      "loss": 0.7601,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49400684202994244,
+      "learning_rate": 3.9934751202203104e-05,
+      "loss": 0.7568,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38894131797132725,
+      "learning_rate": 3.993428288743175e-05,
+      "loss": 0.7011,
+      "step": 616
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4937885213993327,
+      "learning_rate": 3.993381290079616e-05,
+      "loss": 0.7203,
+      "step": 617
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5623060543607711,
+      "learning_rate": 3.993334124233576e-05,
+      "loss": 0.7864,
+      "step": 618
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.548386842111625,
+      "learning_rate": 3.99328679120901e-05,
+      "loss": 0.7769,
+      "step": 619
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4541568321010082,
+      "learning_rate": 3.993239291009888e-05,
+      "loss": 0.7367,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5008002803665998,
+      "learning_rate": 3.993191623640195e-05,
+      "loss": 0.7821,
+      "step": 621
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4411159486526937,
+      "learning_rate": 3.9931437891039266e-05,
+      "loss": 0.6966,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5103855472293527,
+      "learning_rate": 3.9930957874050964e-05,
+      "loss": 0.7169,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4519700505963763,
+      "learning_rate": 3.9930476185477286e-05,
+      "loss": 0.7334,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4615198322929611,
+      "learning_rate": 3.992999282535865e-05,
+      "loss": 0.6904,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4253755603677052,
+      "learning_rate": 3.992950779373558e-05,
+      "loss": 0.7252,
+      "step": 626
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4169575171380782,
+      "learning_rate": 3.992902109064876e-05,
+      "loss": 0.7143,
+      "step": 627
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4207811858353583,
+      "learning_rate": 3.992853271613901e-05,
+      "loss": 0.6767,
+      "step": 628
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4673632113693494,
+      "learning_rate": 3.9928042670247294e-05,
+      "loss": 0.7601,
+      "step": 629
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4838814458536056,
+      "learning_rate": 3.99275509530147e-05,
+      "loss": 0.7528,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.44520215638511945,
+      "learning_rate": 3.992705756448248e-05,
+      "loss": 0.7264,
+      "step": 631
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48728720454481605,
+      "learning_rate": 3.9926562504692e-05,
+      "loss": 0.71,
+      "step": 632
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4641090816032475,
+      "learning_rate": 3.992606577368479e-05,
+      "loss": 0.7568,
+      "step": 633
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4381944755783288,
+      "learning_rate": 3.992556737150251e-05,
+      "loss": 0.7561,
+      "step": 634
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.49126969143383437,
+      "learning_rate": 3.992506729818696e-05,
+      "loss": 0.7043,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.43447666327443046,
+      "learning_rate": 3.9924565553780076e-05,
+      "loss": 0.7096,
+      "step": 636
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4439906347332975,
+      "learning_rate": 3.9924062138323945e-05,
+      "loss": 0.7059,
+      "step": 637
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4588851692557367,
+      "learning_rate": 3.992355705186078e-05,
+      "loss": 0.7656,
+      "step": 638
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7993110166818506,
+      "learning_rate": 3.992305029443295e-05,
+      "loss": 0.6544,
+      "step": 639
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48202418408398734,
+      "learning_rate": 3.992254186608295e-05,
+      "loss": 0.6973,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4123080070346283,
+      "learning_rate": 3.992203176685343e-05,
+      "loss": 0.7114,
+      "step": 641
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5086328119722168,
+      "learning_rate": 3.992151999678716e-05,
+      "loss": 0.7552,
+      "step": 642
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4299683864240533,
+      "learning_rate": 3.9921006555927065e-05,
+      "loss": 0.6695,
+      "step": 643
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.42301035353877564,
+      "learning_rate": 3.9920491444316215e-05,
+      "loss": 0.7215,
+      "step": 644
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.42329049383553824,
+      "learning_rate": 3.99199746619978e-05,
+      "loss": 0.6996,
+      "step": 645
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4814700922482975,
+      "learning_rate": 3.9919456209015174e-05,
+      "loss": 0.6967,
+      "step": 646
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5023630409542857,
+      "learning_rate": 3.991893608541181e-05,
+      "loss": 0.7688,
+      "step": 647
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4628271654299107,
+      "learning_rate": 3.9918414291231326e-05,
+      "loss": 0.7616,
+      "step": 648
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.46905672630205186,
+      "learning_rate": 3.9917890826517496e-05,
+      "loss": 0.7036,
+      "step": 649
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5335670456218143,
+      "learning_rate": 3.9917365691314216e-05,
+      "loss": 0.7875,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40814934409904,
+      "learning_rate": 3.991683888566553e-05,
+      "loss": 0.7249,
+      "step": 651
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4601071326829113,
+      "learning_rate": 3.991631040961562e-05,
+      "loss": 0.7625,
+      "step": 652
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4240926511173165,
+      "learning_rate": 3.991578026320881e-05,
+      "loss": 0.7181,
+      "step": 653
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4636110820060914,
+      "learning_rate": 3.9915248446489564e-05,
+      "loss": 0.7747,
+      "step": 654
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48880374577497443,
+      "learning_rate": 3.9914714959502474e-05,
+      "loss": 0.747,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4568515486302041,
+      "learning_rate": 3.99141798022923e-05,
+      "loss": 0.7422,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5300344489680556,
+      "learning_rate": 3.9913642974903906e-05,
+      "loss": 0.7284,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.47763168348640317,
+      "learning_rate": 3.9913104477382335e-05,
+      "loss": 0.7722,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4413356689177174,
+      "learning_rate": 3.991256430977274e-05,
+      "loss": 0.7517,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.46059365321689244,
+      "learning_rate": 3.991202247212042e-05,
+      "loss": 0.7279,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4621054438084887,
+      "learning_rate": 3.991147896447082e-05,
+      "loss": 0.7028,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5327712431745932,
+      "learning_rate": 3.991093378686953e-05,
+      "loss": 0.7807,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41652322616373055,
+      "learning_rate": 3.9910386939362274e-05,
+      "loss": 0.725,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4352128103582359,
+      "learning_rate": 3.990983842199491e-05,
+      "loss": 0.7373,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5066557736764356,
+      "learning_rate": 3.990928823481344e-05,
+      "loss": 0.7489,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.43772148664693966,
+      "learning_rate": 3.9908736377864005e-05,
+      "loss": 0.7031,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4154548164516414,
+      "learning_rate": 3.9908182851192905e-05,
+      "loss": 0.7363,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40391135648979765,
+      "learning_rate": 3.9907627654846536e-05,
+      "loss": 0.7226,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4527504463399629,
+      "learning_rate": 3.990707078887149e-05,
+      "loss": 0.7188,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41747257391924164,
+      "learning_rate": 3.990651225331446e-05,
+      "loss": 0.7283,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4526406922903793,
+      "learning_rate": 3.9905952048222286e-05,
+      "loss": 0.7261,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.49700656564956247,
+      "learning_rate": 3.990539017364195e-05,
+      "loss": 0.7788,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.43443054291081684,
+      "learning_rate": 3.990482662962058e-05,
+      "loss": 0.7179,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.47672929241892076,
+      "learning_rate": 3.9904261416205444e-05,
+      "loss": 0.7613,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4612123466863538,
+      "learning_rate": 3.990369453344394e-05,
+      "loss": 0.7174,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.44399561165152024,
+      "learning_rate": 3.990312598138361e-05,
+      "loss": 0.7598,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.46923500411249686,
+      "learning_rate": 3.9902555760072144e-05,
+      "loss": 0.7493,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48918176013083964,
+      "learning_rate": 3.990198386955737e-05,
+      "loss": 0.7739,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6669405814761286,
+      "learning_rate": 3.990141030988723e-05,
+      "loss": 0.6176,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5086866542931996,
+      "learning_rate": 3.990083508110985e-05,
+      "loss": 0.7163,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5357689182379822,
+      "learning_rate": 3.990025818327346e-05,
+      "loss": 0.6373,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5819516148392687,
+      "learning_rate": 3.989967961642646e-05,
+      "loss": 0.7975,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4924260776106743,
+      "learning_rate": 3.9899099380617354e-05,
+      "loss": 0.7811,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4602703389502578,
+      "learning_rate": 3.989851747589482e-05,
+      "loss": 0.7113,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4785369780243868,
+      "learning_rate": 3.989793390230765e-05,
+      "loss": 0.7307,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.43340338583046467,
+      "learning_rate": 3.98973486599048e-05,
+      "loss": 0.7624,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4960542632026031,
+      "learning_rate": 3.989676174873535e-05,
+      "loss": 0.8062,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48572184828309783,
+      "learning_rate": 3.9896173168848515e-05,
+      "loss": 0.7129,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6329362546050996,
+      "learning_rate": 3.989558292029367e-05,
+      "loss": 0.6496,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.49781847010900376,
+      "learning_rate": 3.989499100312031e-05,
+      "loss": 0.7448,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4642368616189104,
+      "learning_rate": 3.989439741737808e-05,
+      "loss": 0.7631,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5255556950836021,
+      "learning_rate": 3.989380216311677e-05,
+      "loss": 0.7567,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5135647214624632,
+      "learning_rate": 3.98932052403863e-05,
+      "loss": 0.8076,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4344503726385802,
+      "learning_rate": 3.989260664923672e-05,
+      "loss": 0.7358,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4872984958302139,
+      "learning_rate": 3.989200638971825e-05,
+      "loss": 0.7358,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4601666967528821,
+      "learning_rate": 3.989140446188123e-05,
+      "loss": 0.7375,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5331952149277639,
+      "learning_rate": 3.989080086577614e-05,
+      "loss": 0.7602,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5059667947279325,
+      "learning_rate": 3.9890195601453606e-05,
+      "loss": 0.7465,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4763150508774377,
+      "learning_rate": 3.988958866896439e-05,
+      "loss": 0.6446,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.47517646360673454,
+      "learning_rate": 3.988898006835938e-05,
+      "loss": 0.714,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40880007003903085,
+      "learning_rate": 3.9888369799689644e-05,
+      "loss": 0.6925,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4664870432146247,
+      "learning_rate": 3.988775786300635e-05,
+      "loss": 0.7941,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4328303092243608,
+      "learning_rate": 3.9887144258360824e-05,
+      "loss": 0.7573,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.444522631170833,
+      "learning_rate": 3.988652898580452e-05,
+      "loss": 0.7498,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4721571084193608,
+      "learning_rate": 3.988591204538905e-05,
+      "loss": 0.7058,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5103536920270852,
+      "learning_rate": 3.9885293437166155e-05,
+      "loss": 0.7184,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4335222664605951,
+      "learning_rate": 3.988467316118771e-05,
+      "loss": 0.7242,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.45187954598808266,
+      "learning_rate": 3.9884051217505734e-05,
+      "loss": 0.7636,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4702914735357344,
+      "learning_rate": 3.988342760617241e-05,
+      "loss": 0.727,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4161888996755085,
+      "learning_rate": 3.988280232724003e-05,
+      "loss": 0.7265,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4486301021589009,
+      "learning_rate": 3.9882175380761016e-05,
+      "loss": 0.7517,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.49802150032277887,
+      "learning_rate": 3.988154676678797e-05,
+      "loss": 0.7846,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4663419796617441,
+      "learning_rate": 3.9880916485373606e-05,
+      "loss": 0.7745,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4524823140035991,
+      "learning_rate": 3.9880284536570796e-05,
+      "loss": 0.7445,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.43321099081673803,
+      "learning_rate": 3.9879650920432527e-05,
+      "loss": 0.7244,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48186395136056975,
+      "learning_rate": 3.987901563701194e-05,
+      "loss": 0.7714,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4368603597484503,
+      "learning_rate": 3.987837868636232e-05,
+      "loss": 0.7139,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.45719466236900796,
+      "learning_rate": 3.9877740068537084e-05,
+      "loss": 0.7013,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4949996367848271,
+      "learning_rate": 3.9877099783589795e-05,
+      "loss": 0.7658,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4653735789161951,
+      "learning_rate": 3.987645783157416e-05,
+      "loss": 0.753,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4235289186093677,
+      "learning_rate": 3.9875814212544006e-05,
+      "loss": 0.7319,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.43109763969533443,
+      "learning_rate": 3.9875168926553315e-05,
+      "loss": 0.7332,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4970386909791161,
+      "learning_rate": 3.987452197365622e-05,
+      "loss": 0.7581,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.457488614947762,
+      "learning_rate": 3.987387335390696e-05,
+      "loss": 0.7104,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4960576378395833,
+      "learning_rate": 3.987322306735994e-05,
+      "loss": 0.7499,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4902023666980728,
+      "learning_rate": 3.987257111406971e-05,
+      "loss": 0.7875,
+      "step": 726
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.45148246891709753,
+      "learning_rate": 3.987191749409094e-05,
+      "loss": 0.7159,
+      "step": 727
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4688063725671951,
+      "learning_rate": 3.987126220747845e-05,
+      "loss": 0.723,
+      "step": 728
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.41924511038850604,
+      "learning_rate": 3.987060525428719e-05,
+      "loss": 0.7148,
+      "step": 729
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6242519526203664,
+      "learning_rate": 3.986994663457227e-05,
+      "loss": 0.6508,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5571332255791903,
+      "learning_rate": 3.986928634838893e-05,
+      "loss": 0.7251,
+      "step": 731
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4365429596996432,
+      "learning_rate": 3.986862439579253e-05,
+      "loss": 0.7149,
+      "step": 732
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.46913800796256866,
+      "learning_rate": 3.98679607768386e-05,
+      "loss": 0.7329,
+      "step": 733
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.44293675848588776,
+      "learning_rate": 3.9867295491582784e-05,
+      "loss": 0.7432,
+      "step": 734
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4979765601316522,
+      "learning_rate": 3.9866628540080905e-05,
+      "loss": 0.7578,
+      "step": 735
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4200698977898931,
+      "learning_rate": 3.986595992238887e-05,
+      "loss": 0.7152,
+      "step": 736
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4388092938936519,
+      "learning_rate": 3.986528963856278e-05,
+      "loss": 0.731,
+      "step": 737
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.45802487066773423,
+      "learning_rate": 3.986461768865884e-05,
+      "loss": 0.7269,
+      "step": 738
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3926848436890785,
+      "learning_rate": 3.9863944072733394e-05,
+      "loss": 0.6756,
+      "step": 739
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4648858404479435,
+      "learning_rate": 3.986326879084295e-05,
+      "loss": 0.7349,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4385000860224574,
+      "learning_rate": 3.986259184304414e-05,
+      "loss": 0.731,
+      "step": 741
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4994429050458236,
+      "learning_rate": 3.986191322939375e-05,
+      "loss": 0.7612,
+      "step": 742
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5120360731751818,
+      "learning_rate": 3.986123294994868e-05,
+      "loss": 0.7819,
+      "step": 743
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4276927804500961,
+      "learning_rate": 3.9860551004765985e-05,
+      "loss": 0.7122,
+      "step": 744
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5029414791069715,
+      "learning_rate": 3.9859867393902864e-05,
+      "loss": 0.7648,
+      "step": 745
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5363716384476919,
+      "learning_rate": 3.985918211741665e-05,
+      "loss": 0.764,
+      "step": 746
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4629687890968092,
+      "learning_rate": 3.985849517536481e-05,
+      "loss": 0.7326,
+      "step": 747
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4632793311021556,
+      "learning_rate": 3.985780656780498e-05,
+      "loss": 0.6638,
+      "step": 748
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4845068344788968,
+      "learning_rate": 3.9857116294794874e-05,
+      "loss": 0.7198,
+      "step": 749
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5284007240550077,
+      "learning_rate": 3.9856424356392425e-05,
+      "loss": 0.7293,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7420424860981486,
+      "learning_rate": 3.9855730752655636e-05,
+      "loss": 0.6871,
+      "step": 751
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4346071261017445,
+      "learning_rate": 3.985503548364269e-05,
+      "loss": 0.7051,
+      "step": 752
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.48255169195557585,
+      "learning_rate": 3.9854338549411895e-05,
+      "loss": 0.7163,
+      "step": 753
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4209120023869542,
+      "learning_rate": 3.9853639950021705e-05,
+      "loss": 0.6872,
+      "step": 754
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.47917716220142065,
+      "learning_rate": 3.9852939685530706e-05,
+      "loss": 0.7517,
+      "step": 755
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.491401028979925,
+      "learning_rate": 3.985223775599763e-05,
+      "loss": 0.7784,
+      "step": 756
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.41264753209502497,
+      "learning_rate": 3.985153416148136e-05,
+      "loss": 0.7122,
+      "step": 757
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3981239353978901,
+      "learning_rate": 3.985082890204089e-05,
+      "loss": 0.6933,
+      "step": 758
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.429733798134422,
+      "learning_rate": 3.9850121977735366e-05,
+      "loss": 0.7361,
+      "step": 759
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.42652830717112783,
+      "learning_rate": 3.984941338862409e-05,
+      "loss": 0.688,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4421000687823219,
+      "learning_rate": 3.984870313476649e-05,
+      "loss": 0.7051,
+      "step": 761
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4964008785762364,
+      "learning_rate": 3.984799121622212e-05,
+      "loss": 0.7564,
+      "step": 762
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5080035674385439,
+      "learning_rate": 3.9847277633050704e-05,
+      "loss": 0.7868,
+      "step": 763
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5033186584744674,
+      "learning_rate": 3.984656238531208e-05,
+      "loss": 0.727,
+      "step": 764
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.451789616672876,
+      "learning_rate": 3.984584547306624e-05,
+      "loss": 0.6878,
+      "step": 765
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4254450726533525,
+      "learning_rate": 3.98451268963733e-05,
+      "loss": 0.7283,
+      "step": 766
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4819095669527727,
+      "learning_rate": 3.9844406655293545e-05,
+      "loss": 0.768,
+      "step": 767
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4115112964558116,
+      "learning_rate": 3.9843684749887364e-05,
+      "loss": 0.7302,
+      "step": 768
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.43556777677302966,
+      "learning_rate": 3.984296118021531e-05,
+      "loss": 0.7485,
+      "step": 769
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4404508504324208,
+      "learning_rate": 3.984223594633806e-05,
+      "loss": 0.7397,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.45949984384566284,
+      "learning_rate": 3.984150904831645e-05,
+      "loss": 0.7194,
+      "step": 771
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.44565964384652923,
+      "learning_rate": 3.984078048621144e-05,
+      "loss": 0.6977,
+      "step": 772
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7127720529961585,
+      "learning_rate": 3.9840050260084124e-05,
+      "loss": 0.6809,
+      "step": 773
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4377220602806821,
+      "learning_rate": 3.983931836999576e-05,
+      "loss": 0.6856,
+      "step": 774
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4437519645245356,
+      "learning_rate": 3.983858481600773e-05,
+      "loss": 0.7125,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5098563537227444,
+      "learning_rate": 3.983784959818154e-05,
+      "loss": 0.7589,
+      "step": 776
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5231796363599394,
+      "learning_rate": 3.983711271657886e-05,
+      "loss": 0.7475,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4281848721515563,
+      "learning_rate": 3.983637417126151e-05,
+      "loss": 0.7339,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5058776597511776,
+      "learning_rate": 3.9835633962291396e-05,
+      "loss": 0.7812,
+      "step": 779
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4653421829989136,
+      "learning_rate": 3.983489208973063e-05,
+      "loss": 0.7359,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5930892635050106,
+      "learning_rate": 3.983414855364141e-05,
+      "loss": 0.6607,
+      "step": 781
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8698387716591811,
+      "learning_rate": 3.983340335408611e-05,
+      "loss": 0.7292,
+      "step": 782
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5141802151371111,
+      "learning_rate": 3.983265649112722e-05,
+      "loss": 0.7448,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5546025723265418,
+      "learning_rate": 3.9831907964827386e-05,
+      "loss": 0.8033,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5139288414716754,
+      "learning_rate": 3.983115777524938e-05,
+      "loss": 0.7266,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.447120971923023,
+      "learning_rate": 3.983040592245612e-05,
+      "loss": 0.7061,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.524912035220329,
+      "learning_rate": 3.9829652406510666e-05,
+      "loss": 0.7082,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.526814712221743,
+      "learning_rate": 3.982889722747621e-05,
+      "loss": 0.7207,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.39810744482813243,
+      "learning_rate": 3.982814038541609e-05,
+      "loss": 0.7401,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4331227391422036,
+      "learning_rate": 3.982738188039379e-05,
+      "loss": 0.7452,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5021963059080485,
+      "learning_rate": 3.982662171247291e-05,
+      "loss": 0.7679,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.49993106007197197,
+      "learning_rate": 3.9825859881717214e-05,
+      "loss": 0.7675,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.40150626012648194,
+      "learning_rate": 3.9825096388190594e-05,
+      "loss": 0.6723,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.37792388451777725,
+      "learning_rate": 3.982433123195708e-05,
+      "loss": 0.7623,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4935298319572828,
+      "learning_rate": 3.982356441308085e-05,
+      "loss": 0.7754,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.47581204653580045,
+      "learning_rate": 3.982279593162621e-05,
+      "loss": 0.7626,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4138580230888477,
+      "learning_rate": 3.982202578765762e-05,
+      "loss": 0.7523,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4851266098686905,
+      "learning_rate": 3.982125398123966e-05,
+      "loss": 0.7225,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.41770067628857993,
+      "learning_rate": 3.982048051243707e-05,
+      "loss": 0.6998,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5103646492137653,
+      "learning_rate": 3.981970538131472e-05,
+      "loss": 0.7305,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.40474080846848337,
+      "learning_rate": 3.981892858793761e-05,
+      "loss": 0.7163,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4423007414714302,
+      "learning_rate": 3.98181501323709e-05,
+      "loss": 0.737,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4913154559512235,
+      "learning_rate": 3.981737001467987e-05,
+      "loss": 0.7707,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4246605892302057,
+      "learning_rate": 3.981658823492995e-05,
+      "loss": 0.7067,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.41759718377656757,
+      "learning_rate": 3.981580479318671e-05,
+      "loss": 0.6749,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4865198601513159,
+      "learning_rate": 3.981501968951585e-05,
+      "loss": 0.7432,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4289167354464642,
+      "learning_rate": 3.981423292398321e-05,
+      "loss": 0.7182,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.45164945735826223,
+      "learning_rate": 3.98134444966548e-05,
+      "loss": 0.7593,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4683117122971864,
+      "learning_rate": 3.981265440759672e-05,
+      "loss": 0.7534,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.44667972206408607,
+      "learning_rate": 3.981186265687525e-05,
+      "loss": 0.7276,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.44042766786078513,
+      "learning_rate": 3.981106924455678e-05,
+      "loss": 0.7637,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5265603399474437,
+      "learning_rate": 3.981027417070785e-05,
+      "loss": 0.729,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.42890314416845116,
+      "learning_rate": 3.980947743539516e-05,
+      "loss": 0.6865,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4187445836775986,
+      "learning_rate": 3.9808679038685526e-05,
+      "loss": 0.7259,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4951896327763837,
+      "learning_rate": 3.98078789806459e-05,
+      "loss": 0.7487,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.43825221843387596,
+      "learning_rate": 3.9807077261343374e-05,
+      "loss": 0.7375,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3990694682800376,
+      "learning_rate": 3.980627388084521e-05,
+      "loss": 0.7202,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4166955279919618,
+      "learning_rate": 3.980546883921878e-05,
+      "loss": 0.7333,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.42407484618586394,
+      "learning_rate": 3.980466213653159e-05,
+      "loss": 0.7671,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4173204717697616,
+      "learning_rate": 3.9803853772851304e-05,
+      "loss": 0.7436,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4536782332366876,
+      "learning_rate": 3.980304374824572e-05,
+      "loss": 0.7131,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4357785566567726,
+      "learning_rate": 3.980223206278278e-05,
+      "loss": 0.7171,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.39568944481190754,
+      "learning_rate": 3.9801418716530545e-05,
+      "loss": 0.7034,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4244108242274051,
+      "learning_rate": 3.980060370955723e-05,
+      "loss": 0.6889,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.44540320603460887,
+      "learning_rate": 3.97997870419312e-05,
+      "loss": 0.7586,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.43768081702525075,
+      "learning_rate": 3.979896871372095e-05,
+      "loss": 0.757,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.439476060753004,
+      "learning_rate": 3.97981487249951e-05,
+      "loss": 0.7705,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4467926782508042,
+      "learning_rate": 3.9797327075822425e-05,
+      "loss": 0.7289,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3796168214329519,
+      "learning_rate": 3.9796503766271845e-05,
+      "loss": 0.6934,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5101590276473768,
+      "learning_rate": 3.97956787964124e-05,
+      "loss": 0.7359,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4198988710697302,
+      "learning_rate": 3.979485216631327e-05,
+      "loss": 0.7258,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.43030869092969176,
+      "learning_rate": 3.97940238760438e-05,
+      "loss": 0.7514,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4129805438240203,
+      "learning_rate": 3.9793193925673456e-05,
+      "loss": 0.7713,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.41328361838244604,
+      "learning_rate": 3.979236231527185e-05,
+      "loss": 0.7181,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4153760909440419,
+      "learning_rate": 3.9791529044908704e-05,
+      "loss": 0.7307,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4425482944408306,
+      "learning_rate": 3.979069411465393e-05,
+      "loss": 0.6945,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4721019125219253,
+      "learning_rate": 3.978985752457753e-05,
+      "loss": 0.7429,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4119474960491266,
+      "learning_rate": 3.9789019274749685e-05,
+      "loss": 0.7007,
+      "step": 838
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4027102302602117,
+      "learning_rate": 3.97881793652407e-05,
+      "loss": 0.7319,
+      "step": 839
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.40336557660313865,
+      "learning_rate": 3.978733779612101e-05,
+      "loss": 0.699,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.511337607745208,
+      "learning_rate": 3.9786494567461184e-05,
+      "loss": 0.7567,
+      "step": 841
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4108358106351842,
+      "learning_rate": 3.978564967933196e-05,
+      "loss": 0.7224,
+      "step": 842
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4623075494734944,
+      "learning_rate": 3.9784803131804195e-05,
+      "loss": 0.7417,
+      "step": 843
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4498448137300952,
+      "learning_rate": 3.9783954924948884e-05,
+      "loss": 0.714,
+      "step": 844
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5334717361046395,
+      "learning_rate": 3.978310505883716e-05,
+      "loss": 0.6527,
+      "step": 845
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4988723376841438,
+      "learning_rate": 3.978225353354032e-05,
+      "loss": 0.7292,
+      "step": 846
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.42147400077775915,
+      "learning_rate": 3.978140034912976e-05,
+      "loss": 0.6922,
+      "step": 847
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4549677803339487,
+      "learning_rate": 3.978054550567704e-05,
+      "loss": 0.6956,
+      "step": 848
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.46433253549208064,
+      "learning_rate": 3.977968900325386e-05,
+      "loss": 0.7061,
+      "step": 849
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4391041155923521,
+      "learning_rate": 3.977883084193205e-05,
+      "loss": 0.7328,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4722175997332042,
+      "learning_rate": 3.9777971021783585e-05,
+      "loss": 0.7345,
+      "step": 851
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3938856221549249,
+      "learning_rate": 3.9777109542880574e-05,
+      "loss": 0.7345,
+      "step": 852
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4098773000769218,
+      "learning_rate": 3.977624640529527e-05,
+      "loss": 0.6928,
+      "step": 853
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4925799764866705,
+      "learning_rate": 3.977538160910007e-05,
+      "loss": 0.7449,
+      "step": 854
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.41851457913214474,
+      "learning_rate": 3.977451515436749e-05,
+      "loss": 0.7196,
+      "step": 855
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.39627309875643674,
+      "learning_rate": 3.977364704117021e-05,
+      "loss": 0.6942,
+      "step": 856
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.39915224162622664,
+      "learning_rate": 3.977277726958103e-05,
+      "loss": 0.7455,
+      "step": 857
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.42080455897435876,
+      "learning_rate": 3.97719058396729e-05,
+      "loss": 0.7121,
+      "step": 858
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.41950339168387024,
+      "learning_rate": 3.9771032751518914e-05,
+      "loss": 0.7349,
+      "step": 859
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4365258170183948,
+      "learning_rate": 3.977015800519228e-05,
+      "loss": 0.729,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3988093715865241,
+      "learning_rate": 3.976928160076637e-05,
+      "loss": 0.7311,
+      "step": 861
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.45321414836131024,
+      "learning_rate": 3.976840353831469e-05,
+      "loss": 0.719,
+      "step": 862
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.34017517976475425,
+      "learning_rate": 3.976752381791088e-05,
+      "loss": 0.7168,
+      "step": 863
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.34944200203246967,
+      "learning_rate": 3.976664243962872e-05,
+      "loss": 0.6816,
+      "step": 864
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.38709976301742705,
+      "learning_rate": 3.976575940354213e-05,
+      "loss": 0.7337,
+      "step": 865
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.41970763692016316,
+      "learning_rate": 3.976487470972517e-05,
+      "loss": 0.7342,
+      "step": 866
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.332423502295157,
+      "learning_rate": 3.976398835825204e-05,
+      "loss": 0.6847,
+      "step": 867
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5010092728667959,
+      "learning_rate": 3.976310034919707e-05,
+      "loss": 0.7703,
+      "step": 868
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4360798790718183,
+      "learning_rate": 3.9762210682634746e-05,
+      "loss": 0.7402,
+      "step": 869
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4692533718582949,
+      "learning_rate": 3.976131935863968e-05,
+      "loss": 0.775,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.45851900960811653,
+      "learning_rate": 3.976042637728662e-05,
+      "loss": 0.7352,
+      "step": 871
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5000186474361137,
+      "learning_rate": 3.975953173865047e-05,
+      "loss": 0.7225,
+      "step": 872
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.43542539774910705,
+      "learning_rate": 3.975863544280625e-05,
+      "loss": 0.7137,
+      "step": 873
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.39087627220824567,
+      "learning_rate": 3.975773748982914e-05,
+      "loss": 0.7237,
+      "step": 874
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.43301237671713955,
+      "learning_rate": 3.975683787979445e-05,
+      "loss": 0.7408,
+      "step": 875
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.43971104376938497,
+      "learning_rate": 3.9755936612777626e-05,
+      "loss": 0.7039,
+      "step": 876
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4236274482084417,
+      "learning_rate": 3.975503368885425e-05,
+      "loss": 0.7396,
+      "step": 877
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4815056473502008,
+      "learning_rate": 3.975412910810007e-05,
+      "loss": 0.7117,
+      "step": 878
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.44059042227676815,
+      "learning_rate": 3.975322287059094e-05,
+      "loss": 0.7416,
+      "step": 879
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4703985694147043,
+      "learning_rate": 3.975231497640285e-05,
+      "loss": 0.7247,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.45287017853257955,
+      "learning_rate": 3.9751405425611966e-05,
+      "loss": 0.7312,
+      "step": 881
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5313857507501497,
+      "learning_rate": 3.975049421829456e-05,
+      "loss": 0.766,
+      "step": 882
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4721283471958226,
+      "learning_rate": 3.974958135452706e-05,
+      "loss": 0.7634,
+      "step": 883
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.46744343969474406,
+      "learning_rate": 3.9748666834386015e-05,
+      "loss": 0.7348,
+      "step": 884
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4500216131585384,
+      "learning_rate": 3.9747750657948145e-05,
+      "loss": 0.7064,
+      "step": 885
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3903457625622246,
+      "learning_rate": 3.974683282529027e-05,
+      "loss": 0.7244,
+      "step": 886
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4163817776078578,
+      "learning_rate": 3.974591333648937e-05,
+      "loss": 0.707,
+      "step": 887
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4133638316598582,
+      "learning_rate": 3.9744992191622574e-05,
+      "loss": 0.7202,
+      "step": 888
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4220620946969162,
+      "learning_rate": 3.974406939076712e-05,
+      "loss": 0.72,
+      "step": 889
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.38136413584370554,
+      "learning_rate": 3.974314493400042e-05,
+      "loss": 0.6932,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4659059742713764,
+      "learning_rate": 3.974221882139999e-05,
+      "loss": 0.7112,
+      "step": 891
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.41905737292826367,
+      "learning_rate": 3.974129105304352e-05,
+      "loss": 0.6965,
+      "step": 892
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.433325178233147,
+      "learning_rate": 3.97403616290088e-05,
+      "loss": 0.7271,
+      "step": 893
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.43460370656546765,
+      "learning_rate": 3.9739430549373796e-05,
+      "loss": 0.7322,
+      "step": 894
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.505041310793526,
+      "learning_rate": 3.9738497814216584e-05,
+      "loss": 0.8032,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.41860387986341047,
+      "learning_rate": 3.973756342361541e-05,
+      "loss": 0.7441,
+      "step": 896
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4405678894676713,
+      "learning_rate": 3.973662737764862e-05,
+      "loss": 0.7818,
+      "step": 897
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4000780578439628,
+      "learning_rate": 3.973568967639473e-05,
+      "loss": 0.7561,
+      "step": 898
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.44927315822300035,
+      "learning_rate": 3.973475031993238e-05,
+      "loss": 0.7283,
+      "step": 899
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.41423390906740903,
+      "learning_rate": 3.973380930834036e-05,
+      "loss": 0.6842,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.42842145906680007,
+      "learning_rate": 3.9732866641697586e-05,
+      "loss": 0.7522,
+      "step": 901
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.45652739532581693,
+      "learning_rate": 3.973192232008311e-05,
+      "loss": 0.7313,
+      "step": 902
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4313279202931235,
+      "learning_rate": 3.9730976343576144e-05,
+      "loss": 0.7073,
+      "step": 903
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48227339722542506,
+      "learning_rate": 3.973002871225602e-05,
+      "loss": 0.7676,
+      "step": 904
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4770798787265274,
+      "learning_rate": 3.972907942620221e-05,
+      "loss": 0.7603,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4846724060816875,
+      "learning_rate": 3.972812848549435e-05,
+      "loss": 0.7135,
+      "step": 906
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4530752316551426,
+      "learning_rate": 3.9727175890212166e-05,
+      "loss": 0.7414,
+      "step": 907
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5546816579049781,
+      "learning_rate": 3.972622164043557e-05,
+      "loss": 0.7158,
+      "step": 908
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.46181423963143226,
+      "learning_rate": 3.972526573624459e-05,
+      "loss": 0.7169,
+      "step": 909
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.40541344321778694,
+      "learning_rate": 3.9724308177719395e-05,
+      "loss": 0.712,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4912375850753469,
+      "learning_rate": 3.972334896494029e-05,
+      "loss": 0.7265,
+      "step": 911
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.45104638201441755,
+      "learning_rate": 3.972238809798773e-05,
+      "loss": 0.7077,
+      "step": 912
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.39970824792937104,
+      "learning_rate": 3.972142557694229e-05,
+      "loss": 0.7011,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.47405858819369473,
+      "learning_rate": 3.9720461401884716e-05,
+      "loss": 0.7867,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.44415595474239583,
+      "learning_rate": 3.971949557289586e-05,
+      "loss": 0.7545,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4686034949315643,
+      "learning_rate": 3.971852809005672e-05,
+      "loss": 0.789,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4703443011654633,
+      "learning_rate": 3.971755895344844e-05,
+      "loss": 0.7295,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.38359028031732656,
+      "learning_rate": 3.971658816315231e-05,
+      "loss": 0.7411,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3986235810419239,
+      "learning_rate": 3.971561571924974e-05,
+      "loss": 0.7152,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.42477250157263297,
+      "learning_rate": 3.971464162182229e-05,
+      "loss": 0.6966,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.417061120001355,
+      "learning_rate": 3.9713665870951657e-05,
+      "loss": 0.7085,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48123306139693345,
+      "learning_rate": 3.971268846671968e-05,
+      "loss": 0.7444,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4494475860011618,
+      "learning_rate": 3.971170940920832e-05,
+      "loss": 0.7274,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4519045143460999,
+      "learning_rate": 3.97107286984997e-05,
+      "loss": 0.7752,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.43990687006494006,
+      "learning_rate": 3.970974633467607e-05,
+      "loss": 0.7676,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4382721677527796,
+      "learning_rate": 3.970876231781982e-05,
+      "loss": 0.7557,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.44643600032887526,
+      "learning_rate": 3.970777664801347e-05,
+      "loss": 0.7643,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5086727768018396,
+      "learning_rate": 3.97067893253397e-05,
+      "loss": 0.7386,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.42726180847894085,
+      "learning_rate": 3.9705800349881306e-05,
+      "loss": 0.7576,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4151142446024772,
+      "learning_rate": 3.970480972172124e-05,
+      "loss": 0.7466,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4139485751639977,
+      "learning_rate": 3.9703817440942575e-05,
+      "loss": 0.6556,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4466694965437757,
+      "learning_rate": 3.970282350762854e-05,
+      "loss": 0.6755,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.45946242595348613,
+      "learning_rate": 3.970182792186249e-05,
+      "loss": 0.7412,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4623779165635829,
+      "learning_rate": 3.970083068372792e-05,
+      "loss": 0.6962,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.407286830571221,
+      "learning_rate": 3.969983179330848e-05,
+      "loss": 0.6905,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4709036294004165,
+      "learning_rate": 3.9698831250687944e-05,
+      "loss": 0.7256,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4749261749308425,
+      "learning_rate": 3.9697829055950216e-05,
+      "loss": 0.7101,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4199514501825004,
+      "learning_rate": 3.969682520917936e-05,
+      "loss": 0.7315,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.474694565644087,
+      "learning_rate": 3.969581971045955e-05,
+      "loss": 0.7878,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5293434740727213,
+      "learning_rate": 3.9694812559875135e-05,
+      "loss": 0.6391,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4058719205020863,
+      "learning_rate": 3.9693803757510575e-05,
+      "loss": 0.7032,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4390387924811561,
+      "learning_rate": 3.9692793303450476e-05,
+      "loss": 0.7381,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.375613667353272,
+      "learning_rate": 3.9691781197779584e-05,
+      "loss": 0.7053,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.41775662131900343,
+      "learning_rate": 3.9690767440582784e-05,
+      "loss": 0.7824,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3942624898282956,
+      "learning_rate": 3.9689752031945105e-05,
+      "loss": 0.7287,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3623048292061834,
+      "learning_rate": 3.9688734971951704e-05,
+      "loss": 0.7269,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.43311375122189794,
+      "learning_rate": 3.968771626068787e-05,
+      "loss": 0.7035,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.41951178389294425,
+      "learning_rate": 3.968669589823905e-05,
+      "loss": 0.7477,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.41191834679356826,
+      "learning_rate": 3.968567388469083e-05,
+      "loss": 0.716,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.39900234159682135,
+      "learning_rate": 3.9684650220128914e-05,
+      "loss": 0.7276,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.41658155929847546,
+      "learning_rate": 3.968362490463915e-05,
+      "loss": 0.6924,
+      "step": 951
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.42066236116418587,
+      "learning_rate": 3.968259793830754e-05,
+      "loss": 0.7969,
+      "step": 952
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.39759763661324143,
+      "learning_rate": 3.9681569321220215e-05,
+      "loss": 0.6932,
+      "step": 953
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.41308734681734643,
+      "learning_rate": 3.968053905346343e-05,
+      "loss": 0.7,
+      "step": 954
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4189671109827754,
+      "learning_rate": 3.967950713512361e-05,
+      "loss": 0.7181,
+      "step": 955
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4669129342914768,
+      "learning_rate": 3.96784735662873e-05,
+      "loss": 0.7183,
+      "step": 956
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.43742418438334385,
+      "learning_rate": 3.967743834704117e-05,
+      "loss": 0.7496,
+      "step": 957
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.43874806706732106,
+      "learning_rate": 3.9676401477472053e-05,
+      "loss": 0.7692,
+      "step": 958
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4402714107426946,
+      "learning_rate": 3.967536295766691e-05,
+      "loss": 0.7271,
+      "step": 959
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4442933225220601,
+      "learning_rate": 3.9674322787712825e-05,
+      "loss": 0.7384,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.422943020611276,
+      "learning_rate": 3.967328096769706e-05,
+      "loss": 0.7401,
+      "step": 961
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45670157905897796,
+      "learning_rate": 3.967223749770697e-05,
+      "loss": 0.6873,
+      "step": 962
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5079394691820016,
+      "learning_rate": 3.967119237783009e-05,
+      "loss": 0.7534,
+      "step": 963
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44319297598418217,
+      "learning_rate": 3.967014560815405e-05,
+      "loss": 0.6948,
+      "step": 964
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3779427415073587,
+      "learning_rate": 3.966909718876666e-05,
+      "loss": 0.6976,
+      "step": 965
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.394139512963898,
+      "learning_rate": 3.966804711975585e-05,
+      "loss": 0.7147,
+      "step": 966
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.36566374645522026,
+      "learning_rate": 3.966699540120967e-05,
+      "loss": 0.6929,
+      "step": 967
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.35026692497821277,
+      "learning_rate": 3.966594203321634e-05,
+      "loss": 0.6983,
+      "step": 968
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3729967296282797,
+      "learning_rate": 3.966488701586421e-05,
+      "loss": 0.705,
+      "step": 969
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3991430988040831,
+      "learning_rate": 3.966383034924175e-05,
+      "loss": 0.6813,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.47608437372902906,
+      "learning_rate": 3.966277203343758e-05,
+      "loss": 0.7688,
+      "step": 971
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.41118472370789394,
+      "learning_rate": 3.966171206854047e-05,
+      "loss": 0.7461,
+      "step": 972
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.38763864082456556,
+      "learning_rate": 3.966065045463932e-05,
+      "loss": 0.6653,
+      "step": 973
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4028356578759592,
+      "learning_rate": 3.9659587191823154e-05,
+      "loss": 0.6941,
+      "step": 974
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4269524873078637,
+      "learning_rate": 3.965852228018116e-05,
+      "loss": 0.7322,
+      "step": 975
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.43809543902710396,
+      "learning_rate": 3.965745571980264e-05,
+      "loss": 0.7181,
+      "step": 976
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4179196198683105,
+      "learning_rate": 3.965638751077705e-05,
+      "loss": 0.6743,
+      "step": 977
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6426393439880983,
+      "learning_rate": 3.965531765319397e-05,
+      "loss": 0.6804,
+      "step": 978
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45382153666946157,
+      "learning_rate": 3.9654246147143145e-05,
+      "loss": 0.7109,
+      "step": 979
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4149436197567412,
+      "learning_rate": 3.9653172992714434e-05,
+      "loss": 0.7086,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.41642384822248635,
+      "learning_rate": 3.9652098189997825e-05,
+      "loss": 0.6882,
+      "step": 981
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.43047380565256227,
+      "learning_rate": 3.965102173908349e-05,
+      "loss": 0.7442,
+      "step": 982
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.39490391960325855,
+      "learning_rate": 3.9649943640061686e-05,
+      "loss": 0.7094,
+      "step": 983
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4349462381095686,
+      "learning_rate": 3.964886389302284e-05,
+      "loss": 0.7178,
+      "step": 984
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4543624596208098,
+      "learning_rate": 3.9647782498057515e-05,
+      "loss": 0.7459,
+      "step": 985
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46750949698936495,
+      "learning_rate": 3.9646699455256396e-05,
+      "loss": 0.6914,
+      "step": 986
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44968250655655034,
+      "learning_rate": 3.964561476471033e-05,
+      "loss": 0.6958,
+      "step": 987
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4191960679579849,
+      "learning_rate": 3.964452842651026e-05,
+      "loss": 0.7014,
+      "step": 988
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5042959072135964,
+      "learning_rate": 3.964344044074733e-05,
+      "loss": 0.6597,
+      "step": 989
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4405582730348391,
+      "learning_rate": 3.964235080751277e-05,
+      "loss": 0.7075,
+      "step": 990
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.41529253309612935,
+      "learning_rate": 3.9641259526897974e-05,
+      "loss": 0.7403,
+      "step": 991
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.36260187487236095,
+      "learning_rate": 3.9640166598994454e-05,
+      "loss": 0.7283,
+      "step": 992
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4490442071688043,
+      "learning_rate": 3.963907202389389e-05,
+      "loss": 0.7426,
+      "step": 993
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4366215395748644,
+      "learning_rate": 3.963797580168806e-05,
+      "loss": 0.7189,
+      "step": 994
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.415308664942708,
+      "learning_rate": 3.963687793246893e-05,
+      "loss": 0.7354,
+      "step": 995
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4357284889464029,
+      "learning_rate": 3.9635778416328556e-05,
+      "loss": 0.71,
+      "step": 996
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45546338329475544,
+      "learning_rate": 3.963467725335916e-05,
+      "loss": 0.7324,
+      "step": 997
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3818194280527735,
+      "learning_rate": 3.96335744436531e-05,
+      "loss": 0.7065,
+      "step": 998
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.43291397441805046,
+      "learning_rate": 3.963246998730286e-05,
+      "loss": 0.7404,
+      "step": 999
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4050957716871353,
+      "learning_rate": 3.963136388440107e-05,
+      "loss": 0.6806,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44271537485423185,
+      "learning_rate": 3.96302561350405e-05,
+      "loss": 0.7247,
+      "step": 1001
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4353507191749679,
+      "learning_rate": 3.9629146739314055e-05,
+      "loss": 0.747,
+      "step": 1002
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44831602442169327,
+      "learning_rate": 3.962803569731478e-05,
+      "loss": 0.7314,
+      "step": 1003
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4136497618395222,
+      "learning_rate": 3.962692300913586e-05,
+      "loss": 0.6885,
+      "step": 1004
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4324932256887879,
+      "learning_rate": 3.962580867487061e-05,
+      "loss": 0.7629,
+      "step": 1005
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3917092264888668,
+      "learning_rate": 3.962469269461249e-05,
+      "loss": 0.7017,
+      "step": 1006
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.41365238123247394,
+      "learning_rate": 3.962357506845509e-05,
+      "loss": 0.7257,
+      "step": 1007
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4034297807614838,
+      "learning_rate": 3.9622455796492144e-05,
+      "loss": 0.7446,
+      "step": 1008
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.43218765178673746,
+      "learning_rate": 3.962133487881754e-05,
+      "loss": 0.716,
+      "step": 1009
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4280731751488442,
+      "learning_rate": 3.9620212315525265e-05,
+      "loss": 0.7266,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.40709236844403746,
+      "learning_rate": 3.961908810670948e-05,
+      "loss": 0.7182,
+      "step": 1011
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.42438232655112995,
+      "learning_rate": 3.9617962252464474e-05,
+      "loss": 0.7577,
+      "step": 1012
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.40262380928500935,
+      "learning_rate": 3.961683475288466e-05,
+      "loss": 0.7196,
+      "step": 1013
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4473350286336194,
+      "learning_rate": 3.961570560806461e-05,
+      "loss": 0.7361,
+      "step": 1014
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46113095779546354,
+      "learning_rate": 3.961457481809902e-05,
+      "loss": 0.7219,
+      "step": 1015
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3788683500029072,
+      "learning_rate": 3.961344238308274e-05,
+      "loss": 0.7069,
+      "step": 1016
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.42400282584448484,
+      "learning_rate": 3.9612308303110715e-05,
+      "loss": 0.7136,
+      "step": 1017
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3980686964232055,
+      "learning_rate": 3.961117257827809e-05,
+      "loss": 0.7309,
+      "step": 1018
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.47105047383409754,
+      "learning_rate": 3.96100352086801e-05,
+      "loss": 0.7492,
+      "step": 1019
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45195492764840706,
+      "learning_rate": 3.960889619441214e-05,
+      "loss": 0.7592,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5215143397811902,
+      "learning_rate": 3.9607755535569745e-05,
+      "loss": 0.7533,
+      "step": 1021
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3752919475859752,
+      "learning_rate": 3.9606613232248564e-05,
+      "loss": 0.7418,
+      "step": 1022
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3963384715505007,
+      "learning_rate": 3.9605469284544416e-05,
+      "loss": 0.7732,
+      "step": 1023
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4458655963669448,
+      "learning_rate": 3.960432369255323e-05,
+      "loss": 0.7578,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4446617456894019,
+      "learning_rate": 3.96031764563711e-05,
+      "loss": 0.7142,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.47174889051670715,
+      "learning_rate": 3.960202757609423e-05,
+      "loss": 0.7763,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5517210381263696,
+      "learning_rate": 3.960087705181899e-05,
+      "loss": 0.7709,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4508494431057654,
+      "learning_rate": 3.959972488364185e-05,
+      "loss": 0.7479,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4227229894818293,
+      "learning_rate": 3.959857107165946e-05,
+      "loss": 0.7341,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4412033153518765,
+      "learning_rate": 3.959741561596858e-05,
+      "loss": 0.728,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45930456572170786,
+      "learning_rate": 3.959625851666613e-05,
+      "loss": 0.7654,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.45943331630359585,
+      "learning_rate": 3.959509977384914e-05,
+      "loss": 0.7161,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.41379189073960415,
+      "learning_rate": 3.959393938761479e-05,
+      "loss": 0.7038,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.40050310763831526,
+      "learning_rate": 3.959277735806041e-05,
+      "loss": 0.6838,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3927702277303905,
+      "learning_rate": 3.959161368528346e-05,
+      "loss": 0.6959,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4135407739473675,
+      "learning_rate": 3.959044836938153e-05,
+      "loss": 0.6974,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4201999745319598,
+      "learning_rate": 3.958928141045235e-05,
+      "loss": 0.7221,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4450534983752317,
+      "learning_rate": 3.95881128085938e-05,
+      "loss": 0.7425,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.41778232943476057,
+      "learning_rate": 3.9586942563903884e-05,
+      "loss": 0.7337,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4006565163931187,
+      "learning_rate": 3.958577067648075e-05,
+      "loss": 0.7229,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4064465831023442,
+      "learning_rate": 3.958459714642268e-05,
+      "loss": 0.694,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.36117090173408206,
+      "learning_rate": 3.95834219738281e-05,
+      "loss": 0.7135,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.40920633761906394,
+      "learning_rate": 3.9582245158795576e-05,
+      "loss": 0.7258,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4041623541492188,
+      "learning_rate": 3.9581066701423796e-05,
+      "loss": 0.7082,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3885221076763363,
+      "learning_rate": 3.95798866018116e-05,
+      "loss": 0.6962,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4475510476616913,
+      "learning_rate": 3.957870486005796e-05,
+      "loss": 0.7642,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.428285693360841,
+      "learning_rate": 3.957752147626199e-05,
+      "loss": 0.7535,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.39470178230307634,
+      "learning_rate": 3.957633645052294e-05,
+      "loss": 0.7118,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4495057369649759,
+      "learning_rate": 3.9575149782940194e-05,
+      "loss": 0.7469,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3899947208952405,
+      "learning_rate": 3.957396147361327e-05,
+      "loss": 0.701,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.42387925298056117,
+      "learning_rate": 3.957277152264184e-05,
+      "loss": 0.7278,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4599101151424267,
+      "learning_rate": 3.95715799301257e-05,
+      "loss": 0.7744,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.424089278474961,
+      "learning_rate": 3.957038669616479e-05,
+      "loss": 0.7766,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44713387676891014,
+      "learning_rate": 3.956919182085918e-05,
+      "loss": 0.7269,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4043333879039454,
+      "learning_rate": 3.9567995304309086e-05,
+      "loss": 0.7063,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.382902809540198,
+      "learning_rate": 3.956679714661486e-05,
+      "loss": 0.7104,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.43112031973956233,
+      "learning_rate": 3.9565597347876985e-05,
+      "loss": 0.7321,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4197577759603607,
+      "learning_rate": 3.956439590819609e-05,
+      "loss": 0.7489,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44563136184581337,
+      "learning_rate": 3.956319282767294e-05,
+      "loss": 0.7697,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3989487727314978,
+      "learning_rate": 3.956198810640843e-05,
+      "loss": 0.7498,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.402691448883414,
+      "learning_rate": 3.956078174450361e-05,
+      "loss": 0.7267,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.601026080488942,
+      "learning_rate": 3.955957374205964e-05,
+      "loss": 0.6905,
+      "step": 1062
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.49656995906363116,
+      "learning_rate": 3.955836409917785e-05,
+      "loss": 0.7355,
+      "step": 1063
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.41332099318482013,
+      "learning_rate": 3.9557152815959676e-05,
+      "loss": 0.6969,
+      "step": 1064
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.35617519076891135,
+      "learning_rate": 3.955593989250672e-05,
+      "loss": 0.7181,
+      "step": 1065
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3954602973018096,
+      "learning_rate": 3.95547253289207e-05,
+      "loss": 0.7496,
+      "step": 1066
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3865225755891078,
+      "learning_rate": 3.9553509125303486e-05,
+      "loss": 0.721,
+      "step": 1067
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.399718120485215,
+      "learning_rate": 3.955229128175708e-05,
+      "loss": 0.7211,
+      "step": 1068
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.36636262437102846,
+      "learning_rate": 3.9551071798383605e-05,
+      "loss": 0.6987,
+      "step": 1069
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3862292690923536,
+      "learning_rate": 3.954985067528536e-05,
+      "loss": 0.7414,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.39194250451612617,
+      "learning_rate": 3.9548627912564746e-05,
+      "loss": 0.7612,
+      "step": 1071
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4311821944674314,
+      "learning_rate": 3.9547403510324315e-05,
+      "loss": 0.7071,
+      "step": 1072
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45087943561011185,
+      "learning_rate": 3.954617746866676e-05,
+      "loss": 0.7119,
+      "step": 1073
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44449724199206403,
+      "learning_rate": 3.954494978769491e-05,
+      "loss": 0.7525,
+      "step": 1074
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.42085147612526314,
+      "learning_rate": 3.9543720467511724e-05,
+      "loss": 0.7267,
+      "step": 1075
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.39982826992818243,
+      "learning_rate": 3.95424895082203e-05,
+      "loss": 0.6895,
+      "step": 1076
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.42158696397305157,
+      "learning_rate": 3.954125690992389e-05,
+      "loss": 0.6448,
+      "step": 1077
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.431962683406502,
+      "learning_rate": 3.9540022672725857e-05,
+      "loss": 0.7199,
+      "step": 1078
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.412131399024236,
+      "learning_rate": 3.953878679672973e-05,
+      "loss": 0.7423,
+      "step": 1079
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4941540158770006,
+      "learning_rate": 3.953754928203914e-05,
+      "loss": 0.777,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.40748821203571606,
+      "learning_rate": 3.9536310128757895e-05,
+      "loss": 0.6954,
+      "step": 1081
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4586778127160596,
+      "learning_rate": 3.9535069336989905e-05,
+      "loss": 0.7602,
+      "step": 1082
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.39652589684411976,
+      "learning_rate": 3.953382690683925e-05,
+      "loss": 0.7249,
+      "step": 1083
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4055430124727345,
+      "learning_rate": 3.9532582838410124e-05,
+      "loss": 0.699,
+      "step": 1084
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4048015370109496,
+      "learning_rate": 3.953133713180686e-05,
+      "loss": 0.7196,
+      "step": 1085
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4062109412034345,
+      "learning_rate": 3.953008978713394e-05,
+      "loss": 0.6836,
+      "step": 1086
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.37113406697464413,
+      "learning_rate": 3.952884080449598e-05,
+      "loss": 0.6808,
+      "step": 1087
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4457834861462244,
+      "learning_rate": 3.952759018399772e-05,
+      "loss": 0.7405,
+      "step": 1088
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6629332397746661,
+      "learning_rate": 3.9526337925744056e-05,
+      "loss": 0.6358,
+      "step": 1089
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44160425958072724,
+      "learning_rate": 3.952508402984001e-05,
+      "loss": 0.7458,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4323566076095308,
+      "learning_rate": 3.952382849639075e-05,
+      "loss": 0.7394,
+      "step": 1091
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3891318147443604,
+      "learning_rate": 3.9522571325501576e-05,
+      "loss": 0.6857,
+      "step": 1092
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.41673252608119365,
+      "learning_rate": 3.952131251727792e-05,
+      "loss": 0.733,
+      "step": 1093
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.40987864108676436,
+      "learning_rate": 3.9520052071825355e-05,
+      "loss": 0.7217,
+      "step": 1094
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4724487328825405,
+      "learning_rate": 3.9518789989249605e-05,
+      "loss": 0.7499,
+      "step": 1095
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45491854735732085,
+      "learning_rate": 3.95175262696565e-05,
+      "loss": 0.7231,
+      "step": 1096
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4731763855334973,
+      "learning_rate": 3.9516260913152045e-05,
+      "loss": 0.7473,
+      "step": 1097
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44296435506294746,
+      "learning_rate": 3.951499391984236e-05,
+      "loss": 0.7396,
+      "step": 1098
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4632999622992318,
+      "learning_rate": 3.95137252898337e-05,
+      "loss": 0.7424,
+      "step": 1099
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3878895114741467,
+      "learning_rate": 3.951245502323246e-05,
+      "loss": 0.6856,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.41223404822712534,
+      "learning_rate": 3.951118312014519e-05,
+      "loss": 0.727,
+      "step": 1101
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.40840489856184103,
+      "learning_rate": 3.9509909580678555e-05,
+      "loss": 0.723,
+      "step": 1102
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.38719665871943004,
+      "learning_rate": 3.9508634404939363e-05,
+      "loss": 0.6746,
+      "step": 1103
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4515139928902047,
+      "learning_rate": 3.950735759303456e-05,
+      "loss": 0.7446,
+      "step": 1104
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.41214317391563715,
+      "learning_rate": 3.9506079145071245e-05,
+      "loss": 0.6942,
+      "step": 1105
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4714654463399092,
+      "learning_rate": 3.9504799061156625e-05,
+      "loss": 0.7415,
+      "step": 1106
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.42186937646440364,
+      "learning_rate": 3.950351734139806e-05,
+      "loss": 0.7197,
+      "step": 1107
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.40374630183988136,
+      "learning_rate": 3.950223398590306e-05,
+      "loss": 0.7096,
+      "step": 1108
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4325106345355436,
+      "learning_rate": 3.950094899477923e-05,
+      "loss": 0.7165,
+      "step": 1109
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.46257394049167866,
+      "learning_rate": 3.9499662368134376e-05,
+      "loss": 0.7209,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.48096262214025837,
+      "learning_rate": 3.949837410607638e-05,
+      "loss": 0.7684,
+      "step": 1111
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44001427020296574,
+      "learning_rate": 3.94970842087133e-05,
+      "loss": 0.7889,
+      "step": 1112
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.41805760201017866,
+      "learning_rate": 3.9495792676153314e-05,
+      "loss": 0.7046,
+      "step": 1113
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4343216649303345,
+      "learning_rate": 3.9494499508504734e-05,
+      "loss": 0.7295,
+      "step": 1114
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.42394751604213043,
+      "learning_rate": 3.949320470587603e-05,
+      "loss": 0.7543,
+      "step": 1115
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4064668417841309,
+      "learning_rate": 3.949190826837579e-05,
+      "loss": 0.7253,
+      "step": 1116
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4518689690639633,
+      "learning_rate": 3.949061019611274e-05,
+      "loss": 0.7222,
+      "step": 1117
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.43279675637418125,
+      "learning_rate": 3.9489310489195756e-05,
+      "loss": 0.7517,
+      "step": 1118
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4571184404164909,
+      "learning_rate": 3.948800914773383e-05,
+      "loss": 0.7796,
+      "step": 1119
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.42200578527010807,
+      "learning_rate": 3.948670617183612e-05,
+      "loss": 0.7293,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.38446878766962506,
+      "learning_rate": 3.9485401561611896e-05,
+      "loss": 0.741,
+      "step": 1121
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4183570796289842,
+      "learning_rate": 3.948409531717057e-05,
+      "loss": 0.6824,
+      "step": 1122
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7509856730938185,
+      "learning_rate": 3.9482787438621706e-05,
+      "loss": 0.6387,
+      "step": 1123
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4798883889609756,
+      "learning_rate": 3.9481477926074994e-05,
+      "loss": 0.6875,
+      "step": 1124
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45689509033695963,
+      "learning_rate": 3.948016677964025e-05,
+      "loss": 0.7464,
+      "step": 1125
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.47457058850106826,
+      "learning_rate": 3.9478853999427446e-05,
+      "loss": 0.7057,
+      "step": 1126
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45210619848760286,
+      "learning_rate": 3.9477539585546676e-05,
+      "loss": 0.7377,
+      "step": 1127
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.39904496706823495,
+      "learning_rate": 3.947622353810819e-05,
+      "loss": 0.7153,
+      "step": 1128
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.38223178863242674,
+      "learning_rate": 3.947490585722236e-05,
+      "loss": 0.6724,
+      "step": 1129
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3478946568188464,
+      "learning_rate": 3.947358654299969e-05,
+      "loss": 0.6997,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.38941248804153566,
+      "learning_rate": 3.947226559555083e-05,
+      "loss": 0.7157,
+      "step": 1131
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4254766661094845,
+      "learning_rate": 3.9470943014986574e-05,
+      "loss": 0.747,
+      "step": 1132
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4159042000990034,
+      "learning_rate": 3.9469618801417846e-05,
+      "loss": 0.7119,
+      "step": 1133
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3999731660209764,
+      "learning_rate": 3.94682929549557e-05,
+      "loss": 0.7714,
+      "step": 1134
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45166291154800753,
+      "learning_rate": 3.946696547571134e-05,
+      "loss": 0.7443,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3593664842900916,
+      "learning_rate": 3.946563636379609e-05,
+      "loss": 0.7133,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4082759278286714,
+      "learning_rate": 3.9464305619321436e-05,
+      "loss": 0.7483,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.43774914621200606,
+      "learning_rate": 3.946297324239896e-05,
+      "loss": 0.7477,
+      "step": 1138
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.36548517057567487,
+      "learning_rate": 3.9461639233140433e-05,
+      "loss": 0.7313,
+      "step": 1139
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.37689833640978126,
+      "learning_rate": 3.946030359165773e-05,
+      "loss": 0.6741,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.405271135145775,
+      "learning_rate": 3.945896631806287e-05,
+      "loss": 0.7244,
+      "step": 1141
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.41551145751841484,
+      "learning_rate": 3.945762741246799e-05,
+      "loss": 0.7332,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.48021176641790886,
+      "learning_rate": 3.945628687498541e-05,
+      "loss": 0.7302,
+      "step": 1143
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.42372473425362517,
+      "learning_rate": 3.945494470572755e-05,
+      "loss": 0.7311,
+      "step": 1144
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3690812267633848,
+      "learning_rate": 3.945360090480697e-05,
+      "loss": 0.6661,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.42350618646068405,
+      "learning_rate": 3.9452255472336375e-05,
+      "loss": 0.6854,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4385359622394328,
+      "learning_rate": 3.94509084084286e-05,
+      "loss": 0.7317,
+      "step": 1147
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4422229012759564,
+      "learning_rate": 3.9449559713196636e-05,
+      "loss": 0.7376,
+      "step": 1148
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.49139273825373403,
+      "learning_rate": 3.944820938675358e-05,
+      "loss": 0.7143,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3361165218362832,
+      "learning_rate": 3.9446857429212695e-05,
+      "loss": 0.7168,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4219507131321425,
+      "learning_rate": 3.9445503840687365e-05,
+      "loss": 0.7005,
+      "step": 1151
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.46238796351712247,
+      "learning_rate": 3.9444148621291107e-05,
+      "loss": 0.7555,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4338953999973827,
+      "learning_rate": 3.944279177113759e-05,
+      "loss": 0.7307,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4344426459122168,
+      "learning_rate": 3.944143329034061e-05,
+      "loss": 0.7404,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3881865013149564,
+      "learning_rate": 3.944007317901409e-05,
+      "loss": 0.7117,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3907094471013424,
+      "learning_rate": 3.943871143727212e-05,
+      "loss": 0.6704,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3889927669729832,
+      "learning_rate": 3.9437348065228897e-05,
+      "loss": 0.7614,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4170804999661633,
+      "learning_rate": 3.943598306299876e-05,
+      "loss": 0.7131,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4182323068464507,
+      "learning_rate": 3.94346164306962e-05,
+      "loss": 0.7439,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.38528219397122154,
+      "learning_rate": 3.943324816843584e-05,
+      "loss": 0.6858,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3905624980558444,
+      "learning_rate": 3.9431878276332414e-05,
+      "loss": 0.7221,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.43812655504103926,
+      "learning_rate": 3.943050675450082e-05,
+      "loss": 0.6952,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.40790194075058284,
+      "learning_rate": 3.942913360305611e-05,
+      "loss": 0.7084,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.39540830098490043,
+      "learning_rate": 3.942775882211341e-05,
+      "loss": 0.7043,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.40686083047990984,
+      "learning_rate": 3.942638241178805e-05,
+      "loss": 0.748,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.38484036010664735,
+      "learning_rate": 3.942500437219545e-05,
+      "loss": 0.7342,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3957766463279013,
+      "learning_rate": 3.9423624703451205e-05,
+      "loss": 0.6953,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.36811350181037816,
+      "learning_rate": 3.942224340567101e-05,
+      "loss": 0.7118,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44011792775260183,
+      "learning_rate": 3.9420860478970716e-05,
+      "loss": 0.736,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4120718793840243,
+      "learning_rate": 3.94194759234663e-05,
+      "loss": 0.7398,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.37773131596523596,
+      "learning_rate": 3.9418089739273895e-05,
+      "loss": 0.6987,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3956365398404693,
+      "learning_rate": 3.941670192650976e-05,
+      "loss": 0.7506,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4115880317336099,
+      "learning_rate": 3.941531248529028e-05,
+      "loss": 0.7458,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.43341178839405503,
+      "learning_rate": 3.941392141573199e-05,
+      "loss": 0.6915,
+      "step": 1174
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43505582553049577,
+      "learning_rate": 3.9412528717951556e-05,
+      "loss": 0.7465,
+      "step": 1175
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3672893991277072,
+      "learning_rate": 3.941113439206579e-05,
+      "loss": 0.6822,
+      "step": 1176
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.42187761757696457,
+      "learning_rate": 3.940973843819162e-05,
+      "loss": 0.7113,
+      "step": 1177
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3989672972782668,
+      "learning_rate": 3.940834085644612e-05,
+      "loss": 0.7682,
+      "step": 1178
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.37362292903278826,
+      "learning_rate": 3.940694164694653e-05,
+      "loss": 0.7009,
+      "step": 1179
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.428833289908656,
+      "learning_rate": 3.940554080981016e-05,
+      "loss": 0.7138,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3924377922878619,
+      "learning_rate": 3.940413834515454e-05,
+      "loss": 0.7247,
+      "step": 1181
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.40892577436210803,
+      "learning_rate": 3.940273425309726e-05,
+      "loss": 0.7171,
+      "step": 1182
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45490370092847676,
+      "learning_rate": 3.940132853375609e-05,
+      "loss": 0.7434,
+      "step": 1183
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4062790075799036,
+      "learning_rate": 3.9399921187248925e-05,
+      "loss": 0.7096,
+      "step": 1184
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4421779392917502,
+      "learning_rate": 3.939851221369381e-05,
+      "loss": 0.7227,
+      "step": 1185
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.39360050280009545,
+      "learning_rate": 3.939710161320889e-05,
+      "loss": 0.7386,
+      "step": 1186
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41209410193642054,
+      "learning_rate": 3.939568938591249e-05,
+      "loss": 0.7112,
+      "step": 1187
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4080242794590702,
+      "learning_rate": 3.939427553192304e-05,
+      "loss": 0.7479,
+      "step": 1188
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4627408217203887,
+      "learning_rate": 3.9392860051359135e-05,
+      "loss": 0.7607,
+      "step": 1189
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.34284847200466406,
+      "learning_rate": 3.939144294433947e-05,
+      "loss": 0.6984,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.40059640756217973,
+      "learning_rate": 3.93900242109829e-05,
+      "loss": 0.6945,
+      "step": 1191
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3822844443261866,
+      "learning_rate": 3.938860385140842e-05,
+      "loss": 0.7407,
+      "step": 1192
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45037058363286275,
+      "learning_rate": 3.938718186573516e-05,
+      "loss": 0.7175,
+      "step": 1193
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.40193970977611043,
+      "learning_rate": 3.938575825408236e-05,
+      "loss": 0.6764,
+      "step": 1194
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43473583697291185,
+      "learning_rate": 3.9384333016569433e-05,
+      "loss": 0.7697,
+      "step": 1195
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4506733759781442,
+      "learning_rate": 3.93829061533159e-05,
+      "loss": 0.7801,
+      "step": 1196
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41302935312435973,
+      "learning_rate": 3.9381477664441436e-05,
+      "loss": 0.7779,
+      "step": 1197
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4337482869652652,
+      "learning_rate": 3.938004755006586e-05,
+      "loss": 0.721,
+      "step": 1198
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.40630778020862385,
+      "learning_rate": 3.9378615810309085e-05,
+      "loss": 0.7276,
+      "step": 1199
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.468572928828287,
+      "learning_rate": 3.9377182445291214e-05,
+      "loss": 0.7469,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4370389710325975,
+      "learning_rate": 3.937574745513245e-05,
+      "loss": 0.7257,
+      "step": 1201
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.48381124357851085,
+      "learning_rate": 3.937431083995315e-05,
+      "loss": 0.7656,
+      "step": 1202
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6311837695008448,
+      "learning_rate": 3.93728725998738e-05,
+      "loss": 0.703,
+      "step": 1203
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4413216199751347,
+      "learning_rate": 3.9371432735015017e-05,
+      "loss": 0.7204,
+      "step": 1204
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4127514926222138,
+      "learning_rate": 3.9369991245497564e-05,
+      "loss": 0.7228,
+      "step": 1205
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4966462297973762,
+      "learning_rate": 3.936854813144234e-05,
+      "loss": 0.7494,
+      "step": 1206
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.40007251132127697,
+      "learning_rate": 3.9367103392970385e-05,
+      "loss": 0.6741,
+      "step": 1207
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.47708757814488717,
+      "learning_rate": 3.936565703020285e-05,
+      "loss": 0.7906,
+      "step": 1208
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44003527211221016,
+      "learning_rate": 3.936420904326104e-05,
+      "loss": 0.6942,
+      "step": 1209
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.40500606582200016,
+      "learning_rate": 3.9362759432266415e-05,
+      "loss": 0.7093,
+      "step": 1210
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41255356886240513,
+      "learning_rate": 3.936130819734054e-05,
+      "loss": 0.6891,
+      "step": 1211
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46491301613469843,
+      "learning_rate": 3.935985533860512e-05,
+      "loss": 0.7943,
+      "step": 1212
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.420257734142064,
+      "learning_rate": 3.935840085618202e-05,
+      "loss": 0.7542,
+      "step": 1213
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3976391506911525,
+      "learning_rate": 3.935694475019322e-05,
+      "loss": 0.7518,
+      "step": 1214
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4066082522217229,
+      "learning_rate": 3.9355487020760836e-05,
+      "loss": 0.7357,
+      "step": 1215
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41039898219443904,
+      "learning_rate": 3.935402766800714e-05,
+      "loss": 0.7006,
+      "step": 1216
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3594571771174962,
+      "learning_rate": 3.935256669205451e-05,
+      "loss": 0.7125,
+      "step": 1217
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.40923398611264694,
+      "learning_rate": 3.935110409302548e-05,
+      "loss": 0.6885,
+      "step": 1218
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4283172139409865,
+      "learning_rate": 3.934963987104273e-05,
+      "loss": 0.7475,
+      "step": 1219
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45318335125518283,
+      "learning_rate": 3.9348174026229045e-05,
+      "loss": 0.7489,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4267055323696375,
+      "learning_rate": 3.9346706558707374e-05,
+      "loss": 0.7434,
+      "step": 1221
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4102250378604716,
+      "learning_rate": 3.934523746860078e-05,
+      "loss": 0.7003,
+      "step": 1222
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41847581776968196,
+      "learning_rate": 3.93437667560325e-05,
+      "loss": 0.7224,
+      "step": 1223
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.40303708608361966,
+      "learning_rate": 3.934229442112585e-05,
+      "loss": 0.693,
+      "step": 1224
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46487897606928186,
+      "learning_rate": 3.934082046400433e-05,
+      "loss": 0.7673,
+      "step": 1225
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.417067772068026,
+      "learning_rate": 3.933934488479156e-05,
+      "loss": 0.6947,
+      "step": 1226
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.36692260569837337,
+      "learning_rate": 3.933786768361129e-05,
+      "loss": 0.6881,
+      "step": 1227
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8829607719686574,
+      "learning_rate": 3.9336388860587404e-05,
+      "loss": 0.7564,
+      "step": 1228
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46115703475688774,
+      "learning_rate": 3.9334908415843944e-05,
+      "loss": 0.779,
+      "step": 1229
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43003128062206986,
+      "learning_rate": 3.933342634950506e-05,
+      "loss": 0.6797,
+      "step": 1230
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4455501391102957,
+      "learning_rate": 3.933194266169507e-05,
+      "loss": 0.7315,
+      "step": 1231
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4899661151622647,
+      "learning_rate": 3.933045735253838e-05,
+      "loss": 0.7223,
+      "step": 1232
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.49790350221465174,
+      "learning_rate": 3.932897042215959e-05,
+      "loss": 0.802,
+      "step": 1233
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44604827533644165,
+      "learning_rate": 3.932748187068338e-05,
+      "loss": 0.7168,
+      "step": 1234
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4205146741817137,
+      "learning_rate": 3.932599169823461e-05,
+      "loss": 0.7128,
+      "step": 1235
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3795028277361502,
+      "learning_rate": 3.932449990493827e-05,
+      "loss": 0.6932,
+      "step": 1236
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4320227345326029,
+      "learning_rate": 3.9323006490919445e-05,
+      "loss": 0.7149,
+      "step": 1237
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45383346752690107,
+      "learning_rate": 3.9321511456303414e-05,
+      "loss": 0.7848,
+      "step": 1238
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4647706700107513,
+      "learning_rate": 3.932001480121555e-05,
+      "loss": 0.7465,
+      "step": 1239
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4265286591926267,
+      "learning_rate": 3.931851652578137e-05,
+      "loss": 0.7347,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.34294230771036416,
+      "learning_rate": 3.9317016630126546e-05,
+      "loss": 0.6679,
+      "step": 1241
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4171315669978336,
+      "learning_rate": 3.931551511437686e-05,
+      "loss": 0.7528,
+      "step": 1242
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.37944863685026325,
+      "learning_rate": 3.9314011978658256e-05,
+      "loss": 0.7408,
+      "step": 1243
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41888528111006995,
+      "learning_rate": 3.931250722309679e-05,
+      "loss": 0.7464,
+      "step": 1244
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4135713201128789,
+      "learning_rate": 3.931100084781866e-05,
+      "loss": 0.7028,
+      "step": 1245
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41311616421819086,
+      "learning_rate": 3.9309492852950216e-05,
+      "loss": 0.71,
+      "step": 1246
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3894650186915157,
+      "learning_rate": 3.930798323861793e-05,
+      "loss": 0.7096,
+      "step": 1247
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4074978059102712,
+      "learning_rate": 3.9306472004948404e-05,
+      "loss": 0.6947,
+      "step": 1248
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.382657702379941,
+      "learning_rate": 3.930495915206839e-05,
+      "loss": 0.7178,
+      "step": 1249
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4135339656669443,
+      "learning_rate": 3.9303444680104764e-05,
+      "loss": 0.7127,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41408189067176554,
+      "learning_rate": 3.9301928589184545e-05,
+      "loss": 0.7004,
+      "step": 1251
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44827025874748044,
+      "learning_rate": 3.9300410879434886e-05,
+      "loss": 0.7477,
+      "step": 1252
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4691300652793255,
+      "learning_rate": 3.929889155098308e-05,
+      "loss": 0.7328,
+      "step": 1253
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46288193606590555,
+      "learning_rate": 3.929737060395655e-05,
+      "loss": 0.7734,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4848113405984341,
+      "learning_rate": 3.929584803848284e-05,
+      "loss": 0.7665,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43989124641124927,
+      "learning_rate": 3.929432385468966e-05,
+      "loss": 0.6912,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4208625570695992,
+      "learning_rate": 3.929279805270485e-05,
+      "loss": 0.6864,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4041601694415345,
+      "learning_rate": 3.929127063265636e-05,
+      "loss": 0.7391,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4002018599206256,
+      "learning_rate": 3.9289741594672304e-05,
+      "loss": 0.7228,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4052430785921455,
+      "learning_rate": 3.928821093888091e-05,
+      "loss": 0.7292,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4394165759558913,
+      "learning_rate": 3.928667866541056e-05,
+      "loss": 0.7405,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4094017259407289,
+      "learning_rate": 3.928514477438977e-05,
+      "loss": 0.6989,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4466252375475494,
+      "learning_rate": 3.928360926594717e-05,
+      "loss": 0.7434,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3735972642583066,
+      "learning_rate": 3.928207214021156e-05,
+      "loss": 0.6957,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3967032705986703,
+      "learning_rate": 3.928053339731184e-05,
+      "loss": 0.6765,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4624621740037121,
+      "learning_rate": 3.9278993037377066e-05,
+      "loss": 0.7477,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.34901414932851527,
+      "learning_rate": 3.927745106053643e-05,
+      "loss": 0.6864,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3743166509232723,
+      "learning_rate": 3.9275907466919255e-05,
+      "loss": 0.7277,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.34426237097461426,
+      "learning_rate": 3.927436225665501e-05,
+      "loss": 0.6312,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.38476782562121403,
+      "learning_rate": 3.9272815429873275e-05,
+      "loss": 0.725,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4164708164109382,
+      "learning_rate": 3.927126698670378e-05,
+      "loss": 0.7488,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0282502706288994,
+      "learning_rate": 3.92697169272764e-05,
+      "loss": 0.7542,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4589810063820626,
+      "learning_rate": 3.926816525172113e-05,
+      "loss": 0.7135,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4276036559847881,
+      "learning_rate": 3.9266611960168114e-05,
+      "loss": 0.6667,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4244946081044326,
+      "learning_rate": 3.926505705274762e-05,
+      "loss": 0.7048,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4420603111354717,
+      "learning_rate": 3.926350052959006e-05,
+      "loss": 0.7535,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41551017139737045,
+      "learning_rate": 3.926194239082597e-05,
+      "loss": 0.7505,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4185027263102274,
+      "learning_rate": 3.926038263658603e-05,
+      "loss": 0.7248,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43215396566772757,
+      "learning_rate": 3.925882126700107e-05,
+      "loss": 0.733,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3817074725587132,
+      "learning_rate": 3.925725828220202e-05,
+      "loss": 0.6976,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3933936708942011,
+      "learning_rate": 3.925569368231998e-05,
+      "loss": 0.727,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.42257345780688516,
+      "learning_rate": 3.9254127467486165e-05,
+      "loss": 0.7053,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.39841800037405245,
+      "learning_rate": 3.925255963783194e-05,
+      "loss": 0.6864,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3996617077018786,
+      "learning_rate": 3.925099019348878e-05,
+      "loss": 0.6787,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.416604334627095,
+      "learning_rate": 3.924941913458832e-05,
+      "loss": 0.7677,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43342841757512224,
+      "learning_rate": 3.924784646126233e-05,
+      "loss": 0.7406,
+      "step": 1286
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3810653657678319,
+      "learning_rate": 3.9246272173642704e-05,
+      "loss": 0.7307,
+      "step": 1287
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4643126363583578,
+      "learning_rate": 3.924469627186147e-05,
+      "loss": 0.7501,
+      "step": 1288
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.970545717506486,
+      "learning_rate": 3.924311875605081e-05,
+      "loss": 0.7082,
+      "step": 1289
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5084699926897248,
+      "learning_rate": 3.9241539626343006e-05,
+      "loss": 0.7122,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.45428384334491395,
+      "learning_rate": 3.923995888287052e-05,
+      "loss": 0.7278,
+      "step": 1291
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3958093849619742,
+      "learning_rate": 3.9238376525765915e-05,
+      "loss": 0.7079,
+      "step": 1292
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40169147112740444,
+      "learning_rate": 3.923679255516191e-05,
+      "loss": 0.7248,
+      "step": 1293
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4099927110585457,
+      "learning_rate": 3.9235206971191345e-05,
+      "loss": 0.6992,
+      "step": 1294
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.46437643962652064,
+      "learning_rate": 3.923361977398719e-05,
+      "loss": 0.7456,
+      "step": 1295
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4610423790366667,
+      "learning_rate": 3.9232030963682586e-05,
+      "loss": 0.709,
+      "step": 1296
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4296158213738652,
+      "learning_rate": 3.923044054041077e-05,
+      "loss": 0.6882,
+      "step": 1297
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44062080783237056,
+      "learning_rate": 3.922884850430512e-05,
+      "loss": 0.7329,
+      "step": 1298
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3879054506779071,
+      "learning_rate": 3.9227254855499174e-05,
+      "loss": 0.7172,
+      "step": 1299
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4172605150785946,
+      "learning_rate": 3.922565959412659e-05,
+      "loss": 0.73,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.36549047585593747,
+      "learning_rate": 3.922406272032115e-05,
+      "loss": 0.729,
+      "step": 1301
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4045406832401137,
+      "learning_rate": 3.922246423421679e-05,
+      "loss": 0.7554,
+      "step": 1302
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3460154745545245,
+      "learning_rate": 3.9220864135947566e-05,
+      "loss": 0.6826,
+      "step": 1303
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3975304057178129,
+      "learning_rate": 3.921926242564768e-05,
+      "loss": 0.6814,
+      "step": 1304
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4104366278070622,
+      "learning_rate": 3.9217659103451464e-05,
+      "loss": 0.6917,
+      "step": 1305
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40229764210274194,
+      "learning_rate": 3.9216054169493386e-05,
+      "loss": 0.6857,
+      "step": 1306
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4215097044050228,
+      "learning_rate": 3.921444762390806e-05,
+      "loss": 0.7237,
+      "step": 1307
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.381188652547655,
+      "learning_rate": 3.921283946683021e-05,
+      "loss": 0.6825,
+      "step": 1308
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3940207064846049,
+      "learning_rate": 3.921122969839472e-05,
+      "loss": 0.7287,
+      "step": 1309
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.35632827330593736,
+      "learning_rate": 3.9209618318736595e-05,
+      "loss": 0.6737,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3697847128104496,
+      "learning_rate": 3.9208005327990986e-05,
+      "loss": 0.7076,
+      "step": 1311
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3615879982068235,
+      "learning_rate": 3.920639072629316e-05,
+      "loss": 0.6835,
+      "step": 1312
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3825902104711276,
+      "learning_rate": 3.9204774513778544e-05,
+      "loss": 0.7006,
+      "step": 1313
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.485645695084602,
+      "learning_rate": 3.920315669058268e-05,
+      "loss": 0.7714,
+      "step": 1314
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40532267986426956,
+      "learning_rate": 3.920153725684126e-05,
+      "loss": 0.6776,
+      "step": 1315
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.357126433998723,
+      "learning_rate": 3.9199916212690096e-05,
+      "loss": 0.6919,
+      "step": 1316
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4059001829879514,
+      "learning_rate": 3.919829355826515e-05,
+      "loss": 0.7367,
+      "step": 1317
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 17.065601913253865,
+      "learning_rate": 3.919666929370251e-05,
+      "loss": 1.7949,
+      "step": 1318
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40190554846578613,
+      "learning_rate": 3.9195043419138395e-05,
+      "loss": 0.7046,
+      "step": 1319
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4124155975034453,
+      "learning_rate": 3.9193415934709176e-05,
+      "loss": 0.7322,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42801299124796066,
+      "learning_rate": 3.919178684055133e-05,
+      "loss": 0.74,
+      "step": 1321
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4829548548536914,
+      "learning_rate": 3.9190156136801516e-05,
+      "loss": 0.7425,
+      "step": 1322
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4266997052999744,
+      "learning_rate": 3.9188523823596476e-05,
+      "loss": 0.6845,
+      "step": 1323
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5747662028650811,
+      "learning_rate": 3.918688990107311e-05,
+      "loss": 0.7341,
+      "step": 1324
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5824417201352448,
+      "learning_rate": 3.918525436936847e-05,
+      "loss": 0.7852,
+      "step": 1325
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4157166310082691,
+      "learning_rate": 3.918361722861971e-05,
+      "loss": 0.7047,
+      "step": 1326
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40079548310073165,
+      "learning_rate": 3.918197847896414e-05,
+      "loss": 0.6392,
+      "step": 1327
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4681688174320125,
+      "learning_rate": 3.9180338120539204e-05,
+      "loss": 0.7361,
+      "step": 1328
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4385582416759592,
+      "learning_rate": 3.917869615348248e-05,
+      "loss": 0.7116,
+      "step": 1329
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.463391255406761,
+      "learning_rate": 3.917705257793166e-05,
+      "loss": 0.7606,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4651114052582486,
+      "learning_rate": 3.9175407394024606e-05,
+      "loss": 0.6702,
+      "step": 1331
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42212547685338897,
+      "learning_rate": 3.9173760601899286e-05,
+      "loss": 0.7146,
+      "step": 1332
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3930304225203257,
+      "learning_rate": 3.917211220169383e-05,
+      "loss": 0.7498,
+      "step": 1333
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4395292942566465,
+      "learning_rate": 3.9170462193546476e-05,
+      "loss": 0.716,
+      "step": 1334
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40360295298581617,
+      "learning_rate": 3.9168810577595606e-05,
+      "loss": 0.7165,
+      "step": 1335
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.47414766346244724,
+      "learning_rate": 3.916715735397975e-05,
+      "loss": 0.78,
+      "step": 1336
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.384643695667904,
+      "learning_rate": 3.916550252283754e-05,
+      "loss": 0.6707,
+      "step": 1337
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4204370088061283,
+      "learning_rate": 3.916384608430779e-05,
+      "loss": 0.7219,
+      "step": 1338
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4019037373509994,
+      "learning_rate": 3.916218803852941e-05,
+      "loss": 0.7304,
+      "step": 1339
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.457180478455833,
+      "learning_rate": 3.9160528385641465e-05,
+      "loss": 0.7435,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43625977777254954,
+      "learning_rate": 3.9158867125783144e-05,
+      "loss": 0.6951,
+      "step": 1341
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40714528035706893,
+      "learning_rate": 3.915720425909377e-05,
+      "loss": 0.7048,
+      "step": 1342
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4124874236513415,
+      "learning_rate": 3.915553978571281e-05,
+      "loss": 0.7537,
+      "step": 1343
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4328244845399336,
+      "learning_rate": 3.9153873705779874e-05,
+      "loss": 0.7074,
+      "step": 1344
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4606276310513356,
+      "learning_rate": 3.9152206019434674e-05,
+      "loss": 0.6993,
+      "step": 1345
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.47752926112110367,
+      "learning_rate": 3.915053672681708e-05,
+      "loss": 0.782,
+      "step": 1346
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4643906673687899,
+      "learning_rate": 3.914886582806711e-05,
+      "loss": 0.7452,
+      "step": 1347
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44257887183481315,
+      "learning_rate": 3.914719332332488e-05,
+      "loss": 0.7653,
+      "step": 1348
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4445786194785071,
+      "learning_rate": 3.9145519212730677e-05,
+      "loss": 0.751,
+      "step": 1349
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3754149545416572,
+      "learning_rate": 3.9143843496424896e-05,
+      "loss": 0.6753,
+      "step": 1350
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4575363947673363,
+      "learning_rate": 3.914216617454808e-05,
+      "loss": 0.7665,
+      "step": 1351
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.38861975292922324,
+      "learning_rate": 3.91404872472409e-05,
+      "loss": 0.7317,
+      "step": 1352
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4161666306310224,
+      "learning_rate": 3.913880671464418e-05,
+      "loss": 0.729,
+      "step": 1353
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40598900266937926,
+      "learning_rate": 3.913712457689885e-05,
+      "loss": 0.7418,
+      "step": 1354
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.378016243835612,
+      "learning_rate": 3.913544083414599e-05,
+      "loss": 0.6979,
+      "step": 1355
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4815916939344821,
+      "learning_rate": 3.913375548652683e-05,
+      "loss": 0.731,
+      "step": 1356
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.37235212428816933,
+      "learning_rate": 3.9132068534182694e-05,
+      "loss": 0.6989,
+      "step": 1357
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43986433931594565,
+      "learning_rate": 3.913037997725508e-05,
+      "loss": 0.723,
+      "step": 1358
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42676959311706897,
+      "learning_rate": 3.9128689815885594e-05,
+      "loss": 0.7255,
+      "step": 1359
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3767467574080448,
+      "learning_rate": 3.9126998050216e-05,
+      "loss": 0.7664,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4543412191523173,
+      "learning_rate": 3.912530468038818e-05,
+      "loss": 0.7898,
+      "step": 1361
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42223110284961984,
+      "learning_rate": 3.9123609706544155e-05,
+      "loss": 0.7292,
+      "step": 1362
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4051506450328164,
+      "learning_rate": 3.912191312882607e-05,
+      "loss": 0.7109,
+      "step": 1363
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.41307248355564463,
+      "learning_rate": 3.912021494737624e-05,
+      "loss": 0.7234,
+      "step": 1364
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4344353344559594,
+      "learning_rate": 3.9118515162337065e-05,
+      "loss": 0.7588,
+      "step": 1365
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3332209364750361,
+      "learning_rate": 3.911681377385111e-05,
+      "loss": 0.6802,
+      "step": 1366
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 18.04116641819322,
+      "learning_rate": 3.911511078206108e-05,
+      "loss": 2.3546,
+      "step": 1367
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 47.19524818386799,
+      "learning_rate": 3.911340618710978e-05,
+      "loss": 1.9114,
+      "step": 1368
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4078937458255927,
+      "learning_rate": 3.91116999891402e-05,
+      "loss": 0.7986,
+      "step": 1369
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0605043821364792,
+      "learning_rate": 3.910999218829541e-05,
+      "loss": 0.7314,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.700654377810319,
+      "learning_rate": 3.910828278471866e-05,
+      "loss": 0.7994,
+      "step": 1371
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.685722184519679,
+      "learning_rate": 3.910657177855331e-05,
+      "loss": 0.6936,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5560688237378115,
+      "learning_rate": 3.910485916994286e-05,
+      "loss": 0.7195,
+      "step": 1373
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 339.57523478556897,
+      "learning_rate": 3.910314495903095e-05,
+      "loss": 5.0741,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1164135014771834,
+      "learning_rate": 3.910142914596134e-05,
+      "loss": 0.7401,
+      "step": 1375
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.19721654527043,
+      "learning_rate": 3.9099711730877935e-05,
+      "loss": 0.8009,
+      "step": 1376
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.2275647816676836,
+      "learning_rate": 3.909799271392477e-05,
+      "loss": 0.7791,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2752427668674604,
+      "learning_rate": 3.9096272095246026e-05,
+      "loss": 0.7643,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 4.350997648974718,
+      "learning_rate": 3.9094549874986e-05,
+      "loss": 0.7799,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 38.541251580004435,
+      "learning_rate": 3.909282605328914e-05,
+      "loss": 1.3673,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 88.25505945637792,
+      "learning_rate": 3.9091100630300014e-05,
+      "loss": 7.0043,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 79.77542187504156,
+      "learning_rate": 3.908937360616333e-05,
+      "loss": 6.3246,
+      "step": 1382
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 29.775797736276015,
+      "learning_rate": 3.9087644981023946e-05,
+      "loss": 5.4757,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 59.176209587134174,
+      "learning_rate": 3.908591475502682e-05,
+      "loss": 8.8484,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 53.26857201786358,
+      "learning_rate": 3.908418292831709e-05,
+      "loss": 9.4012,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 23.126483263617715,
+      "learning_rate": 3.908244950103996e-05,
+      "loss": 11.7095,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 20.493942405462892,
+      "learning_rate": 3.9080714473340854e-05,
+      "loss": 7.8606,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 24.693348166193477,
+      "learning_rate": 3.907897784536527e-05,
+      "loss": 8.1133,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 43.71907570653974,
+      "learning_rate": 3.907723961725885e-05,
+      "loss": 7.328,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 17.45205107965086,
+      "learning_rate": 3.907549978916738e-05,
+      "loss": 6.0996,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 12.947540318626952,
+      "learning_rate": 3.907375836123679e-05,
+      "loss": 5.8273,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.995456188862198,
+      "learning_rate": 3.907201533361311e-05,
+      "loss": 5.3495,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.002262058142962,
+      "learning_rate": 3.907027070644254e-05,
+      "loss": 5.0306,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 7.828933861693323,
+      "learning_rate": 3.90685244798714e-05,
+      "loss": 5.1764,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.319496545349419,
+      "learning_rate": 3.9066776654046144e-05,
+      "loss": 4.6278,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.081408225336294,
+      "learning_rate": 3.9065027229113346e-05,
+      "loss": 4.4863,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 9.67281250138606,
+      "learning_rate": 3.906327620521975e-05,
+      "loss": 4.986,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 9.005532408288055,
+      "learning_rate": 3.9061523582512196e-05,
+      "loss": 4.3787,
+      "step": 1398
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.632309535861057,
+      "learning_rate": 3.905976936113768e-05,
+      "loss": 5.2062,
+      "step": 1399
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.677519358537119,
+      "learning_rate": 3.9058013541243326e-05,
+      "loss": 4.4711,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.392451474290603,
+      "learning_rate": 3.905625612297639e-05,
+      "loss": 4.6181,
+      "step": 1401
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.87463145478808,
+      "learning_rate": 3.9054497106484275e-05,
+      "loss": 4.3112,
+      "step": 1402
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.452225939444054,
+      "learning_rate": 3.90527364919145e-05,
+      "loss": 4.3792,
+      "step": 1403
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.835721666693584,
+      "learning_rate": 3.9050974279414714e-05,
+      "loss": 4.2666,
+      "step": 1404
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.760487117825942,
+      "learning_rate": 3.904921046913273e-05,
+      "loss": 4.6522,
+      "step": 1405
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.566202974373264,
+      "learning_rate": 3.9047445061216465e-05,
+      "loss": 4.2432,
+      "step": 1406
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.538748040138855,
+      "learning_rate": 3.9045678055813994e-05,
+      "loss": 4.437,
+      "step": 1407
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.6295583142887162,
+      "learning_rate": 3.90439094530735e-05,
+      "loss": 4.2611,
+      "step": 1408
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.183294017523728,
+      "learning_rate": 3.904213925314332e-05,
+      "loss": 3.9905,
+      "step": 1409
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.7022754517917313,
+      "learning_rate": 3.904036745617192e-05,
+      "loss": 4.1275,
+      "step": 1410
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4266541752884674,
+      "learning_rate": 3.903859406230789e-05,
+      "loss": 4.6405,
+      "step": 1411
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.514938915484207,
+      "learning_rate": 3.903681907169998e-05,
+      "loss": 4.0836,
+      "step": 1412
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.970114805074592,
+      "learning_rate": 3.9035042484497035e-05,
+      "loss": 4.2445,
+      "step": 1413
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.980507102019282,
+      "learning_rate": 3.903326430084807e-05,
+      "loss": 4.3315,
+      "step": 1414
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0366378885183476,
+      "learning_rate": 3.903148452090221e-05,
+      "loss": 4.3313,
+      "step": 1415
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8562300721168234,
+      "learning_rate": 3.902970314480874e-05,
+      "loss": 3.5646,
+      "step": 1416
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 9.163742087449439,
+      "learning_rate": 3.902792017271704e-05,
+      "loss": 9.4741,
+      "step": 1417
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5499266884863547,
+      "learning_rate": 3.902613560477665e-05,
+      "loss": 4.2491,
+      "step": 1418
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.968404715601953,
+      "learning_rate": 3.902434944113726e-05,
+      "loss": 4.2476,
+      "step": 1419
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.273219788744293,
+      "learning_rate": 3.902256168194864e-05,
+      "loss": 4.0545,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.5476528007339905,
+      "learning_rate": 3.902077232736075e-05,
+      "loss": 3.6932,
+      "step": 1421
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.972846384232807,
+      "learning_rate": 3.9018981377523664e-05,
+      "loss": 4.3583,
+      "step": 1422
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5494655144581486,
+      "learning_rate": 3.9017188832587576e-05,
+      "loss": 4.0575,
+      "step": 1423
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.4430408814667857,
+      "learning_rate": 3.901539469270283e-05,
+      "loss": 3.7078,
+      "step": 1424
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.569020931702631,
+      "learning_rate": 3.901359895801989e-05,
+      "loss": 3.9479,
+      "step": 1425
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.298356087123428,
+      "learning_rate": 3.901180162868937e-05,
+      "loss": 3.977,
+      "step": 1426
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.781416410697411,
+      "learning_rate": 3.9010002704862004e-05,
+      "loss": 4.1534,
+      "step": 1427
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1321805545034453,
+      "learning_rate": 3.900820218668868e-05,
+      "loss": 4.5095,
+      "step": 1428
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.081420957053082,
+      "learning_rate": 3.9006400074320385e-05,
+      "loss": 3.9597,
+      "step": 1429
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1218330315986513,
+      "learning_rate": 3.900459636790827e-05,
+      "loss": 4.546,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0839442078220656,
+      "learning_rate": 3.900279106760362e-05,
+      "loss": 4.1313,
+      "step": 1431
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.5574749527870253,
+      "learning_rate": 3.900098417355782e-05,
+      "loss": 3.9653,
+      "step": 1432
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2084669118767484,
+      "learning_rate": 3.899917568592243e-05,
+      "loss": 3.6,
+      "step": 1433
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9819313809161907,
+      "learning_rate": 3.899736560484912e-05,
+      "loss": 4.1209,
+      "step": 1434
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6780943427175603,
+      "learning_rate": 3.89955539304897e-05,
+      "loss": 4.6015,
+      "step": 1435
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6484036130697477,
+      "learning_rate": 3.8993740662996115e-05,
+      "loss": 4.2341,
+      "step": 1436
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0856386856102054,
+      "learning_rate": 3.8991925802520434e-05,
+      "loss": 3.6282,
+      "step": 1437
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9078104555256612,
+      "learning_rate": 3.8990109349214875e-05,
+      "loss": 3.9317,
+      "step": 1438
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7304677942788176,
+      "learning_rate": 3.8988291303231774e-05,
+      "loss": 3.8882,
+      "step": 1439
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3660361181354033,
+      "learning_rate": 3.8986471664723624e-05,
+      "loss": 3.8359,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1538437792378289,
+      "learning_rate": 3.8984650433843014e-05,
+      "loss": 4.2416,
+      "step": 1441
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.078868190890925,
+      "learning_rate": 3.898282761074271e-05,
+      "loss": 3.8148,
+      "step": 1442
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9043953038127436,
+      "learning_rate": 3.898100319557558e-05,
+      "loss": 4.0536,
+      "step": 1443
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9045107001067699,
+      "learning_rate": 3.8979177188494624e-05,
+      "loss": 4.2437,
+      "step": 1444
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3844076527397902,
+      "learning_rate": 3.897734958965301e-05,
+      "loss": 3.7467,
+      "step": 1445
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6450244536434206,
+      "learning_rate": 3.8975520399204e-05,
+      "loss": 3.1823,
+      "step": 1446
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0292094598627677,
+      "learning_rate": 3.8973689617301e-05,
+      "loss": 3.4326,
+      "step": 1447
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.306918710284133,
+      "learning_rate": 3.897185724409758e-05,
+      "loss": 3.8394,
+      "step": 1448
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3666693644963317,
+      "learning_rate": 3.897002327974741e-05,
+      "loss": 3.79,
+      "step": 1449
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2144868298784786,
+      "learning_rate": 3.896818772440428e-05,
+      "loss": 3.5534,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8328062104975894,
+      "learning_rate": 3.896635057822217e-05,
+      "loss": 3.2506,
+      "step": 1451
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0901573134590765,
+      "learning_rate": 3.8964511841355136e-05,
+      "loss": 3.5414,
+      "step": 1452
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3198517643610408,
+      "learning_rate": 3.896267151395739e-05,
+      "loss": 3.5215,
+      "step": 1453
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.89921662514426,
+      "learning_rate": 3.89608295961833e-05,
+      "loss": 3.539,
+      "step": 1454
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2513681249177593,
+      "learning_rate": 3.895898608818732e-05,
+      "loss": 3.6887,
+      "step": 1455
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2640481895818692,
+      "learning_rate": 3.8957140990124076e-05,
+      "loss": 3.7964,
+      "step": 1456
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1716692122755328,
+      "learning_rate": 3.895529430214831e-05,
+      "loss": 3.5664,
+      "step": 1457
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8301974759596213,
+      "learning_rate": 3.89534460244149e-05,
+      "loss": 3.5679,
+      "step": 1458
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.9690094409409715,
+      "learning_rate": 3.895159615707887e-05,
+      "loss": 3.585,
+      "step": 1459
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9225925781406249,
+      "learning_rate": 3.894974470029535e-05,
+      "loss": 3.473,
+      "step": 1460
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9577459689158079,
+      "learning_rate": 3.894789165421964e-05,
+      "loss": 3.9351,
+      "step": 1461
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 31.0979216930714,
+      "learning_rate": 3.894603701900713e-05,
+      "loss": 2.9965,
+      "step": 1462
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.91036017861903,
+      "learning_rate": 3.894418079481337e-05,
+      "loss": 3.6193,
+      "step": 1463
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.2239493885487205,
+      "learning_rate": 3.8942322981794055e-05,
+      "loss": 3.4311,
+      "step": 1464
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.768599527358904,
+      "learning_rate": 3.894046358010499e-05,
+      "loss": 3.6046,
+      "step": 1465
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8865452718398756,
+      "learning_rate": 3.893860258990212e-05,
+      "loss": 3.1451,
+      "step": 1466
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.2831058821720707,
+      "learning_rate": 3.893674001134152e-05,
+      "loss": 3.3884,
+      "step": 1467
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0472440127808467,
+      "learning_rate": 3.8934875844579395e-05,
+      "loss": 3.6401,
+      "step": 1468
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.0488206972182073,
+      "learning_rate": 3.893301008977211e-05,
+      "loss": 3.5783,
+      "step": 1469
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.865032769030806,
+      "learning_rate": 3.893114274707613e-05,
+      "loss": 2.9898,
+      "step": 1470
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9225082494792436,
+      "learning_rate": 3.892927381664808e-05,
+      "loss": 3.0564,
+      "step": 1471
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.053575656118969,
+      "learning_rate": 3.8927403298644685e-05,
+      "loss": 3.1004,
+      "step": 1472
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3768972841862714,
+      "learning_rate": 3.892553119322284e-05,
+      "loss": 3.463,
+      "step": 1473
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3246256311369864,
+      "learning_rate": 3.892365750053955e-05,
+      "loss": 2.7872,
+      "step": 1474
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9162711110680666,
+      "learning_rate": 3.892178222075196e-05,
+      "loss": 3.2049,
+      "step": 1475
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3444065737280324,
+      "learning_rate": 3.8919905354017346e-05,
+      "loss": 2.4947,
+      "step": 1476
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8169956396584723,
+      "learning_rate": 3.8918026900493113e-05,
+      "loss": 2.6102,
+      "step": 1477
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.045440797443741,
+      "learning_rate": 3.891614686033682e-05,
+      "loss": 2.8603,
+      "step": 1478
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.5919749809613384,
+      "learning_rate": 3.891426523370613e-05,
+      "loss": 2.3336,
+      "step": 1479
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 44.200773584096076,
+      "learning_rate": 3.8912382020758854e-05,
+      "loss": 9.5405,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.629041698925087,
+      "learning_rate": 3.891049722165295e-05,
+      "loss": 3.4208,
+      "step": 1481
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.780252929439216,
+      "learning_rate": 3.890861083654647e-05,
+      "loss": 3.7826,
+      "step": 1482
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 31.84642197227763,
+      "learning_rate": 3.8906722865597644e-05,
+      "loss": 3.5817,
+      "step": 1483
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.448803310957018,
+      "learning_rate": 3.89048333089648e-05,
+      "loss": 2.942,
+      "step": 1484
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.05325857944285,
+      "learning_rate": 3.890294216680642e-05,
+      "loss": 3.565,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.910122144072295,
+      "learning_rate": 3.8901049439281105e-05,
+      "loss": 3.0791,
+      "step": 1486
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1317196312683833,
+      "learning_rate": 3.88991551265476e-05,
+      "loss": 2.887,
+      "step": 1487
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 8.289123530207178,
+      "learning_rate": 3.889725922876479e-05,
+      "loss": 9.5472,
+      "step": 1488
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.4484414949831232,
+      "learning_rate": 3.889536174609166e-05,
+      "loss": 2.9492,
+      "step": 1489
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.467957520807212,
+      "learning_rate": 3.8893462678687364e-05,
+      "loss": 3.3085,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 5.031434241932885,
+      "learning_rate": 3.8891562026711165e-05,
+      "loss": 2.8762,
+      "step": 1491
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 4.0074364581999085,
+      "learning_rate": 3.888965979032248e-05,
+      "loss": 3.1375,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1511421849660604,
+      "learning_rate": 3.888775596968084e-05,
+      "loss": 3.4048,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8585378597431457,
+      "learning_rate": 3.8885850564945914e-05,
+      "loss": 2.5434,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.1894868887614614,
+      "learning_rate": 3.888394357627752e-05,
+      "loss": 2.6779,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.1195462532409297,
+      "learning_rate": 3.888203500383557e-05,
+      "loss": 2.7547,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 3.0567465734257047,
+      "learning_rate": 3.888012484778016e-05,
+      "loss": 2.8687,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.322157893476532,
+      "learning_rate": 3.887821310827147e-05,
+      "loss": 3.603,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.3614068452088737,
+      "learning_rate": 3.887629978546986e-05,
+      "loss": 3.0509,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6684438675843134,
+      "learning_rate": 3.887438487953577e-05,
+      "loss": 2.5048,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6420993647576199,
+      "learning_rate": 3.887246839062981e-05,
+      "loss": 2.7774,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3723356508341578,
+      "learning_rate": 3.887055031891273e-05,
+      "loss": 2.4834,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9084412248462337,
+      "learning_rate": 3.886863066454539e-05,
+      "loss": 2.7786,
+      "step": 1503
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1359101523717492,
+      "learning_rate": 3.886670942768877e-05,
+      "loss": 2.5924,
+      "step": 1504
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7151562007833434,
+      "learning_rate": 3.886478660850403e-05,
+      "loss": 3.0463,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5553522417327073,
+      "learning_rate": 3.88628622071524e-05,
+      "loss": 2.7334,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.196253929064681,
+      "learning_rate": 3.886093622379531e-05,
+      "loss": 2.5145,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.3419899812018226,
+      "learning_rate": 3.8859008658594283e-05,
+      "loss": 2.4732,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0265057050013962,
+      "learning_rate": 3.885707951171097e-05,
+      "loss": 1.9071,
+      "step": 1509
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.700721267121911,
+      "learning_rate": 3.885514878330717e-05,
+      "loss": 3.21,
+      "step": 1510
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2627212476409118,
+      "learning_rate": 3.885321647354481e-05,
+      "loss": 2.2101,
+      "step": 1511
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1323772021220393,
+      "learning_rate": 3.885128258258596e-05,
+      "loss": 2.6371,
+      "step": 1512
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0213790527401052,
+      "learning_rate": 3.88493471105928e-05,
+      "loss": 2.4138,
+      "step": 1513
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.533212281726283,
+      "learning_rate": 3.8847410057727664e-05,
+      "loss": 2.8262,
+      "step": 1514
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9612533525297318,
+      "learning_rate": 3.884547142415301e-05,
+      "loss": 2.8586,
+      "step": 1515
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.915179323182184,
+      "learning_rate": 3.884353121003143e-05,
+      "loss": 2.8137,
+      "step": 1516
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6704048114733794,
+      "learning_rate": 3.884158941552564e-05,
+      "loss": 2.8061,
+      "step": 1517
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.6592307375150512,
+      "learning_rate": 3.88396460407985e-05,
+      "loss": 2.56,
+      "step": 1518
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.736069527127518,
+      "learning_rate": 3.8837701086013e-05,
+      "loss": 3.1031,
+      "step": 1519
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.482539254673155,
+      "learning_rate": 3.883575455133226e-05,
+      "loss": 2.5165,
+      "step": 1520
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.104707245724428,
+      "learning_rate": 3.883380643691954e-05,
+      "loss": 2.3371,
+      "step": 1521
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.7560931734203264,
+      "learning_rate": 3.883185674293821e-05,
+      "loss": 2.4045,
+      "step": 1522
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0883540929010116,
+      "learning_rate": 3.8829905469551806e-05,
+      "loss": 2.711,
+      "step": 1523
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6156709889521552,
+      "learning_rate": 3.882795261692397e-05,
+      "loss": 2.3153,
+      "step": 1524
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.5930308387505545,
+      "learning_rate": 3.882599818521848e-05,
+      "loss": 2.5867,
+      "step": 1525
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 5.214798265406714,
+      "learning_rate": 3.882404217459926e-05,
+      "loss": 2.9913,
+      "step": 1526
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 12.131641655261474,
+      "learning_rate": 3.882208458523036e-05,
+      "loss": 2.9257,
+      "step": 1527
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7609884262825242,
+      "learning_rate": 3.882012541727596e-05,
+      "loss": 2.1318,
+      "step": 1528
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8454002656717547,
+      "learning_rate": 3.881816467090036e-05,
+      "loss": 2.3653,
+      "step": 1529
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4803345381617632,
+      "learning_rate": 3.8816202346268026e-05,
+      "loss": 2.0116,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6985743648226026,
+      "learning_rate": 3.881423844354352e-05,
+      "loss": 3.0743,
+      "step": 1531
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 16.08641643378324,
+      "learning_rate": 3.881227296289156e-05,
+      "loss": 2.1522,
+      "step": 1532
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6578248509726126,
+      "learning_rate": 3.8810305904476984e-05,
+      "loss": 2.3555,
+      "step": 1533
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 14.388299616905623,
+      "learning_rate": 3.880833726846477e-05,
+      "loss": 2.6564,
+      "step": 1534
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.5328780418716104,
+      "learning_rate": 3.880636705502002e-05,
+      "loss": 2.1924,
+      "step": 1535
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4450034528420255,
+      "learning_rate": 3.880439526430798e-05,
+      "loss": 2.5472,
+      "step": 1536
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9273615488752696,
+      "learning_rate": 3.8802421896494024e-05,
+      "loss": 2.4394,
+      "step": 1537
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.3502075368927975,
+      "learning_rate": 3.880044695174365e-05,
+      "loss": 2.4542,
+      "step": 1538
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 25.15777436754738,
+      "learning_rate": 3.879847043022249e-05,
+      "loss": 2.0769,
+      "step": 1539
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7353247724810512,
+      "learning_rate": 3.879649233209632e-05,
+      "loss": 2.5287,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.292651482864617,
+      "learning_rate": 3.879451265753104e-05,
+      "loss": 2.1944,
+      "step": 1541
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 6.325084121325147,
+      "learning_rate": 3.879253140669268e-05,
+      "loss": 1.8235,
+      "step": 1542
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.563276210257134,
+      "learning_rate": 3.879054857974741e-05,
+      "loss": 2.2032,
+      "step": 1543
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.397362373688821,
+      "learning_rate": 3.8788564176861514e-05,
+      "loss": 2.2345,
+      "step": 1544
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 7.375270699784678,
+      "learning_rate": 3.878657819820144e-05,
+      "loss": 2.4755,
+      "step": 1545
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2440247352543397,
+      "learning_rate": 3.878459064393374e-05,
+      "loss": 2.3002,
+      "step": 1546
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.723493331476765,
+      "learning_rate": 3.8782601514225116e-05,
+      "loss": 2.3843,
+      "step": 1547
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.550137047600981,
+      "learning_rate": 3.878061080924237e-05,
+      "loss": 2.787,
+      "step": 1548
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.742592357548979,
+      "learning_rate": 3.8778618529152484e-05,
+      "loss": 2.6799,
+      "step": 1549
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.561831178116687,
+      "learning_rate": 3.877662467412255e-05,
+      "loss": 2.0807,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5277285718107954,
+      "learning_rate": 3.877462924431977e-05,
+      "loss": 2.13,
+      "step": 1551
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.104135546193453,
+      "learning_rate": 3.8772632239911514e-05,
+      "loss": 1.8998,
+      "step": 1552
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5835610222183214,
+      "learning_rate": 3.877063366106526e-05,
+      "loss": 2.6025,
+      "step": 1553
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.090247414441834,
+      "learning_rate": 3.876863350794863e-05,
+      "loss": 2.4861,
+      "step": 1554
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.202568130666563,
+      "learning_rate": 3.876663178072938e-05,
+      "loss": 2.7796,
+      "step": 1555
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4293916229658674,
+      "learning_rate": 3.8764628479575386e-05,
+      "loss": 2.4605,
+      "step": 1556
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8937401732421194,
+      "learning_rate": 3.8762623604654656e-05,
+      "loss": 2.0699,
+      "step": 1557
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.316565356784012,
+      "learning_rate": 3.876061715613535e-05,
+      "loss": 2.5672,
+      "step": 1558
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4352218681297242,
+      "learning_rate": 3.8758609134185736e-05,
+      "loss": 2.7145,
+      "step": 1559
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5238928564267262,
+      "learning_rate": 3.875659953897423e-05,
+      "loss": 2.3671,
+      "step": 1560
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0581835470422851,
+      "learning_rate": 3.875458837066938e-05,
+      "loss": 2.4823,
+      "step": 1561
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3013618548257628,
+      "learning_rate": 3.875257562943984e-05,
+      "loss": 1.963,
+      "step": 1562
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.091275134535315,
+      "learning_rate": 3.875056131545444e-05,
+      "loss": 2.3327,
+      "step": 1563
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4876284854967292,
+      "learning_rate": 3.87485454288821e-05,
+      "loss": 1.9314,
+      "step": 1564
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0890848681202885,
+      "learning_rate": 3.8746527969891905e-05,
+      "loss": 2.5346,
+      "step": 1565
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4902016769097097,
+      "learning_rate": 3.874450893865305e-05,
+      "loss": 2.1462,
+      "step": 1566
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9869282529684841,
+      "learning_rate": 3.874248833533486e-05,
+      "loss": 2.611,
+      "step": 1567
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8394613324414156,
+      "learning_rate": 3.874046616010681e-05,
+      "loss": 2.4041,
+      "step": 1568
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8644124518110893,
+      "learning_rate": 3.87384424131385e-05,
+      "loss": 2.2576,
+      "step": 1569
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8333918105096911,
+      "learning_rate": 3.873641709459965e-05,
+      "loss": 2.2948,
+      "step": 1570
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5149280288757707,
+      "learning_rate": 3.8734390204660134e-05,
+      "loss": 2.2752,
+      "step": 1571
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3551527252730124,
+      "learning_rate": 3.8732361743489934e-05,
+      "loss": 2.1908,
+      "step": 1572
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.263207145993294,
+      "learning_rate": 3.873033171125918e-05,
+      "loss": 2.0947,
+      "step": 1573
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.439934408506761,
+      "learning_rate": 3.872830010813812e-05,
+      "loss": 2.0917,
+      "step": 1574
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0826346382103442,
+      "learning_rate": 3.8726266934297144e-05,
+      "loss": 2.1366,
+      "step": 1575
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0192566950874073,
+      "learning_rate": 3.8724232189906784e-05,
+      "loss": 1.7918,
+      "step": 1576
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9033833090591745,
+      "learning_rate": 3.872219587513768e-05,
+      "loss": 1.926,
+      "step": 1577
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5623477914101649,
+      "learning_rate": 3.8720157990160615e-05,
+      "loss": 2.5331,
+      "step": 1578
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.523003596348439,
+      "learning_rate": 3.871811853514652e-05,
+      "loss": 2.1922,
+      "step": 1579
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.087333100761452,
+      "learning_rate": 3.871607751026642e-05,
+      "loss": 1.7139,
+      "step": 1580
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8282190215186163,
+      "learning_rate": 3.87140349156915e-05,
+      "loss": 2.7783,
+      "step": 1581
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5898828404736125,
+      "learning_rate": 3.871199075159308e-05,
+      "loss": 1.9558,
+      "step": 1582
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3405031155740952,
+      "learning_rate": 3.870994501814259e-05,
+      "loss": 2.4636,
+      "step": 1583
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.3275632729783347,
+      "learning_rate": 3.87078977155116e-05,
+      "loss": 2.0361,
+      "step": 1584
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.578756247365489,
+      "learning_rate": 3.8705848843871837e-05,
+      "loss": 1.8291,
+      "step": 1585
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7607235306058557,
+      "learning_rate": 3.870379840339511e-05,
+      "loss": 1.9944,
+      "step": 1586
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1347373101597664,
+      "learning_rate": 3.870174639425341e-05,
+      "loss": 2.093,
+      "step": 1587
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.4535080963500344,
+      "learning_rate": 3.8699692816618816e-05,
+      "loss": 2.0248,
+      "step": 1588
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6875320483788168,
+      "learning_rate": 3.869763767066358e-05,
+      "loss": 2.1876,
+      "step": 1589
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.0942074711315573,
+      "learning_rate": 3.869558095656004e-05,
+      "loss": 2.1402,
+      "step": 1590
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3325394754819784,
+      "learning_rate": 3.869352267448071e-05,
+      "loss": 2.4655,
+      "step": 1591
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.2853463615358134,
+      "learning_rate": 3.869146282459821e-05,
+      "loss": 2.123,
+      "step": 1592
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0381729857496789,
+      "learning_rate": 3.86894014070853e-05,
+      "loss": 1.8238,
+      "step": 1593
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3474876509688076,
+      "learning_rate": 3.8687338422114867e-05,
+      "loss": 2.8399,
+      "step": 1594
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6276235713903708,
+      "learning_rate": 3.8685273869859916e-05,
+      "loss": 1.8226,
+      "step": 1595
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4977609331638422,
+      "learning_rate": 3.868320775049362e-05,
+      "loss": 2.4318,
+      "step": 1596
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0648408821418185,
+      "learning_rate": 3.8681140064189255e-05,
+      "loss": 1.6441,
+      "step": 1597
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9755244299138326,
+      "learning_rate": 3.867907081112023e-05,
+      "loss": 2.2894,
+      "step": 1598
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6169124295880044,
+      "learning_rate": 3.86769999914601e-05,
+      "loss": 2.39,
+      "step": 1599
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6666392363360614,
+      "learning_rate": 3.867492760538253e-05,
+      "loss": 2.4423,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4677167401902715,
+      "learning_rate": 3.867285365306134e-05,
+      "loss": 2.5824,
+      "step": 1601
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8012898844122256,
+      "learning_rate": 3.867077813467046e-05,
+      "loss": 2.2531,
+      "step": 1602
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2770146921391994,
+      "learning_rate": 3.8668701050383973e-05,
+      "loss": 1.7935,
+      "step": 1603
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3439897414234097,
+      "learning_rate": 3.8666622400376076e-05,
+      "loss": 2.1241,
+      "step": 1604
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2971502060284101,
+      "learning_rate": 3.8664542184821095e-05,
+      "loss": 2.4743,
+      "step": 1605
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3502580793848555,
+      "learning_rate": 3.86624604038935e-05,
+      "loss": 2.2945,
+      "step": 1606
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.263410700117351,
+      "learning_rate": 3.8660377057767895e-05,
+      "loss": 2.476,
+      "step": 1607
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2795144528365088,
+      "learning_rate": 3.8658292146619005e-05,
+      "loss": 2.1903,
+      "step": 1608
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0852145891474447,
+      "learning_rate": 3.8656205670621685e-05,
+      "loss": 1.5694,
+      "step": 1609
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4558598924805077,
+      "learning_rate": 3.8654117629950925e-05,
+      "loss": 2.0449,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.3945087343318117,
+      "learning_rate": 3.8652028024781845e-05,
+      "loss": 2.1038,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2266543624213309,
+      "learning_rate": 3.86499368552897e-05,
+      "loss": 1.8863,
+      "step": 1612
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9575583364906477,
+      "learning_rate": 3.864784412164988e-05,
+      "loss": 2.162,
+      "step": 1613
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8294686161190683,
+      "learning_rate": 3.864574982403789e-05,
+      "loss": 1.8466,
+      "step": 1614
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2779416884845947,
+      "learning_rate": 3.8643653962629386e-05,
+      "loss": 2.1441,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 3.296398449424663,
+      "learning_rate": 3.8641556537600135e-05,
+      "loss": 2.055,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.845877468789236,
+      "learning_rate": 3.863945754912605e-05,
+      "loss": 2.0415,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4570848646560413,
+      "learning_rate": 3.8637356997383176e-05,
+      "loss": 1.8685,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.313390581418334,
+      "learning_rate": 3.863525488254767e-05,
+      "loss": 2.358,
+      "step": 1619
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 4.232913901728299,
+      "learning_rate": 3.863315120479585e-05,
+      "loss": 2.4012,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.524711891376185,
+      "learning_rate": 3.8631045964304144e-05,
+      "loss": 2.1243,
+      "step": 1621
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.29694973213382,
+      "learning_rate": 3.8628939161249105e-05,
+      "loss": 1.834,
+      "step": 1622
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.973701486857545,
+      "learning_rate": 3.862683079580744e-05,
+      "loss": 2.2829,
+      "step": 1623
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3716209211510062,
+      "learning_rate": 3.8624720868155975e-05,
+      "loss": 2.3364,
+      "step": 1624
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4745824042939153,
+      "learning_rate": 3.862260937847166e-05,
+      "loss": 1.5496,
+      "step": 1625
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.158936811877794,
+      "learning_rate": 3.8620496326931587e-05,
+      "loss": 2.3027,
+      "step": 1626
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5678042034210908,
+      "learning_rate": 3.8618381713712974e-05,
+      "loss": 2.2845,
+      "step": 1627
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.235456322058117,
+      "learning_rate": 3.861626553899318e-05,
+      "loss": 2.1907,
+      "step": 1628
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.121541963027202,
+      "learning_rate": 3.861414780294967e-05,
+      "loss": 2.3592,
+      "step": 1629
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8167832191684272,
+      "learning_rate": 3.861202850576007e-05,
+      "loss": 2.5692,
+      "step": 1630
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6264475687544468,
+      "learning_rate": 3.860990764760211e-05,
+      "loss": 2.0356,
+      "step": 1631
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7712887330635507,
+      "learning_rate": 3.860778522865367e-05,
+      "loss": 2.1612,
+      "step": 1632
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5090928256512062,
+      "learning_rate": 3.860566124909277e-05,
+      "loss": 1.8492,
+      "step": 1633
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3262963065229276,
+      "learning_rate": 3.8603535709097514e-05,
+      "loss": 1.9707,
+      "step": 1634
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3280809691645767,
+      "learning_rate": 3.8601408608846186e-05,
+      "loss": 2.1811,
+      "step": 1635
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.313945278563808,
+      "learning_rate": 3.8599279948517195e-05,
+      "loss": 1.7862,
+      "step": 1636
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7516889733399517,
+      "learning_rate": 3.859714972828905e-05,
+      "loss": 2.017,
+      "step": 1637
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.555502362671458,
+      "learning_rate": 3.859501794834041e-05,
+      "loss": 2.0153,
+      "step": 1638
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1808417992896083,
+      "learning_rate": 3.859288460885008e-05,
+      "loss": 1.4449,
+      "step": 1639
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6898634750344041,
+      "learning_rate": 3.859074970999697e-05,
+      "loss": 2.2171,
+      "step": 1640
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.325355847431627,
+      "learning_rate": 3.858861325196014e-05,
+      "loss": 2.0959,
+      "step": 1641
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1747086987629696,
+      "learning_rate": 3.8586475234918766e-05,
+      "loss": 1.7139,
+      "step": 1642
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3782928279210718,
+      "learning_rate": 3.858433565905215e-05,
+      "loss": 2.0198,
+      "step": 1643
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0343555251463727,
+      "learning_rate": 3.858219452453975e-05,
+      "loss": 2.1542,
+      "step": 1644
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9700996633766805,
+      "learning_rate": 3.858005183156114e-05,
+      "loss": 1.7505,
+      "step": 1645
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5894404564946139,
+      "learning_rate": 3.8577907580296026e-05,
+      "loss": 2.5457,
+      "step": 1646
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4447813101683928,
+      "learning_rate": 3.857576177092423e-05,
+      "loss": 1.6698,
+      "step": 1647
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2516984123155257,
+      "learning_rate": 3.857361440362573e-05,
+      "loss": 1.9499,
+      "step": 1648
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.393225816359324,
+      "learning_rate": 3.857146547858063e-05,
+      "loss": 2.0465,
+      "step": 1649
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.089884336303116,
+      "learning_rate": 3.8569314995969144e-05,
+      "loss": 1.822,
+      "step": 1650
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7082057108527484,
+      "learning_rate": 3.856716295597164e-05,
+      "loss": 1.8203,
+      "step": 1651
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.208976303579272,
+      "learning_rate": 3.856500935876858e-05,
+      "loss": 2.1316,
+      "step": 1652
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6041741568637755,
+      "learning_rate": 3.856285420454063e-05,
+      "loss": 2.0214,
+      "step": 1653
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.38671021141194,
+      "learning_rate": 3.856069749346851e-05,
+      "loss": 1.775,
+      "step": 1654
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7031935747501956,
+      "learning_rate": 3.855853922573311e-05,
+      "loss": 1.9639,
+      "step": 1655
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.351507416912302,
+      "learning_rate": 3.8556379401515435e-05,
+      "loss": 1.6797,
+      "step": 1656
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5427932530708017,
+      "learning_rate": 3.8554218020996626e-05,
+      "loss": 1.9192,
+      "step": 1657
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9817770052902988,
+      "learning_rate": 3.855205508435796e-05,
+      "loss": 2.462,
+      "step": 1658
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.483459582272449,
+      "learning_rate": 3.854989059178084e-05,
+      "loss": 2.0378,
+      "step": 1659
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9266188318728452,
+      "learning_rate": 3.85477245434468e-05,
+      "loss": 1.6873,
+      "step": 1660
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.360180200815957,
+      "learning_rate": 3.85455569395375e-05,
+      "loss": 1.8817,
+      "step": 1661
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 11.271280909025778,
+      "learning_rate": 3.854338778023475e-05,
+      "loss": 2.1331,
+      "step": 1662
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.193489754146838,
+      "learning_rate": 3.854121706572045e-05,
+      "loss": 2.2228,
+      "step": 1663
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.263476479969966,
+      "learning_rate": 3.853904479617666e-05,
+      "loss": 2.4502,
+      "step": 1664
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.9847097377715297,
+      "learning_rate": 3.8536870971785584e-05,
+      "loss": 2.1643,
+      "step": 1665
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.449819350104968,
+      "learning_rate": 3.853469559272952e-05,
+      "loss": 2.0924,
+      "step": 1666
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.955155464545263,
+      "learning_rate": 3.853251865919092e-05,
+      "loss": 2.0128,
+      "step": 1667
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.816087489936284,
+      "learning_rate": 3.853034017135237e-05,
+      "loss": 2.0989,
+      "step": 1668
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.565913254044061,
+      "learning_rate": 3.852816012939657e-05,
+      "loss": 1.9971,
+      "step": 1669
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9710979379016065,
+      "learning_rate": 3.852597853350635e-05,
+      "loss": 2.3642,
+      "step": 1670
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4172398167046927,
+      "learning_rate": 3.8523795383864685e-05,
+      "loss": 1.8722,
+      "step": 1671
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 6.3751452290238175,
+      "learning_rate": 3.852161068065467e-05,
+      "loss": 9.5999,
+      "step": 1672
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.349440383923166,
+      "learning_rate": 3.851942442405953e-05,
+      "loss": 2.1974,
+      "step": 1673
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6839583071690727,
+      "learning_rate": 3.851723661426264e-05,
+      "loss": 1.9246,
+      "step": 1674
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.57605212542232,
+      "learning_rate": 3.851504725144748e-05,
+      "loss": 2.4635,
+      "step": 1675
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9078438994734077,
+      "learning_rate": 3.851285633579766e-05,
+      "loss": 2.3175,
+      "step": 1676
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7499964196438547,
+      "learning_rate": 3.851066386749694e-05,
+      "loss": 2.3624,
+      "step": 1677
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0395882538660244,
+      "learning_rate": 3.8508469846729195e-05,
+      "loss": 2.3519,
+      "step": 1678
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4548957498923387,
+      "learning_rate": 3.850627427367844e-05,
+      "loss": 2.1153,
+      "step": 1679
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0805746448185305,
+      "learning_rate": 3.8504077148528806e-05,
+      "loss": 1.9029,
+      "step": 1680
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2039916815548724,
+      "learning_rate": 3.8501878471464574e-05,
+      "loss": 1.8428,
+      "step": 1681
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4706187754045605,
+      "learning_rate": 3.849967824267014e-05,
+      "loss": 2.0283,
+      "step": 1682
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0498303373244542,
+      "learning_rate": 3.8497476462330034e-05,
+      "loss": 1.805,
+      "step": 1683
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0650275110451082,
+      "learning_rate": 3.849527313062891e-05,
+      "loss": 2.3156,
+      "step": 1684
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1731634800028976,
+      "learning_rate": 3.8493068247751565e-05,
+      "loss": 1.9545,
+      "step": 1685
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0724138871479099,
+      "learning_rate": 3.849086181388293e-05,
+      "loss": 1.8009,
+      "step": 1686
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1927005225356095,
+      "learning_rate": 3.848865382920804e-05,
+      "loss": 1.7303,
+      "step": 1687
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.343366430172977,
+      "learning_rate": 3.848644429391208e-05,
+      "loss": 1.8526,
+      "step": 1688
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.294327068694301,
+      "learning_rate": 3.8484233208180366e-05,
+      "loss": 2.0543,
+      "step": 1689
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2141610234773876,
+      "learning_rate": 3.848202057219833e-05,
+      "loss": 1.6618,
+      "step": 1690
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1013181128576333,
+      "learning_rate": 3.847980638615155e-05,
+      "loss": 2.0169,
+      "step": 1691
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1229908516182452,
+      "learning_rate": 3.8477590650225735e-05,
+      "loss": 2.1373,
+      "step": 1692
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.696597566130838,
+      "learning_rate": 3.84753733646067e-05,
+      "loss": 8.4552,
+      "step": 1693
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.2859597608649005,
+      "learning_rate": 3.847315452948042e-05,
+      "loss": 2.0811,
+      "step": 1694
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.692196939911033,
+      "learning_rate": 3.847093414503298e-05,
+      "loss": 2.1882,
+      "step": 1695
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7505197836600168,
+      "learning_rate": 3.8468712211450594e-05,
+      "loss": 1.8692,
+      "step": 1696
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3379364077618168,
+      "learning_rate": 3.8466488728919626e-05,
+      "loss": 1.8857,
+      "step": 1697
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9399069488147471,
+      "learning_rate": 3.846426369762655e-05,
+      "loss": 2.1397,
+      "step": 1698
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9941856871368049,
+      "learning_rate": 3.846203711775797e-05,
+      "loss": 2.4595,
+      "step": 1699
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0314150753447873,
+      "learning_rate": 3.845980898950063e-05,
+      "loss": 1.8961,
+      "step": 1700
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2421156918241967,
+      "learning_rate": 3.8457579313041414e-05,
+      "loss": 1.9737,
+      "step": 1701
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.539681516215296,
+      "learning_rate": 3.845534808856731e-05,
+      "loss": 1.7146,
+      "step": 1702
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2937912300254963,
+      "learning_rate": 3.845311531626545e-05,
+      "loss": 1.8198,
+      "step": 1703
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1396178417847551,
+      "learning_rate": 3.845088099632309e-05,
+      "loss": 1.7297,
+      "step": 1704
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.158091262326807,
+      "learning_rate": 3.844864512892762e-05,
+      "loss": 2.6921,
+      "step": 1705
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0912750958152635,
+      "learning_rate": 3.8446407714266575e-05,
+      "loss": 1.7297,
+      "step": 1706
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2061307556931382,
+      "learning_rate": 3.844416875252759e-05,
+      "loss": 1.7289,
+      "step": 1707
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.0742951593823102,
+      "learning_rate": 3.844192824389844e-05,
+      "loss": 2.2985,
+      "step": 1708
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.971377034599811,
+      "learning_rate": 3.843968618856705e-05,
+      "loss": 1.7654,
+      "step": 1709
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.2178474607906313,
+      "learning_rate": 3.843744258672144e-05,
+      "loss": 2.2042,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1097963077295023,
+      "learning_rate": 3.84351974385498e-05,
+      "loss": 2.0875,
+      "step": 1711
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.385307673454144,
+      "learning_rate": 3.8432950744240396e-05,
+      "loss": 1.6615,
+      "step": 1712
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 9.993555888655695,
+      "learning_rate": 3.843070250398168e-05,
+      "loss": 8.8763,
+      "step": 1713
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.547411966038372,
+      "learning_rate": 3.842845271796221e-05,
+      "loss": 2.3305,
+      "step": 1714
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8096722304202224,
+      "learning_rate": 3.842620138637067e-05,
+      "loss": 2.3076,
+      "step": 1715
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9127421282878267,
+      "learning_rate": 3.8423948509395865e-05,
+      "loss": 1.952,
+      "step": 1716
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.855699018260501,
+      "learning_rate": 3.842169408722675e-05,
+      "loss": 2.1193,
+      "step": 1717
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.183038480571973,
+      "learning_rate": 3.8419438120052396e-05,
+      "loss": 2.1083,
+      "step": 1718
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7182679640368312,
+      "learning_rate": 3.841718060806202e-05,
+      "loss": 1.8303,
+      "step": 1719
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.674512130974247,
+      "learning_rate": 3.8414921551444945e-05,
+      "loss": 1.8991,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5507023941063038,
+      "learning_rate": 3.841266095039064e-05,
+      "loss": 1.9451,
+      "step": 1721
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6057588121094015,
+      "learning_rate": 3.84103988050887e-05,
+      "loss": 1.7725,
+      "step": 1722
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.068439875354089,
+      "learning_rate": 3.840813511572884e-05,
+      "loss": 8.2508,
+      "step": 1723
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.233603232978281,
+      "learning_rate": 3.840586988250093e-05,
+      "loss": 1.802,
+      "step": 1724
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.366648185248925,
+      "learning_rate": 3.840360310559493e-05,
+      "loss": 2.0404,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.579127476049318,
+      "learning_rate": 3.8401334785200966e-05,
+      "loss": 1.7603,
+      "step": 1726
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4835732401871378,
+      "learning_rate": 3.839906492150928e-05,
+      "loss": 1.965,
+      "step": 1727
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 5.274588120919532,
+      "learning_rate": 3.8396793514710235e-05,
+      "loss": 1.9823,
+      "step": 1728
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.376740352954503,
+      "learning_rate": 3.839452056499434e-05,
+      "loss": 1.5501,
+      "step": 1729
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.668710309625274,
+      "learning_rate": 3.839224607255222e-05,
+      "loss": 7.9912,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3325583636002896,
+      "learning_rate": 3.8389970037574634e-05,
+      "loss": 1.9931,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7306144821779237,
+      "learning_rate": 3.8387692460252465e-05,
+      "loss": 1.8944,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.3073873890438894,
+      "learning_rate": 3.8385413340776746e-05,
+      "loss": 2.2308,
+      "step": 1733
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.0772224578980913,
+      "learning_rate": 3.838313267933861e-05,
+      "loss": 1.8311,
+      "step": 1734
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.741149820996405,
+      "learning_rate": 3.8380850476129336e-05,
+      "loss": 1.9658,
+      "step": 1735
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4180951798059327,
+      "learning_rate": 3.837856673134033e-05,
+      "loss": 1.8055,
+      "step": 1736
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.66479593412073,
+      "learning_rate": 3.837628144516313e-05,
+      "loss": 1.7656,
+      "step": 1737
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.565552883588831,
+      "learning_rate": 3.8373994617789395e-05,
+      "loss": 1.9581,
+      "step": 1738
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.486103587805695,
+      "learning_rate": 3.8371706249410925e-05,
+      "loss": 1.8858,
+      "step": 1739
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9503514269149482,
+      "learning_rate": 3.8369416340219636e-05,
+      "loss": 1.9281,
+      "step": 1740
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8412449713946657,
+      "learning_rate": 3.8367124890407586e-05,
+      "loss": 1.9592,
+      "step": 1741
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6248393170827318,
+      "learning_rate": 3.836483190016695e-05,
+      "loss": 1.4632,
+      "step": 1742
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9971109781217509,
+      "learning_rate": 3.836253736969004e-05,
+      "loss": 1.7103,
+      "step": 1743
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.5610304197732274,
+      "learning_rate": 3.836024129916931e-05,
+      "loss": 1.8441,
+      "step": 1744
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6207494185628262,
+      "learning_rate": 3.8357943688797305e-05,
+      "loss": 1.4763,
+      "step": 1745
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.685391286025051,
+      "learning_rate": 3.8355644538766734e-05,
+      "loss": 1.3968,
+      "step": 1746
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.01186021282977,
+      "learning_rate": 3.835334384927043e-05,
+      "loss": 1.6665,
+      "step": 1747
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.7922203428566257,
+      "learning_rate": 3.8351041620501336e-05,
+      "loss": 1.5277,
+      "step": 1748
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7154921216258854,
+      "learning_rate": 3.8348737852652546e-05,
+      "loss": 1.3742,
+      "step": 1749
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5495109398425289,
+      "learning_rate": 3.834643254591727e-05,
+      "loss": 1.3011,
+      "step": 1750
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.15563132433074,
+      "learning_rate": 3.8344125700488854e-05,
+      "loss": 1.445,
+      "step": 1751
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.3229667836013095,
+      "learning_rate": 3.8341817316560776e-05,
+      "loss": 1.2876,
+      "step": 1752
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.551418754327099,
+      "learning_rate": 3.833950739432662e-05,
+      "loss": 1.5233,
+      "step": 1753
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 5.791860796213126,
+      "learning_rate": 3.833719593398013e-05,
+      "loss": 1.3143,
+      "step": 1754
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7317193081467472,
+      "learning_rate": 3.8334882935715164e-05,
+      "loss": 1.2103,
+      "step": 1755
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2321238062573405,
+      "learning_rate": 3.8332568399725716e-05,
+      "loss": 1.1831,
+      "step": 1756
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6337628944001579,
+      "learning_rate": 3.833025232620589e-05,
+      "loss": 1.2096,
+      "step": 1757
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.767506671945708,
+      "learning_rate": 3.832793471534994e-05,
+      "loss": 1.1853,
+      "step": 1758
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6306069931015759,
+      "learning_rate": 3.8325615567352235e-05,
+      "loss": 1.2006,
+      "step": 1759
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.268368155611063,
+      "learning_rate": 3.8323294882407296e-05,
+      "loss": 1.1444,
+      "step": 1760
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.240410642115301,
+      "learning_rate": 3.832097266070974e-05,
+      "loss": 1.1321,
+      "step": 1761
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0570668162688028,
+      "learning_rate": 3.831864890245433e-05,
+      "loss": 1.0945,
+      "step": 1762
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.3268285490850178,
+      "learning_rate": 3.831632360783595e-05,
+      "loss": 1.1604,
+      "step": 1763
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4630780447860945,
+      "learning_rate": 3.8313996777049644e-05,
+      "loss": 1.113,
+      "step": 1764
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0842228802777711,
+      "learning_rate": 3.831166841029054e-05,
+      "loss": 1.071,
+      "step": 1765
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0253604781466166,
+      "learning_rate": 3.830933850775393e-05,
+      "loss": 1.0851,
+      "step": 1766
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4056948248732826,
+      "learning_rate": 3.83070070696352e-05,
+      "loss": 1.1047,
+      "step": 1767
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9303189744818031,
+      "learning_rate": 3.83046740961299e-05,
+      "loss": 1.083,
+      "step": 1768
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 4.242422711132556,
+      "learning_rate": 3.830233958743369e-05,
+      "loss": 1.0506,
+      "step": 1769
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1297156683612941,
+      "learning_rate": 3.830000354374236e-05,
+      "loss": 1.1226,
+      "step": 1770
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1625673679175257,
+      "learning_rate": 3.829766596525183e-05,
+      "loss": 1.013,
+      "step": 1771
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2921871033944958,
+      "learning_rate": 3.829532685215817e-05,
+      "loss": 1.0859,
+      "step": 1772
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.539650294684828,
+      "learning_rate": 3.829298620465753e-05,
+      "loss": 1.0296,
+      "step": 1773
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5316926655323067,
+      "learning_rate": 3.8290644022946234e-05,
+      "loss": 1.0758,
+      "step": 1774
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1046365396245803,
+      "learning_rate": 3.828830030722071e-05,
+      "loss": 1.0516,
+      "step": 1775
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.449314446091933,
+      "learning_rate": 3.8285955057677525e-05,
+      "loss": 1.0392,
+      "step": 1776
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4989586397774497,
+      "learning_rate": 3.828360827451337e-05,
+      "loss": 1.0467,
+      "step": 1777
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9773223946611554,
+      "learning_rate": 3.828125995792508e-05,
+      "loss": 1.0401,
+      "step": 1778
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1067869391548455,
+      "learning_rate": 3.827891010810959e-05,
+      "loss": 0.9826,
+      "step": 1779
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.160085232512416,
+      "learning_rate": 3.827655872526398e-05,
+      "loss": 1.0509,
+      "step": 1780
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7029084141561648,
+      "learning_rate": 3.827420580958547e-05,
+      "loss": 1.0567,
+      "step": 1781
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2469910974283513,
+      "learning_rate": 3.827185136127139e-05,
+      "loss": 1.0424,
+      "step": 1782
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.154545109014722,
+      "learning_rate": 3.8269495380519194e-05,
+      "loss": 8.0477,
+      "step": 1783
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.3887551054087246,
+      "learning_rate": 3.826713786752649e-05,
+      "loss": 1.0735,
+      "step": 1784
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4640538209529037,
+      "learning_rate": 3.8264778822490994e-05,
+      "loss": 1.1109,
+      "step": 1785
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8209650739286716,
+      "learning_rate": 3.8262418245610556e-05,
+      "loss": 1.0267,
+      "step": 1786
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8811835760053275,
+      "learning_rate": 3.826005613708316e-05,
+      "loss": 1.0163,
+      "step": 1787
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.4903733655439249,
+      "learning_rate": 3.82576924971069e-05,
+      "loss": 1.0281,
+      "step": 1788
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6517350337719576,
+      "learning_rate": 3.825532732588002e-05,
+      "loss": 1.0452,
+      "step": 1789
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6176833794844114,
+      "learning_rate": 3.825296062360089e-05,
+      "loss": 0.9956,
+      "step": 1790
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7571843819319738,
+      "learning_rate": 3.825059239046799e-05,
+      "loss": 1.03,
+      "step": 1791
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.030234398561132,
+      "learning_rate": 3.824822262667995e-05,
+      "loss": 1.0016,
+      "step": 1792
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.107470076506068,
+      "learning_rate": 3.824585133243552e-05,
+      "loss": 1.0143,
+      "step": 1793
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2878423228542897,
+      "learning_rate": 3.824347850793357e-05,
+      "loss": 1.0061,
+      "step": 1794
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0073300824105789,
+      "learning_rate": 3.824110415337311e-05,
+      "loss": 1.0124,
+      "step": 1795
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8199527682466337,
+      "learning_rate": 3.823872826895328e-05,
+      "loss": 1.0692,
+      "step": 1796
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.073993648789656,
+      "learning_rate": 3.823635085487334e-05,
+      "loss": 1.0303,
+      "step": 1797
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.2310911883949458,
+      "learning_rate": 3.823397191133266e-05,
+      "loss": 1.0317,
+      "step": 1798
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.163843195327525,
+      "learning_rate": 3.8231591438530794e-05,
+      "loss": 1.0646,
+      "step": 1799
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6780937597089078,
+      "learning_rate": 3.822920943666736e-05,
+      "loss": 1.0085,
+      "step": 1800
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9725822936608967,
+      "learning_rate": 3.822682590594216e-05,
+      "loss": 1.0065,
+      "step": 1801
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6934925789387608,
+      "learning_rate": 3.822444084655507e-05,
+      "loss": 0.9944,
+      "step": 1802
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9894833167773317,
+      "learning_rate": 3.822205425870614e-05,
+      "loss": 1.043,
+      "step": 1803
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8386158353930395,
+      "learning_rate": 3.8219666142595526e-05,
+      "loss": 1.0109,
+      "step": 1804
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8116554289764508,
+      "learning_rate": 3.821727649842352e-05,
+      "loss": 1.027,
+      "step": 1805
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6900825737769456,
+      "learning_rate": 3.821488532639053e-05,
+      "loss": 0.9966,
+      "step": 1806
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9806755252310013,
+      "learning_rate": 3.821249262669711e-05,
+      "loss": 1.0069,
+      "step": 1807
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9326193277237798,
+      "learning_rate": 3.8210098399543927e-05,
+      "loss": 1.0189,
+      "step": 1808
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.4165670710481573,
+      "learning_rate": 3.8207702645131784e-05,
+      "loss": 8.1089,
+      "step": 1809
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.657326217385156,
+      "learning_rate": 3.820530536366161e-05,
+      "loss": 1.002,
+      "step": 1810
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9988034500375527,
+      "learning_rate": 3.8202906555334454e-05,
+      "loss": 0.9942,
+      "step": 1811
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6167083032317967,
+      "learning_rate": 3.820050622035152e-05,
+      "loss": 1.0051,
+      "step": 1812
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9582192454667517,
+      "learning_rate": 3.8198104358914105e-05,
+      "loss": 1.039,
+      "step": 1813
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8946741547792898,
+      "learning_rate": 3.8195700971223655e-05,
+      "loss": 0.9996,
+      "step": 1814
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9682242797290423,
+      "learning_rate": 3.819329605748174e-05,
+      "loss": 1.0228,
+      "step": 1815
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6149544840164844,
+      "learning_rate": 3.8190889617890054e-05,
+      "loss": 0.9949,
+      "step": 1816
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8546471174592802,
+      "learning_rate": 3.818848165265043e-05,
+      "loss": 0.9752,
+      "step": 1817
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8454226524363837,
+      "learning_rate": 3.818607216196481e-05,
+      "loss": 1.0028,
+      "step": 1818
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5640877857121088,
+      "learning_rate": 3.818366114603528e-05,
+      "loss": 1.0144,
+      "step": 1819
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9690475308613412,
+      "learning_rate": 3.818124860506406e-05,
+      "loss": 1.0014,
+      "step": 1820
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6270387745965742,
+      "learning_rate": 3.8178834539253464e-05,
+      "loss": 1.0266,
+      "step": 1821
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7093044022531366,
+      "learning_rate": 3.817641894880599e-05,
+      "loss": 1.0175,
+      "step": 1822
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7304779617618384,
+      "learning_rate": 3.817400183392419e-05,
+      "loss": 1.0132,
+      "step": 1823
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6333908195261108,
+      "learning_rate": 3.817158319481081e-05,
+      "loss": 1.001,
+      "step": 1824
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9428294516709022,
+      "learning_rate": 3.81691630316687e-05,
+      "loss": 0.9812,
+      "step": 1825
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6723050555304716,
+      "learning_rate": 3.816674134470082e-05,
+      "loss": 1.0019,
+      "step": 1826
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8488571137328722,
+      "learning_rate": 3.8164318134110285e-05,
+      "loss": 0.9868,
+      "step": 1827
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.773237900722747,
+      "learning_rate": 3.816189340010033e-05,
+      "loss": 0.981,
+      "step": 1828
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.764877131464061,
+      "learning_rate": 3.815946714287431e-05,
+      "loss": 1.0106,
+      "step": 1829
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6725679510637078,
+      "learning_rate": 3.8157039362635703e-05,
+      "loss": 1.0312,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6903379934932108,
+      "learning_rate": 3.815461005958814e-05,
+      "loss": 1.0152,
+      "step": 1831
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8449925422374996,
+      "learning_rate": 3.8152179233935355e-05,
+      "loss": 0.9782,
+      "step": 1832
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7369365250588161,
+      "learning_rate": 3.8149746885881216e-05,
+      "loss": 1.0048,
+      "step": 1833
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6841864014052853,
+      "learning_rate": 3.814731301562973e-05,
+      "loss": 1.0025,
+      "step": 1834
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7864640561838039,
+      "learning_rate": 3.8144877623385016e-05,
+      "loss": 0.9804,
+      "step": 1835
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7396882635049554,
+      "learning_rate": 3.814244070935133e-05,
+      "loss": 1.0086,
+      "step": 1836
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6518320587000878,
+      "learning_rate": 3.814000227373305e-05,
+      "loss": 1.0134,
+      "step": 1837
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5043248792773318,
+      "learning_rate": 3.813756231673469e-05,
+      "loss": 1.0057,
+      "step": 1838
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6395450489337337,
+      "learning_rate": 3.8135120838560885e-05,
+      "loss": 0.9811,
+      "step": 1839
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7297469214400609,
+      "learning_rate": 3.81326778394164e-05,
+      "loss": 1.0008,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6308649140811685,
+      "learning_rate": 3.813023331950611e-05,
+      "loss": 1.0035,
+      "step": 1841
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6987331901496818,
+      "learning_rate": 3.812778727903506e-05,
+      "loss": 0.9906,
+      "step": 1842
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5394302936796274,
+      "learning_rate": 3.8125339718208376e-05,
+      "loss": 0.9992,
+      "step": 1843
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7615286473589584,
+      "learning_rate": 3.812289063723134e-05,
+      "loss": 0.9753,
+      "step": 1844
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5890286942382108,
+      "learning_rate": 3.812044003630936e-05,
+      "loss": 0.9778,
+      "step": 1845
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6857806125096,
+      "learning_rate": 3.811798791564795e-05,
+      "loss": 0.9741,
+      "step": 1846
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8846883956809151,
+      "learning_rate": 3.8115534275452774e-05,
+      "loss": 0.9692,
+      "step": 1847
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7946772558921203,
+      "learning_rate": 3.811307911592963e-05,
+      "loss": 0.9746,
+      "step": 1848
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5475387543230582,
+      "learning_rate": 3.81106224372844e-05,
+      "loss": 0.9856,
+      "step": 1849
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8360969455441156,
+      "learning_rate": 3.810816423972314e-05,
+      "loss": 1.0048,
+      "step": 1850
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6977056691470485,
+      "learning_rate": 3.810570452345202e-05,
+      "loss": 0.9748,
+      "step": 1851
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7420618382563371,
+      "learning_rate": 3.810324328867733e-05,
+      "loss": 1.0008,
+      "step": 1852
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5570802621546608,
+      "learning_rate": 3.810078053560548e-05,
+      "loss": 1.0079,
+      "step": 1853
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6210392423530283,
+      "learning_rate": 3.8098316264443033e-05,
+      "loss": 0.9884,
+      "step": 1854
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.49075279882745815,
+      "learning_rate": 3.809585047539666e-05,
+      "loss": 0.9942,
+      "step": 1855
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7831377342328955,
+      "learning_rate": 3.809338316867316e-05,
+      "loss": 0.9783,
+      "step": 1856
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6514316589822858,
+      "learning_rate": 3.809091434447946e-05,
+      "loss": 1.0073,
+      "step": 1857
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6245893304184089,
+      "learning_rate": 3.808844400302263e-05,
+      "loss": 0.994,
+      "step": 1858
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.48194791404858706,
+      "learning_rate": 3.8085972144509836e-05,
+      "loss": 0.9637,
+      "step": 1859
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9024817589671661,
+      "learning_rate": 3.808349876914841e-05,
+      "loss": 0.9724,
+      "step": 1860
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5371422559452416,
+      "learning_rate": 3.808102387714578e-05,
+      "loss": 0.9965,
+      "step": 1861
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4720854993907793,
+      "learning_rate": 3.807854746870951e-05,
+      "loss": 0.9583,
+      "step": 1862
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5428820811085859,
+      "learning_rate": 3.80760695440473e-05,
+      "loss": 1.0302,
+      "step": 1863
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6485740215494747,
+      "learning_rate": 3.807359010336697e-05,
+      "loss": 0.9542,
+      "step": 1864
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4985464895543278,
+      "learning_rate": 3.8071109146876455e-05,
+      "loss": 0.9553,
+      "step": 1865
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5750267398248837,
+      "learning_rate": 3.806862667478385e-05,
+      "loss": 0.9658,
+      "step": 1866
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6833123486587026,
+      "learning_rate": 3.8066142687297346e-05,
+      "loss": 0.9884,
+      "step": 1867
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5945156649791127,
+      "learning_rate": 3.806365718462527e-05,
+      "loss": 0.9822,
+      "step": 1868
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.59757827536242,
+      "learning_rate": 3.8061170166976085e-05,
+      "loss": 0.9769,
+      "step": 1869
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8549841487461234,
+      "learning_rate": 3.805868163455837e-05,
+      "loss": 0.9825,
+      "step": 1870
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7395906976698896,
+      "learning_rate": 3.805619158758084e-05,
+      "loss": 0.9603,
+      "step": 1871
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6125344917377242,
+      "learning_rate": 3.805370002625232e-05,
+      "loss": 0.9917,
+      "step": 1872
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8293603821970065,
+      "learning_rate": 3.805120695078178e-05,
+      "loss": 1.0049,
+      "step": 1873
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5314589885467663,
+      "learning_rate": 3.804871236137832e-05,
+      "loss": 0.9584,
+      "step": 1874
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6948857789994267,
+      "learning_rate": 3.804621625825115e-05,
+      "loss": 1.002,
+      "step": 1875
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5907244510930254,
+      "learning_rate": 3.8043718641609615e-05,
+      "loss": 1.0168,
+      "step": 1876
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7049259358204232,
+      "learning_rate": 3.804121951166319e-05,
+      "loss": 0.9733,
+      "step": 1877
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7325276424072316,
+      "learning_rate": 3.8038718868621474e-05,
+      "loss": 0.965,
+      "step": 1878
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.693168705607207,
+      "learning_rate": 3.803621671269419e-05,
+      "loss": 0.969,
+      "step": 1879
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7958009240586993,
+      "learning_rate": 3.803371304409119e-05,
+      "loss": 0.9536,
+      "step": 1880
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7131165436163169,
+      "learning_rate": 3.8031207863022465e-05,
+      "loss": 0.9513,
+      "step": 1881
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7747515170194297,
+      "learning_rate": 3.8028701169698104e-05,
+      "loss": 0.9758,
+      "step": 1882
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6286789800940346,
+      "learning_rate": 3.802619296432835e-05,
+      "loss": 0.9897,
+      "step": 1883
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.416915003081324,
+      "learning_rate": 3.802368324712357e-05,
+      "loss": 7.9804,
+      "step": 1884
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7159500639658842,
+      "learning_rate": 3.8021172018294236e-05,
+      "loss": 7.768,
+      "step": 1885
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.422663044053607,
+      "learning_rate": 3.801865927805097e-05,
+      "loss": 0.9872,
+      "step": 1886
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.2226256282250914,
+      "learning_rate": 3.801614502660451e-05,
+      "loss": 1.0185,
+      "step": 1887
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9913457729739066,
+      "learning_rate": 3.801362926416573e-05,
+      "loss": 1.0025,
+      "step": 1888
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1169463454098532,
+      "learning_rate": 3.801111199094561e-05,
+      "loss": 1.0198,
+      "step": 1889
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8204654329830579,
+      "learning_rate": 3.800859320715529e-05,
+      "loss": 0.9643,
+      "step": 1890
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.3895634291532453,
+      "learning_rate": 3.8006072913006e-05,
+      "loss": 7.7458,
+      "step": 1891
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9360082676793013,
+      "learning_rate": 3.800355110870912e-05,
+      "loss": 0.9926,
+      "step": 1892
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.519384918280522,
+      "learning_rate": 3.800102779447615e-05,
+      "loss": 0.9922,
+      "step": 1893
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5586308847636043,
+      "learning_rate": 3.799850297051873e-05,
+      "loss": 0.9734,
+      "step": 1894
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5940628910176505,
+      "learning_rate": 3.799597663704859e-05,
+      "loss": 0.9989,
+      "step": 1895
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6425721908350632,
+      "learning_rate": 3.799344879427763e-05,
+      "loss": 0.9823,
+      "step": 1896
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7015145468874328,
+      "learning_rate": 3.7990919442417846e-05,
+      "loss": 0.9554,
+      "step": 1897
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6904557432800054,
+      "learning_rate": 3.7988388581681375e-05,
+      "loss": 0.9861,
+      "step": 1898
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6788563535534483,
+      "learning_rate": 3.798585621228048e-05,
+      "loss": 0.9843,
+      "step": 1899
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6580173968547892,
+      "learning_rate": 3.798332233442755e-05,
+      "loss": 1.0073,
+      "step": 1900
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.711457230012597,
+      "learning_rate": 3.798078694833509e-05,
+      "loss": 0.9875,
+      "step": 1901
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5449654290305826,
+      "learning_rate": 3.7978250054215745e-05,
+      "loss": 0.9644,
+      "step": 1902
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6247963291880084,
+      "learning_rate": 3.797571165228227e-05,
+      "loss": 0.9835,
+      "step": 1903
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5569142110378007,
+      "learning_rate": 3.7973171742747584e-05,
+      "loss": 0.9675,
+      "step": 1904
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6034087022592481,
+      "learning_rate": 3.797063032582468e-05,
+      "loss": 0.976,
+      "step": 1905
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0753867487928976,
+      "learning_rate": 3.796808740172671e-05,
+      "loss": 0.9406,
+      "step": 1906
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5595968042097655,
+      "learning_rate": 3.796554297066696e-05,
+      "loss": 0.9808,
+      "step": 1907
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7515792930947157,
+      "learning_rate": 3.7962997032858814e-05,
+      "loss": 0.9472,
+      "step": 1908
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1754940499219022,
+      "learning_rate": 3.7960449588515806e-05,
+      "loss": 0.97,
+      "step": 1909
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.197878590974552,
+      "learning_rate": 3.795790063785158e-05,
+      "loss": 7.8376,
+      "step": 1910
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.294493807354091,
+      "learning_rate": 3.795535018107991e-05,
+      "loss": 1.0044,
+      "step": 1911
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 7.422019221481564,
+      "learning_rate": 3.7952798218414704e-05,
+      "loss": 1.164,
+      "step": 1912
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 18.681191125931438,
+      "learning_rate": 3.795024475007e-05,
+      "loss": 2.6593,
+      "step": 1913
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 9.700885449608595,
+      "learning_rate": 3.794768977625994e-05,
+      "loss": 1.712,
+      "step": 1914
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.4385964971613374,
+      "learning_rate": 3.794513329719882e-05,
+      "loss": 1.2659,
+      "step": 1915
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.191080886701433,
+      "learning_rate": 3.7942575313101036e-05,
+      "loss": 1.3269,
+      "step": 1916
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 5.4145917422212655,
+      "learning_rate": 3.794001582418113e-05,
+      "loss": 1.4486,
+      "step": 1917
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.621276042397796,
+      "learning_rate": 3.793745483065377e-05,
+      "loss": 1.3114,
+      "step": 1918
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.0768726004020808,
+      "learning_rate": 3.793489233273373e-05,
+      "loss": 1.2394,
+      "step": 1919
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8931553543990138,
+      "learning_rate": 3.7932328330635936e-05,
+      "loss": 1.1782,
+      "step": 1920
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6851871869159194,
+      "learning_rate": 3.7929762824575414e-05,
+      "loss": 1.1501,
+      "step": 1921
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.447958940937785,
+      "learning_rate": 3.7927195814767346e-05,
+      "loss": 1.2077,
+      "step": 1922
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7360683113035051,
+      "learning_rate": 3.792462730142701e-05,
+      "loss": 1.1495,
+      "step": 1923
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4065968175488297,
+      "learning_rate": 3.792205728476983e-05,
+      "loss": 1.1086,
+      "step": 1924
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1649816778131095,
+      "learning_rate": 3.7919485765011346e-05,
+      "loss": 1.1002,
+      "step": 1925
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1495772388970653,
+      "learning_rate": 3.791691274236724e-05,
+      "loss": 1.0905,
+      "step": 1926
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0387719034247016,
+      "learning_rate": 3.791433821705329e-05,
+      "loss": 1.0659,
+      "step": 1927
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4734643392350841,
+      "learning_rate": 3.7911762189285444e-05,
+      "loss": 1.0931,
+      "step": 1928
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0383367774472239,
+      "learning_rate": 3.790918465927973e-05,
+      "loss": 1.0825,
+      "step": 1929
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9659114058548549,
+      "learning_rate": 3.790660562725232e-05,
+      "loss": 1.0675,
+      "step": 1930
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8329574982760837,
+      "learning_rate": 3.7904025093419525e-05,
+      "loss": 1.0162,
+      "step": 1931
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8207150497733929,
+      "learning_rate": 3.790144305799777e-05,
+      "loss": 1.0341,
+      "step": 1932
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 3.812442440796094,
+      "learning_rate": 3.78988595212036e-05,
+      "loss": 1.0681,
+      "step": 1933
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8608412737313847,
+      "learning_rate": 3.789627448325371e-05,
+      "loss": 1.0396,
+      "step": 1934
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.777330449600005,
+      "learning_rate": 3.7893687944364884e-05,
+      "loss": 1.0335,
+      "step": 1935
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.3717399810229065,
+      "learning_rate": 3.789109990475406e-05,
+      "loss": 1.0063,
+      "step": 1936
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6574783969480316,
+      "learning_rate": 3.78885103646383e-05,
+      "loss": 1.0399,
+      "step": 1937
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7273020135953594,
+      "learning_rate": 3.788591932423477e-05,
+      "loss": 1.049,
+      "step": 1938
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.392338621496521,
+      "learning_rate": 3.7883326783760794e-05,
+      "loss": 0.9842,
+      "step": 1939
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7949171186371595,
+      "learning_rate": 3.788073274343379e-05,
+      "loss": 1.0239,
+      "step": 1940
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6536404492306394,
+      "learning_rate": 3.7878137203471334e-05,
+      "loss": 1.0248,
+      "step": 1941
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5232104226184451,
+      "learning_rate": 3.78755401640911e-05,
+      "loss": 1.005,
+      "step": 1942
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5990827949663391,
+      "learning_rate": 3.78729416255109e-05,
+      "loss": 0.9704,
+      "step": 1943
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6950087970721345,
+      "learning_rate": 3.787034158794867e-05,
+      "loss": 0.9884,
+      "step": 1944
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.2168695553641768,
+      "learning_rate": 3.7867740051622477e-05,
+      "loss": 1.0124,
+      "step": 1945
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.921961867083176,
+      "learning_rate": 3.78651370167505e-05,
+      "loss": 1.0091,
+      "step": 1946
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6524628225481564,
+      "learning_rate": 3.786253248355106e-05,
+      "loss": 0.9856,
+      "step": 1947
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.763609266996727,
+      "learning_rate": 3.785992645224259e-05,
+      "loss": 1.0189,
+      "step": 1948
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.390114577591212,
+      "learning_rate": 3.785731892304366e-05,
+      "loss": 1.0073,
+      "step": 1949
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.02482805681558,
+      "learning_rate": 3.785470989617296e-05,
+      "loss": 1.0007,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8134205429898977,
+      "learning_rate": 3.78520993718493e-05,
+      "loss": 0.9689,
+      "step": 1951
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6749411473698648,
+      "learning_rate": 3.784948735029163e-05,
+      "loss": 0.9742,
+      "step": 1952
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.053295381657855,
+      "learning_rate": 3.784687383171901e-05,
+      "loss": 0.9633,
+      "step": 1953
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 18.22572153520309,
+      "learning_rate": 3.7844258816350634e-05,
+      "loss": 8.5214,
+      "step": 1954
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 15.46282488713616,
+      "learning_rate": 3.784164230440582e-05,
+      "loss": 1.1523,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 9.074157663347698,
+      "learning_rate": 3.783902429610401e-05,
+      "loss": 1.0565,
+      "step": 1956
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1435811622199792,
+      "learning_rate": 3.783640479166478e-05,
+      "loss": 1.0136,
+      "step": 1957
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.6881886626227725,
+      "learning_rate": 3.783378379130783e-05,
+      "loss": 1.0444,
+      "step": 1958
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.9047739909221777,
+      "learning_rate": 3.783116129525296e-05,
+      "loss": 1.0196,
+      "step": 1959
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.1169638328760647,
+      "learning_rate": 3.7828537303720136e-05,
+      "loss": 1.0099,
+      "step": 1960
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0783221484552694,
+      "learning_rate": 3.782591181692941e-05,
+      "loss": 1.0489,
+      "step": 1961
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7019126829460671,
+      "learning_rate": 3.7823284835100994e-05,
+      "loss": 1.0466,
+      "step": 1962
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 5.75254403505191,
+      "learning_rate": 3.782065635845521e-05,
+      "loss": 8.2724,
+      "step": 1963
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3088686688923992,
+      "learning_rate": 3.7818026387212486e-05,
+      "loss": 1.0344,
+      "step": 1964
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8468188831524451,
+      "learning_rate": 3.7815394921593416e-05,
+      "loss": 1.0336,
+      "step": 1965
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2902855796161699,
+      "learning_rate": 3.781276196181868e-05,
+      "loss": 1.0207,
+      "step": 1966
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1973052122496528,
+      "learning_rate": 3.7810127508109125e-05,
+      "loss": 1.011,
+      "step": 1967
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8376239080205199,
+      "learning_rate": 3.7807491560685676e-05,
+      "loss": 1.0258,
+      "step": 1968
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8117994996773549,
+      "learning_rate": 3.780485411976942e-05,
+      "loss": 1.0088,
+      "step": 1969
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0216646123009092,
+      "learning_rate": 3.780221518558155e-05,
+      "loss": 1.0132,
+      "step": 1970
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1053377563045976,
+      "learning_rate": 3.779957475834339e-05,
+      "loss": 1.0039,
+      "step": 1971
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9320164728084801,
+      "learning_rate": 3.779693283827639e-05,
+      "loss": 1.0235,
+      "step": 1972
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7479541647338632,
+      "learning_rate": 3.7794289425602135e-05,
+      "loss": 0.9988,
+      "step": 1973
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7207254580134912,
+      "learning_rate": 3.779164452054231e-05,
+      "loss": 0.9872,
+      "step": 1974
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.066191000119419,
+      "learning_rate": 3.778899812331874e-05,
+      "loss": 1.0323,
+      "step": 1975
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.032535836833498,
+      "learning_rate": 3.7786350234153386e-05,
+      "loss": 0.9944,
+      "step": 1976
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6821477231907579,
+      "learning_rate": 3.7783700853268325e-05,
+      "loss": 1.0247,
+      "step": 1977
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.613855309026656,
+      "learning_rate": 3.7781049980885746e-05,
+      "loss": 0.9939,
+      "step": 1978
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.135272106919033,
+      "learning_rate": 3.777839761722797e-05,
+      "loss": 0.9848,
+      "step": 1979
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8345315775616692,
+      "learning_rate": 3.777574376251747e-05,
+      "loss": 1.0113,
+      "step": 1980
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7560533324284229,
+      "learning_rate": 3.77730884169768e-05,
+      "loss": 0.9731,
+      "step": 1981
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0474846123575727,
+      "learning_rate": 3.777043158082868e-05,
+      "loss": 1.0044,
+      "step": 1982
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8538939573631721,
+      "learning_rate": 3.7767773254295914e-05,
+      "loss": 0.9897,
+      "step": 1983
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9062812040006373,
+      "learning_rate": 3.7765113437601464e-05,
+      "loss": 0.9884,
+      "step": 1984
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8567546413668945,
+      "learning_rate": 3.776245213096841e-05,
+      "loss": 0.9762,
+      "step": 1985
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8914577999853245,
+      "learning_rate": 3.775978933461994e-05,
+      "loss": 1.0133,
+      "step": 1986
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8630030883601402,
+      "learning_rate": 3.7757125048779394e-05,
+      "loss": 1.0141,
+      "step": 1987
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7031819624563802,
+      "learning_rate": 3.775445927367021e-05,
+      "loss": 0.9793,
+      "step": 1988
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0942040719634694,
+      "learning_rate": 3.775179200951597e-05,
+      "loss": 1.0165,
+      "step": 1989
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7997379044026076,
+      "learning_rate": 3.774912325654038e-05,
+      "loss": 0.9962,
+      "step": 1990
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6378659004547688,
+      "learning_rate": 3.774645301496725e-05,
+      "loss": 1.0447,
+      "step": 1991
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7850784449893107,
+      "learning_rate": 3.774378128502054e-05,
+      "loss": 0.954,
+      "step": 1992
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5936302868982067,
+      "learning_rate": 3.7741108066924316e-05,
+      "loss": 1.0106,
+      "step": 1993
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6494491097071815,
+      "learning_rate": 3.77384333609028e-05,
+      "loss": 1.0062,
+      "step": 1994
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7885006967466419,
+      "learning_rate": 3.773575716718028e-05,
+      "loss": 0.9804,
+      "step": 1995
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7943904306173242,
+      "learning_rate": 3.7733079485981246e-05,
+      "loss": 1.008,
+      "step": 1996
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7518235902458053,
+      "learning_rate": 3.773040031753024e-05,
+      "loss": 0.9656,
+      "step": 1997
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5546188161100511,
+      "learning_rate": 3.7727719662051977e-05,
+      "loss": 0.956,
+      "step": 1998
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7626449152497516,
+      "learning_rate": 3.7725037519771275e-05,
+      "loss": 0.9782,
+      "step": 1999
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5133873834901085,
+      "learning_rate": 3.7722353890913084e-05,
+      "loss": 0.9917,
+      "step": 2000
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7212038478145738,
+      "learning_rate": 3.771966877570247e-05,
+      "loss": 0.9781,
+      "step": 2001
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6134181995289962,
+      "learning_rate": 3.7716982174364635e-05,
+      "loss": 0.985,
+      "step": 2002
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7925593565310034,
+      "learning_rate": 3.7714294087124904e-05,
+      "loss": 0.9809,
+      "step": 2003
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9865796508341537,
+      "learning_rate": 3.771160451420873e-05,
+      "loss": 0.9603,
+      "step": 2004
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7770838625758234,
+      "learning_rate": 3.770891345584166e-05,
+      "loss": 0.9964,
+      "step": 2005
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9164343431280467,
+      "learning_rate": 3.770622091224942e-05,
+      "loss": 0.9851,
+      "step": 2006
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9753767277947106,
+      "learning_rate": 3.7703526883657805e-05,
+      "loss": 0.9779,
+      "step": 2007
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6033872149339352,
+      "learning_rate": 3.7700831370292774e-05,
+      "loss": 1.0079,
+      "step": 2008
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6230526862443199,
+      "learning_rate": 3.7698134372380394e-05,
+      "loss": 0.9902,
+      "step": 2009
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6725341401917747,
+      "learning_rate": 3.769543589014686e-05,
+      "loss": 0.9995,
+      "step": 2010
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9569898429081605,
+      "learning_rate": 3.769273592381849e-05,
+      "loss": 0.9921,
+      "step": 2011
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8778838867500756,
+      "learning_rate": 3.7690034473621725e-05,
+      "loss": 1.0003,
+      "step": 2012
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6029303255776591,
+      "learning_rate": 3.768733153978312e-05,
+      "loss": 0.988,
+      "step": 2013
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8920425461743844,
+      "learning_rate": 3.768462712252939e-05,
+      "loss": 0.9764,
+      "step": 2014
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8363468527147688,
+      "learning_rate": 3.768192122208734e-05,
+      "loss": 0.9734,
+      "step": 2015
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5613447743558752,
+      "learning_rate": 3.767921383868392e-05,
+      "loss": 1.0077,
+      "step": 2016
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6307452790079415,
+      "learning_rate": 3.7676504972546176e-05,
+      "loss": 0.9552,
+      "step": 2017
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7629750058886002,
+      "learning_rate": 3.767379462390131e-05,
+      "loss": 0.9837,
+      "step": 2018
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7257295664152335,
+      "learning_rate": 3.767108279297663e-05,
+      "loss": 0.9577,
+      "step": 2019
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6835656403382394,
+      "learning_rate": 3.766836947999959e-05,
+      "loss": 0.9353,
+      "step": 2020
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4677574493971592,
+      "learning_rate": 3.766565468519772e-05,
+      "loss": 0.962,
+      "step": 2021
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7816169533584967,
+      "learning_rate": 3.766293840879874e-05,
+      "loss": 0.987,
+      "step": 2022
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7125808700368677,
+      "learning_rate": 3.766022065103045e-05,
+      "loss": 0.9658,
+      "step": 2023
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6640949572642338,
+      "learning_rate": 3.765750141212077e-05,
+      "loss": 0.9763,
+      "step": 2024
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8626212356053838,
+      "learning_rate": 3.765478069229778e-05,
+      "loss": 0.9754,
+      "step": 2025
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8138254032144248,
+      "learning_rate": 3.765205849178966e-05,
+      "loss": 0.9989,
+      "step": 2026
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7420589337537503,
+      "learning_rate": 3.76493348108247e-05,
+      "loss": 0.9918,
+      "step": 2027
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7829416365641347,
+      "learning_rate": 3.764660964963135e-05,
+      "loss": 0.9937,
+      "step": 2028
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8474909964552725,
+      "learning_rate": 3.7643883008438156e-05,
+      "loss": 0.9512,
+      "step": 2029
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6690238411204236,
+      "learning_rate": 3.764115488747381e-05,
+      "loss": 0.9655,
+      "step": 2030
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7839997753612417,
+      "learning_rate": 3.76384252869671e-05,
+      "loss": 0.9754,
+      "step": 2031
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9919917922635719,
+      "learning_rate": 3.763569420714697e-05,
+      "loss": 0.9632,
+      "step": 2032
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6969839271624305,
+      "learning_rate": 3.763296164824246e-05,
+      "loss": 0.9697,
+      "step": 2033
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6687551402937115,
+      "learning_rate": 3.763022761048276e-05,
+      "loss": 0.9687,
+      "step": 2034
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8066100711422906,
+      "learning_rate": 3.762749209409716e-05,
+      "loss": 0.9684,
+      "step": 2035
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7327124609802164,
+      "learning_rate": 3.7624755099315084e-05,
+      "loss": 0.9486,
+      "step": 2036
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7091716331236143,
+      "learning_rate": 3.762201662636608e-05,
+      "loss": 0.9817,
+      "step": 2037
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5959973209270243,
+      "learning_rate": 3.761927667547983e-05,
+      "loss": 0.9899,
+      "step": 2038
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7388402704027336,
+      "learning_rate": 3.761653524688612e-05,
+      "loss": 0.9673,
+      "step": 2039
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7784951836235265,
+      "learning_rate": 3.7613792340814874e-05,
+      "loss": 0.9822,
+      "step": 2040
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6874480951289142,
+      "learning_rate": 3.761104795749614e-05,
+      "loss": 0.9542,
+      "step": 2041
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5729008254264534,
+      "learning_rate": 3.7608302097160076e-05,
+      "loss": 0.9491,
+      "step": 2042
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9775436950799796,
+      "learning_rate": 3.760555476003699e-05,
+      "loss": 0.9749,
+      "step": 2043
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.901097383299905,
+      "learning_rate": 3.760280594635728e-05,
+      "loss": 0.9469,
+      "step": 2044
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8212182063730246,
+      "learning_rate": 3.76000556563515e-05,
+      "loss": 0.9415,
+      "step": 2045
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8426167095408184,
+      "learning_rate": 3.7597303890250305e-05,
+      "loss": 0.9487,
+      "step": 2046
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8424624033138901,
+      "learning_rate": 3.759455064828449e-05,
+      "loss": 0.9823,
+      "step": 2047
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8200674573503258,
+      "learning_rate": 3.7591795930684946e-05,
+      "loss": 0.9619,
+      "step": 2048
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7788211134184637,
+      "learning_rate": 3.758903973768274e-05,
+      "loss": 0.9624,
+      "step": 2049
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6011954338810722,
+      "learning_rate": 3.7586282069509e-05,
+      "loss": 0.9372,
+      "step": 2050
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6663210136408823,
+      "learning_rate": 3.7583522926395025e-05,
+      "loss": 0.9582,
+      "step": 2051
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8869656595383276,
+      "learning_rate": 3.7580762308572226e-05,
+      "loss": 0.9504,
+      "step": 2052
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9880263278721558,
+      "learning_rate": 3.757800021627212e-05,
+      "loss": 0.9622,
+      "step": 2053
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8850414521707465,
+      "learning_rate": 3.7575236649726364e-05,
+      "loss": 0.9481,
+      "step": 2054
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9376691408338363,
+      "learning_rate": 3.757247160916673e-05,
+      "loss": 0.9721,
+      "step": 2055
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8418063576942804,
+      "learning_rate": 3.7569705094825136e-05,
+      "loss": 0.9949,
+      "step": 2056
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0651970755305862,
+      "learning_rate": 3.756693710693358e-05,
+      "loss": 0.9453,
+      "step": 2057
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4266018823529132,
+      "learning_rate": 3.756416764572423e-05,
+      "loss": 0.9489,
+      "step": 2058
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8268087016010058,
+      "learning_rate": 3.756139671142936e-05,
+      "loss": 0.9476,
+      "step": 2059
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.2297294777727346,
+      "learning_rate": 3.755862430428135e-05,
+      "loss": 0.9657,
+      "step": 2060
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.1228331294942748,
+      "learning_rate": 3.755585042451273e-05,
+      "loss": 0.949,
+      "step": 2061
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.3437712849203471,
+      "learning_rate": 3.755307507235613e-05,
+      "loss": 0.9508,
+      "step": 2062
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9254857875733422,
+      "learning_rate": 3.755029824804433e-05,
+      "loss": 0.9424,
+      "step": 2063
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5615799152484822,
+      "learning_rate": 3.7547519951810196e-05,
+      "loss": 0.9555,
+      "step": 2064
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9657180728802042,
+      "learning_rate": 3.754474018388677e-05,
+      "loss": 0.9485,
+      "step": 2065
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9431115817293649,
+      "learning_rate": 3.754195894450717e-05,
+      "loss": 0.9583,
+      "step": 2066
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7574693781381374,
+      "learning_rate": 3.7539176233904654e-05,
+      "loss": 0.9231,
+      "step": 2067
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7589612457328366,
+      "learning_rate": 3.753639205231261e-05,
+      "loss": 0.9454,
+      "step": 2068
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7216579502712618,
+      "learning_rate": 3.753360639996455e-05,
+      "loss": 0.9223,
+      "step": 2069
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0239571302194004,
+      "learning_rate": 3.75308192770941e-05,
+      "loss": 0.9499,
+      "step": 2070
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.880481765465359,
+      "learning_rate": 3.7528030683935e-05,
+      "loss": 0.9244,
+      "step": 2071
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6593690524363901,
+      "learning_rate": 3.752524062072113e-05,
+      "loss": 0.9403,
+      "step": 2072
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7694150501379926,
+      "learning_rate": 3.752244908768651e-05,
+      "loss": 0.975,
+      "step": 2073
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4620409851366467,
+      "learning_rate": 3.7519656085065234e-05,
+      "loss": 0.9267,
+      "step": 2074
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6148287197271388,
+      "learning_rate": 3.751686161309157e-05,
+      "loss": 0.9585,
+      "step": 2075
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9622707135594311,
+      "learning_rate": 3.751406567199987e-05,
+      "loss": 0.9494,
+      "step": 2076
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2657555756193664,
+      "learning_rate": 3.7511268262024635e-05,
+      "loss": 0.9524,
+      "step": 2077
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8723855311595899,
+      "learning_rate": 3.750846938340048e-05,
+      "loss": 0.9839,
+      "step": 2078
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9970251654837307,
+      "learning_rate": 3.750566903636214e-05,
+      "loss": 0.9518,
+      "step": 2079
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8037460487766165,
+      "learning_rate": 3.7502867221144486e-05,
+      "loss": 0.9507,
+      "step": 2080
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6643678634973704,
+      "learning_rate": 3.750006393798249e-05,
+      "loss": 0.954,
+      "step": 2081
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8079909154919869,
+      "learning_rate": 3.749725918711127e-05,
+      "loss": 0.9296,
+      "step": 2082
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.235676966722044,
+      "learning_rate": 3.749445296876605e-05,
+      "loss": 0.9236,
+      "step": 2083
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3983614792442176,
+      "learning_rate": 3.749164528318219e-05,
+      "loss": 0.9508,
+      "step": 2084
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1338439896187356,
+      "learning_rate": 3.7488836130595165e-05,
+      "loss": 0.9324,
+      "step": 2085
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.380506045366588,
+      "learning_rate": 3.7486025511240574e-05,
+      "loss": 0.9562,
+      "step": 2086
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0856630073498903,
+      "learning_rate": 3.7483213425354145e-05,
+      "loss": 0.9566,
+      "step": 2087
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.4056490125141448,
+      "learning_rate": 3.748039987317171e-05,
+      "loss": 0.962,
+      "step": 2088
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.965413019991476,
+      "learning_rate": 3.747758485492926e-05,
+      "loss": 0.9408,
+      "step": 2089
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3001940577827762,
+      "learning_rate": 3.747476837086287e-05,
+      "loss": 0.9561,
+      "step": 2090
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7532956498019048,
+      "learning_rate": 3.7471950421208756e-05,
+      "loss": 0.9598,
+      "step": 2091
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8083547260650636,
+      "learning_rate": 3.746913100620327e-05,
+      "loss": 0.9844,
+      "step": 2092
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7463640623474515,
+      "learning_rate": 3.746631012608286e-05,
+      "loss": 0.9239,
+      "step": 2093
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7318263325196422,
+      "learning_rate": 3.7463487781084116e-05,
+      "loss": 0.9241,
+      "step": 2094
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2876463838303662,
+      "learning_rate": 3.746066397144374e-05,
+      "loss": 0.9322,
+      "step": 2095
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7835734006978523,
+      "learning_rate": 3.745783869739857e-05,
+      "loss": 0.9467,
+      "step": 2096
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8087195686162525,
+      "learning_rate": 3.745501195918554e-05,
+      "loss": 0.9531,
+      "step": 2097
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.717559554509749,
+      "learning_rate": 3.745218375704174e-05,
+      "loss": 0.9262,
+      "step": 2098
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6074662073702076,
+      "learning_rate": 3.7449354091204376e-05,
+      "loss": 0.9314,
+      "step": 2099
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6520705649904557,
+      "learning_rate": 3.744652296191075e-05,
+      "loss": 0.9425,
+      "step": 2100
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7884875285112254,
+      "learning_rate": 3.744369036939832e-05,
+      "loss": 0.9218,
+      "step": 2101
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8079378317759096,
+      "learning_rate": 3.744085631390463e-05,
+      "loss": 0.9163,
+      "step": 2102
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6072571910772571,
+      "learning_rate": 3.7438020795667397e-05,
+      "loss": 0.9352,
+      "step": 2103
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9960534260192992,
+      "learning_rate": 3.743518381492442e-05,
+      "loss": 0.9055,
+      "step": 2104
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7792214430829347,
+      "learning_rate": 3.743234537191363e-05,
+      "loss": 0.9374,
+      "step": 2105
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8129069487631961,
+      "learning_rate": 3.742950546687309e-05,
+      "loss": 0.9098,
+      "step": 2106
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.961665168385432,
+      "learning_rate": 3.742666410004097e-05,
+      "loss": 0.9456,
+      "step": 2107
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9274309386297817,
+      "learning_rate": 3.742382127165558e-05,
+      "loss": 0.9418,
+      "step": 2108
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.92851241317021,
+      "learning_rate": 3.742097698195534e-05,
+      "loss": 0.9577,
+      "step": 2109
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6905450231632975,
+      "learning_rate": 3.74181312311788e-05,
+      "loss": 0.9364,
+      "step": 2110
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.905085126080825,
+      "learning_rate": 3.7415284019564635e-05,
+      "loss": 0.9426,
+      "step": 2111
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0746498860674445,
+      "learning_rate": 3.741243534735163e-05,
+      "loss": 0.9561,
+      "step": 2112
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8239165019183583,
+      "learning_rate": 3.740958521477869e-05,
+      "loss": 0.9677,
+      "step": 2113
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9082682272031656,
+      "learning_rate": 3.7406733622084875e-05,
+      "loss": 0.9006,
+      "step": 2114
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1074678831592855,
+      "learning_rate": 3.740388056950933e-05,
+      "loss": 0.949,
+      "step": 2115
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7846040833640462,
+      "learning_rate": 3.7401026057291344e-05,
+      "loss": 0.9375,
+      "step": 2116
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5318644795461113,
+      "learning_rate": 3.739817008567031e-05,
+      "loss": 0.8965,
+      "step": 2117
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.4426420564113904,
+      "learning_rate": 3.739531265488578e-05,
+      "loss": 7.719,
+      "step": 2118
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3031329683806139,
+      "learning_rate": 3.739245376517737e-05,
+      "loss": 0.9493,
+      "step": 2119
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7528796789896992,
+      "learning_rate": 3.738959341678487e-05,
+      "loss": 0.9229,
+      "step": 2120
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6861292495604749,
+      "learning_rate": 3.738673160994817e-05,
+      "loss": 0.9222,
+      "step": 2121
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8043592225382384,
+      "learning_rate": 3.7383868344907295e-05,
+      "loss": 0.9316,
+      "step": 2122
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7476726816143803,
+      "learning_rate": 3.7381003621902366e-05,
+      "loss": 0.9348,
+      "step": 2123
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7632767014443147,
+      "learning_rate": 3.737813744117366e-05,
+      "loss": 0.9139,
+      "step": 2124
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9286782769473592,
+      "learning_rate": 3.7375269802961555e-05,
+      "loss": 0.9117,
+      "step": 2125
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9486371213860357,
+      "learning_rate": 3.737240070750655e-05,
+      "loss": 0.9283,
+      "step": 2126
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7320819038996084,
+      "learning_rate": 3.736953015504929e-05,
+      "loss": 0.9292,
+      "step": 2127
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8619238066282194,
+      "learning_rate": 3.7366658145830506e-05,
+      "loss": 0.9371,
+      "step": 2128
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8244669157600494,
+      "learning_rate": 3.7363784680091074e-05,
+      "loss": 0.9516,
+      "step": 2129
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1425516636073285,
+      "learning_rate": 3.7360909758072e-05,
+      "loss": 0.9484,
+      "step": 2130
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8211481489217092,
+      "learning_rate": 3.735803338001438e-05,
+      "loss": 0.898,
+      "step": 2131
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5809464725630202,
+      "learning_rate": 3.7355155546159466e-05,
+      "loss": 0.916,
+      "step": 2132
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7378725274996535,
+      "learning_rate": 3.735227625674863e-05,
+      "loss": 0.9272,
+      "step": 2133
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8703825040578916,
+      "learning_rate": 3.734939551202332e-05,
+      "loss": 0.9277,
+      "step": 2134
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8543412990396902,
+      "learning_rate": 3.7346513312225175e-05,
+      "loss": 0.9233,
+      "step": 2135
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8223367697636589,
+      "learning_rate": 3.73436296575959e-05,
+      "loss": 0.9332,
+      "step": 2136
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 4.12778074432374,
+      "learning_rate": 3.7340744548377357e-05,
+      "loss": 7.8608,
+      "step": 2137
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0272287915436527,
+      "learning_rate": 3.733785798481151e-05,
+      "loss": 0.8962,
+      "step": 2138
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1325540288174427,
+      "learning_rate": 3.7334969967140447e-05,
+      "loss": 0.9219,
+      "step": 2139
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0471769367770536,
+      "learning_rate": 3.733208049560639e-05,
+      "loss": 0.9445,
+      "step": 2140
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9417747474900553,
+      "learning_rate": 3.7329189570451684e-05,
+      "loss": 0.8815,
+      "step": 2141
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9637181501228671,
+      "learning_rate": 3.732629719191876e-05,
+      "loss": 0.9092,
+      "step": 2142
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1103028327531088,
+      "learning_rate": 3.732340336025023e-05,
+      "loss": 0.8953,
+      "step": 2143
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2259805637407992,
+      "learning_rate": 3.732050807568878e-05,
+      "loss": 0.956,
+      "step": 2144
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9665530380159955,
+      "learning_rate": 3.7317611338477225e-05,
+      "loss": 0.941,
+      "step": 2145
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6730656782267664,
+      "learning_rate": 3.731471314885853e-05,
+      "loss": 0.9027,
+      "step": 2146
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.037165296763896,
+      "learning_rate": 3.731181350707576e-05,
+      "loss": 0.9221,
+      "step": 2147
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0463653128596897,
+      "learning_rate": 3.7308912413372086e-05,
+      "loss": 0.91,
+      "step": 2148
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8749176189663391,
+      "learning_rate": 3.7306009867990836e-05,
+      "loss": 0.9201,
+      "step": 2149
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0698850425264599,
+      "learning_rate": 3.730310587117544e-05,
+      "loss": 0.9187,
+      "step": 2150
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0401458336514142,
+      "learning_rate": 3.7300200423169453e-05,
+      "loss": 0.8995,
+      "step": 2151
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1050430470613677,
+      "learning_rate": 3.729729352421655e-05,
+      "loss": 0.842,
+      "step": 2152
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0227709830011023,
+      "learning_rate": 3.729438517456052e-05,
+      "loss": 0.9007,
+      "step": 2153
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6280373602323007,
+      "learning_rate": 3.72914753744453e-05,
+      "loss": 0.8998,
+      "step": 2154
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0111213698819326,
+      "learning_rate": 3.728856412411492e-05,
+      "loss": 0.9148,
+      "step": 2155
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1219914869466148,
+      "learning_rate": 3.7285651423813545e-05,
+      "loss": 0.9125,
+      "step": 2156
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2449121440627389,
+      "learning_rate": 3.7282737273785466e-05,
+      "loss": 0.8856,
+      "step": 2157
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8738538135829106,
+      "learning_rate": 3.727982167427509e-05,
+      "loss": 0.9485,
+      "step": 2158
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.030418894211106,
+      "learning_rate": 3.7276904625526926e-05,
+      "loss": 0.877,
+      "step": 2159
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.1115523776338418,
+      "learning_rate": 3.7273986127785644e-05,
+      "loss": 0.9414,
+      "step": 2160
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.2782515959089256,
+      "learning_rate": 3.7271066181296e-05,
+      "loss": 0.927,
+      "step": 2161
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.212589485039888,
+      "learning_rate": 3.72681447863029e-05,
+      "loss": 0.9375,
+      "step": 2162
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6476391589040735,
+      "learning_rate": 3.726522194305135e-05,
+      "loss": 0.8824,
+      "step": 2163
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.964856300801811,
+      "learning_rate": 3.726229765178648e-05,
+      "loss": 0.9207,
+      "step": 2164
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9971787357347416,
+      "learning_rate": 3.725937191275356e-05,
+      "loss": 0.9195,
+      "step": 2165
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7150388884635723,
+      "learning_rate": 3.725644472619796e-05,
+      "loss": 0.918,
+      "step": 2166
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0758027284160447,
+      "learning_rate": 3.7253516092365184e-05,
+      "loss": 0.9376,
+      "step": 2167
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7912554764414688,
+      "learning_rate": 3.725058601150085e-05,
+      "loss": 0.9122,
+      "step": 2168
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9252633342263652,
+      "learning_rate": 3.724765448385069e-05,
+      "loss": 0.9217,
+      "step": 2169
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6678941550016222,
+      "learning_rate": 3.72447215096606e-05,
+      "loss": 0.9135,
+      "step": 2170
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0545423993946337,
+      "learning_rate": 3.724178708917653e-05,
+      "loss": 0.9104,
+      "step": 2171
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8224041997450082,
+      "learning_rate": 3.72388512226446e-05,
+      "loss": 0.8808,
+      "step": 2172
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6842038609303543,
+      "learning_rate": 3.7235913910311034e-05,
+      "loss": 0.9197,
+      "step": 2173
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.047974312153035,
+      "learning_rate": 3.723297515242219e-05,
+      "loss": 0.94,
+      "step": 2174
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.550412655421197,
+      "learning_rate": 3.723003494922453e-05,
+      "loss": 0.9191,
+      "step": 2175
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6287870950592035,
+      "learning_rate": 3.7227093300964646e-05,
+      "loss": 0.9273,
+      "step": 2176
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7769585760132174,
+      "learning_rate": 3.722415020788925e-05,
+      "loss": 0.9385,
+      "step": 2177
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6552170065243489,
+      "learning_rate": 3.7221205670245186e-05,
+      "loss": 0.8849,
+      "step": 2178
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8507585570329254,
+      "learning_rate": 3.721825968827939e-05,
+      "loss": 0.9321,
+      "step": 2179
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.965931821006843,
+      "learning_rate": 3.7215312262238956e-05,
+      "loss": 0.9081,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8838585286584537,
+      "learning_rate": 3.7212363392371074e-05,
+      "loss": 0.9207,
+      "step": 2181
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0152312068453522,
+      "learning_rate": 3.720941307892305e-05,
+      "loss": 0.9152,
+      "step": 2182
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8402094695556429,
+      "learning_rate": 3.720646132214235e-05,
+      "loss": 0.9188,
+      "step": 2183
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8013216584216067,
+      "learning_rate": 3.7203508122276517e-05,
+      "loss": 0.8999,
+      "step": 2184
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7710323037346198,
+      "learning_rate": 3.7200553479573225e-05,
+      "loss": 0.8861,
+      "step": 2185
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0373836349303007,
+      "learning_rate": 3.7197597394280294e-05,
+      "loss": 0.91,
+      "step": 2186
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7460991738840418,
+      "learning_rate": 3.719463986664563e-05,
+      "loss": 0.8916,
+      "step": 2187
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1395542467882125,
+      "learning_rate": 3.7191680896917296e-05,
+      "loss": 0.8873,
+      "step": 2188
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0069446575772807,
+      "learning_rate": 3.718872048534344e-05,
+      "loss": 0.9208,
+      "step": 2189
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.955019898924206,
+      "learning_rate": 3.7185758632172365e-05,
+      "loss": 8.0661,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1153988079028392,
+      "learning_rate": 3.718279533765246e-05,
+      "loss": 0.9248,
+      "step": 2191
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9004792976061196,
+      "learning_rate": 3.717983060203226e-05,
+      "loss": 0.9076,
+      "step": 2192
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7256839961767696,
+      "learning_rate": 3.7176864425560426e-05,
+      "loss": 0.9162,
+      "step": 2193
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6510234082543034,
+      "learning_rate": 3.717389680848571e-05,
+      "loss": 0.9213,
+      "step": 2194
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9608458234832332,
+      "learning_rate": 3.717092775105701e-05,
+      "loss": 0.8845,
+      "step": 2195
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.917258819626189,
+      "learning_rate": 3.7167957253523334e-05,
+      "loss": 0.9321,
+      "step": 2196
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0684688189234668,
+      "learning_rate": 3.716498531613382e-05,
+      "loss": 0.9599,
+      "step": 2197
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9281915583218022,
+      "learning_rate": 3.7162011939137715e-05,
+      "loss": 0.8968,
+      "step": 2198
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.465221590118682,
+      "learning_rate": 3.71590371227844e-05,
+      "loss": 0.9579,
+      "step": 2199
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1031984837090099,
+      "learning_rate": 3.7156060867323356e-05,
+      "loss": 0.9128,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7457855607651123,
+      "learning_rate": 3.715308317300421e-05,
+      "loss": 0.9401,
+      "step": 2201
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9893020383650246,
+      "learning_rate": 3.71501040400767e-05,
+      "loss": 0.9171,
+      "step": 2202
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.767242817539106,
+      "learning_rate": 3.7147123468790674e-05,
+      "loss": 8.1464,
+      "step": 2203
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6473877358773518,
+      "learning_rate": 3.7144141459396106e-05,
+      "loss": 0.9133,
+      "step": 2204
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8585376745911752,
+      "learning_rate": 3.7141158012143106e-05,
+      "loss": 0.9063,
+      "step": 2205
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0547500367968006,
+      "learning_rate": 3.7138173127281875e-05,
+      "loss": 0.8837,
+      "step": 2206
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7514108850414096,
+      "learning_rate": 3.713518680506277e-05,
+      "loss": 0.9004,
+      "step": 2207
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7620711343389799,
+      "learning_rate": 3.7132199045736236e-05,
+      "loss": 0.9203,
+      "step": 2208
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5883102088556793,
+      "learning_rate": 3.7129209849552865e-05,
+      "loss": 0.9555,
+      "step": 2209
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9516078493944095,
+      "learning_rate": 3.712621921676335e-05,
+      "loss": 0.9255,
+      "step": 2210
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9287019499356719,
+      "learning_rate": 3.712322714761851e-05,
+      "loss": 0.8823,
+      "step": 2211
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7138765480130058,
+      "learning_rate": 3.712023364236928e-05,
+      "loss": 0.9494,
+      "step": 2212
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8230899487327326,
+      "learning_rate": 3.7117238701266744e-05,
+      "loss": 0.8769,
+      "step": 2213
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7020917093246589,
+      "learning_rate": 3.7114242324562066e-05,
+      "loss": 0.8846,
+      "step": 2214
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6802438697674749,
+      "learning_rate": 3.711124451250655e-05,
+      "loss": 0.8983,
+      "step": 2215
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7403534007691095,
+      "learning_rate": 3.710824526535163e-05,
+      "loss": 0.8917,
+      "step": 2216
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8085536338677736,
+      "learning_rate": 3.710524458334883e-05,
+      "loss": 0.906,
+      "step": 2217
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7937927080570278,
+      "learning_rate": 3.710224246674983e-05,
+      "loss": 0.9017,
+      "step": 2218
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.138343701181303,
+      "learning_rate": 3.7099238915806415e-05,
+      "loss": 0.8884,
+      "step": 2219
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.64389680049089,
+      "learning_rate": 3.709623393077047e-05,
+      "loss": 0.8962,
+      "step": 2220
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6123759937346777,
+      "learning_rate": 3.709322751189404e-05,
+      "loss": 0.8933,
+      "step": 2221
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.470734546164863,
+      "learning_rate": 3.709021965942926e-05,
+      "loss": 7.6293,
+      "step": 2222
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.5636771069371007,
+      "learning_rate": 3.70872103736284e-05,
+      "loss": 0.9046,
+      "step": 2223
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0275788446099838,
+      "learning_rate": 3.708419965474384e-05,
+      "loss": 0.8855,
+      "step": 2224
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6666032565654557,
+      "learning_rate": 3.7081187503028084e-05,
+      "loss": 0.9081,
+      "step": 2225
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7033426842681916,
+      "learning_rate": 3.7078173918733764e-05,
+      "loss": 0.879,
+      "step": 2226
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.963319443054436,
+      "learning_rate": 3.707515890211363e-05,
+      "loss": 0.8931,
+      "step": 2227
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1582877364191249,
+      "learning_rate": 3.707214245342053e-05,
+      "loss": 7.6011,
+      "step": 2228
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0933918002663736,
+      "learning_rate": 3.706912457290746e-05,
+      "loss": 0.9162,
+      "step": 2229
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7759736417141794,
+      "learning_rate": 3.7066105260827534e-05,
+      "loss": 0.8887,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8538581697305944,
+      "learning_rate": 3.7063084517433966e-05,
+      "loss": 0.8719,
+      "step": 2231
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9041695311532228,
+      "learning_rate": 3.706006234298011e-05,
+      "loss": 0.9197,
+      "step": 2232
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6499813416817392,
+      "learning_rate": 3.705703873771942e-05,
+      "loss": 0.9238,
+      "step": 2233
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6991264290148989,
+      "learning_rate": 3.70540137019055e-05,
+      "loss": 0.904,
+      "step": 2234
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8272243346096877,
+      "learning_rate": 3.7050987235792044e-05,
+      "loss": 0.9418,
+      "step": 2235
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8114594294173325,
+      "learning_rate": 3.704795933963288e-05,
+      "loss": 0.8964,
+      "step": 2236
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.754740590475309,
+      "learning_rate": 3.704493001368195e-05,
+      "loss": 0.9078,
+      "step": 2237
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9169449731884536,
+      "learning_rate": 3.704189925819333e-05,
+      "loss": 0.9033,
+      "step": 2238
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9002588675367763,
+      "learning_rate": 3.70388670734212e-05,
+      "loss": 0.8862,
+      "step": 2239
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8355380034548875,
+      "learning_rate": 3.703583345961986e-05,
+      "loss": 0.9049,
+      "step": 2240
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7999229484161744,
+      "learning_rate": 3.703279841704376e-05,
+      "loss": 0.8608,
+      "step": 2241
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6738588023300085,
+      "learning_rate": 3.702976194594741e-05,
+      "loss": 0.8969,
+      "step": 2242
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9926317117565295,
+      "learning_rate": 3.7026724046585506e-05,
+      "loss": 0.8806,
+      "step": 2243
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.870704614870112,
+      "learning_rate": 3.702368471921281e-05,
+      "loss": 0.9265,
+      "step": 2244
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6351484697551021,
+      "learning_rate": 3.7020643964084234e-05,
+      "loss": 0.8798,
+      "step": 2245
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.596979280013882,
+      "learning_rate": 3.701760178145481e-05,
+      "loss": 7.6876,
+      "step": 2246
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8521876500359453,
+      "learning_rate": 3.701455817157967e-05,
+      "loss": 0.9083,
+      "step": 2247
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7404078997600728,
+      "learning_rate": 3.70115131347141e-05,
+      "loss": 0.8877,
+      "step": 2248
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8553668107096121,
+      "learning_rate": 3.7008466671113444e-05,
+      "loss": 0.9049,
+      "step": 2249
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9926447767296213,
+      "learning_rate": 3.700541878103325e-05,
+      "loss": 0.8718,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9188527296171305,
+      "learning_rate": 3.700236946472911e-05,
+      "loss": 0.8807,
+      "step": 2251
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.756867144599537,
+      "learning_rate": 3.699931872245678e-05,
+      "loss": 0.8687,
+      "step": 2252
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7774465364591066,
+      "learning_rate": 3.6996266554472115e-05,
+      "loss": 0.8864,
+      "step": 2253
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6287321628133723,
+      "learning_rate": 3.69932129610311e-05,
+      "loss": 0.8921,
+      "step": 2254
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8215112751605155,
+      "learning_rate": 3.699015794238984e-05,
+      "loss": 0.8703,
+      "step": 2255
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1466374032062574,
+      "learning_rate": 3.6987101498804556e-05,
+      "loss": 0.8502,
+      "step": 2256
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.911053461761421,
+      "learning_rate": 3.698404363053158e-05,
+      "loss": 0.8617,
+      "step": 2257
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0042167276842393,
+      "learning_rate": 3.698098433782738e-05,
+      "loss": 0.8592,
+      "step": 2258
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4615715536329936,
+      "learning_rate": 3.697792362094853e-05,
+      "loss": 7.5699,
+      "step": 2259
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.694968190469987,
+      "learning_rate": 3.697486148015172e-05,
+      "loss": 0.9079,
+      "step": 2260
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7949807337446789,
+      "learning_rate": 3.697179791569379e-05,
+      "loss": 0.8337,
+      "step": 2261
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.3073351452429358,
+      "learning_rate": 3.696873292783166e-05,
+      "loss": 0.8388,
+      "step": 2262
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2210269340147162,
+      "learning_rate": 3.6965666516822395e-05,
+      "loss": 0.9058,
+      "step": 2263
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.212239702491679,
+      "learning_rate": 3.696259868292317e-05,
+      "loss": 0.8503,
+      "step": 2264
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0072990825217016,
+      "learning_rate": 3.695952942639128e-05,
+      "loss": 0.8583,
+      "step": 2265
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0226481065059718,
+      "learning_rate": 3.6956458747484136e-05,
+      "loss": 0.8561,
+      "step": 2266
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5995406615431084,
+      "learning_rate": 3.695338664645927e-05,
+      "loss": 0.8733,
+      "step": 2267
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.128979358295932,
+      "learning_rate": 3.695031312357435e-05,
+      "loss": 0.8813,
+      "step": 2268
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6715859649040231,
+      "learning_rate": 3.6947238179087134e-05,
+      "loss": 0.8403,
+      "step": 2269
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0802509523367012,
+      "learning_rate": 3.6944161813255514e-05,
+      "loss": 0.8751,
+      "step": 2270
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6257606997172344,
+      "learning_rate": 3.6941084026337514e-05,
+      "loss": 0.8534,
+      "step": 2271
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6632190473440063,
+      "learning_rate": 3.693800481859126e-05,
+      "loss": 0.882,
+      "step": 2272
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5101285681098336,
+      "learning_rate": 3.693492419027499e-05,
+      "loss": 0.8764,
+      "step": 2273
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5098597961258792,
+      "learning_rate": 3.693184214164708e-05,
+      "loss": 0.8419,
+      "step": 2274
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6190304737715709,
+      "learning_rate": 3.692875867296602e-05,
+      "loss": 0.8563,
+      "step": 2275
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.547900478615309,
+      "learning_rate": 3.692567378449042e-05,
+      "loss": 0.8545,
+      "step": 2276
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6091135056861874,
+      "learning_rate": 3.692258747647899e-05,
+      "loss": 0.8511,
+      "step": 2277
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 4.577674581337048,
+      "learning_rate": 3.6919499749190594e-05,
+      "loss": 7.539,
+      "step": 2278
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6071022529439896,
+      "learning_rate": 3.691641060288419e-05,
+      "loss": 0.8411,
+      "step": 2279
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7006191663113873,
+      "learning_rate": 3.691332003781885e-05,
+      "loss": 0.8186,
+      "step": 2280
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.549891692602598,
+      "learning_rate": 3.691022805425378e-05,
+      "loss": 0.8655,
+      "step": 2281
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7706179731421533,
+      "learning_rate": 3.690713465244832e-05,
+      "loss": 0.8305,
+      "step": 2282
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5917349413621008,
+      "learning_rate": 3.690403983266189e-05,
+      "loss": 0.8137,
+      "step": 2283
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6962924903750811,
+      "learning_rate": 3.6900943595154055e-05,
+      "loss": 0.839,
+      "step": 2284
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5726260274844539,
+      "learning_rate": 3.689784594018449e-05,
+      "loss": 0.8092,
+      "step": 2285
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6598266675425164,
+      "learning_rate": 3.6894746868012995e-05,
+      "loss": 0.8447,
+      "step": 2286
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5672242127935972,
+      "learning_rate": 3.6891646378899484e-05,
+      "loss": 0.8015,
+      "step": 2287
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6484805363270184,
+      "learning_rate": 3.688854447310398e-05,
+      "loss": 0.8243,
+      "step": 2288
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7318589710727421,
+      "learning_rate": 3.6885441150886664e-05,
+      "loss": 0.8249,
+      "step": 2289
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.49070114259886394,
+      "learning_rate": 3.6882336412507786e-05,
+      "loss": 0.7587,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7758208392195949,
+      "learning_rate": 3.687923025822774e-05,
+      "loss": 0.8626,
+      "step": 2291
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8441259868302853,
+      "learning_rate": 3.6876122688307034e-05,
+      "loss": 7.4867,
+      "step": 2292
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.686548629082471,
+      "learning_rate": 3.6873013703006304e-05,
+      "loss": 0.8372,
+      "step": 2293
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8708421879663935,
+      "learning_rate": 3.6869903302586283e-05,
+      "loss": 7.6701,
+      "step": 2294
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8272283557140951,
+      "learning_rate": 3.686679148730786e-05,
+      "loss": 0.788,
+      "step": 2295
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5446502453386763,
+      "learning_rate": 3.6863678257432e-05,
+      "loss": 0.7966,
+      "step": 2296
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5371024553817686,
+      "learning_rate": 3.6860563613219805e-05,
+      "loss": 0.7847,
+      "step": 2297
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5383375196416568,
+      "learning_rate": 3.68574475549325e-05,
+      "loss": 0.8261,
+      "step": 2298
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6118046999872234,
+      "learning_rate": 3.685433008283144e-05,
+      "loss": 0.8306,
+      "step": 2299
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.547414967634106,
+      "learning_rate": 3.685121119717806e-05,
+      "loss": 0.8174,
+      "step": 2300
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.557478404669372,
+      "learning_rate": 3.684809089823396e-05,
+      "loss": 0.8499,
+      "step": 2301
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7629586640572783,
+      "learning_rate": 3.684496918626081e-05,
+      "loss": 0.7848,
+      "step": 2302
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6139039990589346,
+      "learning_rate": 3.6841846061520446e-05,
+      "loss": 0.794,
+      "step": 2303
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5411088629692621,
+      "learning_rate": 3.683872152427479e-05,
+      "loss": 0.8298,
+      "step": 2304
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 4.818283522176468,
+      "learning_rate": 3.6835595574785895e-05,
+      "loss": 7.4249,
+      "step": 2305
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9512954955418755,
+      "learning_rate": 3.6832468213315926e-05,
+      "loss": 7.5358,
+      "step": 2306
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.86383500489586,
+      "learning_rate": 3.6829339440127186e-05,
+      "loss": 0.776,
+      "step": 2307
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5636293962410024,
+      "learning_rate": 3.682620925548207e-05,
+      "loss": 0.7396,
+      "step": 2308
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7263467147017998,
+      "learning_rate": 3.68230776596431e-05,
+      "loss": 0.7659,
+      "step": 2309
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6068886435249901,
+      "learning_rate": 3.681994465287293e-05,
+      "loss": 0.8183,
+      "step": 2310
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.3147896184719265,
+      "learning_rate": 3.6816810235434316e-05,
+      "loss": 0.7814,
+      "step": 2311
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 3.4684619114870765,
+      "learning_rate": 3.681367440759013e-05,
+      "loss": 7.3994,
+      "step": 2312
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7101028723576944,
+      "learning_rate": 3.681053716960338e-05,
+      "loss": 0.7612,
+      "step": 2313
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7387595331953126,
+      "learning_rate": 3.680739852173718e-05,
+      "loss": 0.7498,
+      "step": 2314
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5349010007014263,
+      "learning_rate": 3.680425846425477e-05,
+      "loss": 0.7279,
+      "step": 2315
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.190934470384835,
+      "learning_rate": 3.6801116997419486e-05,
+      "loss": 0.772,
+      "step": 2316
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4629226646801553,
+      "learning_rate": 3.679797412149482e-05,
+      "loss": 0.7367,
+      "step": 2317
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4899253775889012,
+      "learning_rate": 3.6794829836744344e-05,
+      "loss": 0.7516,
+      "step": 2318
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.537606446492008,
+      "learning_rate": 3.6791684143431774e-05,
+      "loss": 0.804,
+      "step": 2319
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6100354810920157,
+      "learning_rate": 3.6788537041820934e-05,
+      "loss": 0.6937,
+      "step": 2320
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6736837099508904,
+      "learning_rate": 3.6785388532175773e-05,
+      "loss": 0.7214,
+      "step": 2321
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4960809868568116,
+      "learning_rate": 3.6782238614760344e-05,
+      "loss": 0.7649,
+      "step": 2322
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5292660079201222,
+      "learning_rate": 3.677908728983883e-05,
+      "loss": 0.7527,
+      "step": 2323
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5379116764381567,
+      "learning_rate": 3.677593455767553e-05,
+      "loss": 0.7763,
+      "step": 2324
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5435723299666465,
+      "learning_rate": 3.6772780418534856e-05,
+      "loss": 0.7517,
+      "step": 2325
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5938526625006452,
+      "learning_rate": 3.676962487268134e-05,
+      "loss": 0.7443,
+      "step": 2326
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5068977069515198,
+      "learning_rate": 3.676646792037965e-05,
+      "loss": 0.7082,
+      "step": 2327
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5215807406325709,
+      "learning_rate": 3.6763309561894544e-05,
+      "loss": 0.6629,
+      "step": 2328
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6544769155990071,
+      "learning_rate": 3.67601497974909e-05,
+      "loss": 0.7257,
+      "step": 2329
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5648614930541149,
+      "learning_rate": 3.675698862743374e-05,
+      "loss": 0.7284,
+      "step": 2330
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.49600887982449837,
+      "learning_rate": 3.675382605198818e-05,
+      "loss": 0.7408,
+      "step": 2331
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5278533913623604,
+      "learning_rate": 3.675066207141946e-05,
+      "loss": 0.7338,
+      "step": 2332
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5634808772311186,
+      "learning_rate": 3.6747496685992945e-05,
+      "loss": 0.7252,
+      "step": 2333
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5405597478601002,
+      "learning_rate": 3.674432989597411e-05,
+      "loss": 0.7087,
+      "step": 2334
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4477718682268489,
+      "learning_rate": 3.674116170162855e-05,
+      "loss": 0.7055,
+      "step": 2335
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4871701406478126,
+      "learning_rate": 3.6737992103221974e-05,
+      "loss": 0.6923,
+      "step": 2336
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5711094225453882,
+      "learning_rate": 3.673482110102021e-05,
+      "loss": 0.7087,
+      "step": 2337
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.45260108338846705,
+      "learning_rate": 3.6731648695289214e-05,
+      "loss": 0.739,
+      "step": 2338
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5164516061094354,
+      "learning_rate": 3.6728474886295053e-05,
+      "loss": 0.7011,
+      "step": 2339
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5437607840131722,
+      "learning_rate": 3.67252996743039e-05,
+      "loss": 0.7326,
+      "step": 2340
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.45685521577084326,
+      "learning_rate": 3.672212305958206e-05,
+      "loss": 0.7004,
+      "step": 2341
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4751944697463827,
+      "learning_rate": 3.671894504239596e-05,
+      "loss": 0.7318,
+      "step": 2342
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4475072112630098,
+      "learning_rate": 3.671576562301213e-05,
+      "loss": 0.696,
+      "step": 2343
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4585132665518122,
+      "learning_rate": 3.6712584801697224e-05,
+      "loss": 0.7563,
+      "step": 2344
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.47087097692174934,
+      "learning_rate": 3.670940257871801e-05,
+      "loss": 0.7268,
+      "step": 2345
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.44771332562218413,
+      "learning_rate": 3.670621895434139e-05,
+      "loss": 0.7313,
+      "step": 2346
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4176705289252314,
+      "learning_rate": 3.670303392883436e-05,
+      "loss": 0.7102,
+      "step": 2347
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.47693413285261127,
+      "learning_rate": 3.669984750246404e-05,
+      "loss": 0.6869,
+      "step": 2348
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5325908316849872,
+      "learning_rate": 3.669665967549769e-05,
+      "loss": 0.7288,
+      "step": 2349
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5421211659594769,
+      "learning_rate": 3.6693470448202654e-05,
+      "loss": 0.7003,
+      "step": 2350
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4339156691966681,
+      "learning_rate": 3.66902798208464e-05,
+      "loss": 0.7205,
+      "step": 2351
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5200657715672793,
+      "learning_rate": 3.668708779369655e-05,
+      "loss": 0.7006,
+      "step": 2352
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4279377249385664,
+      "learning_rate": 3.6683894367020785e-05,
+      "loss": 0.6861,
+      "step": 2353
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.44887190315558795,
+      "learning_rate": 3.6680699541086964e-05,
+      "loss": 0.7205,
+      "step": 2354
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7628697764605055,
+      "learning_rate": 3.667750331616301e-05,
+      "loss": 0.6992,
+      "step": 2355
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.45526619356866255,
+      "learning_rate": 3.667430569251699e-05,
+      "loss": 0.7469,
+      "step": 2356
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0357970784319583,
+      "learning_rate": 3.6671106670417096e-05,
+      "loss": 0.6868,
+      "step": 2357
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4377430629154334,
+      "learning_rate": 3.666790625013161e-05,
+      "loss": 0.7384,
+      "step": 2358
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.46294750063112305,
+      "learning_rate": 3.666470443192897e-05,
+      "loss": 0.6768,
+      "step": 2359
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.48626921366705717,
+      "learning_rate": 3.666150121607769e-05,
+      "loss": 0.7382,
+      "step": 2360
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.40695718566732664,
+      "learning_rate": 3.665829660284642e-05,
+      "loss": 0.6611,
+      "step": 2361
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3936966583021227,
+      "learning_rate": 3.665509059250394e-05,
+      "loss": 0.7328,
+      "step": 2362
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5529385829010128,
+      "learning_rate": 3.6651883185319114e-05,
+      "loss": 0.7082,
+      "step": 2363
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.507680305287082,
+      "learning_rate": 3.664867438156096e-05,
+      "loss": 0.7216,
+      "step": 2364
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6170793276450899,
+      "learning_rate": 3.6645464181498596e-05,
+      "loss": 0.6728,
+      "step": 2365
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5176081068950867,
+      "learning_rate": 3.6642252585401257e-05,
+      "loss": 0.724,
+      "step": 2366
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5959063560846646,
+      "learning_rate": 3.6639039593538286e-05,
+      "loss": 0.7557,
+      "step": 2367
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4819464844533694,
+      "learning_rate": 3.6635825206179164e-05,
+      "loss": 0.7295,
+      "step": 2368
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.45234561098399095,
+      "learning_rate": 3.663260942359347e-05,
+      "loss": 0.6993,
+      "step": 2369
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.445874647149771,
+      "learning_rate": 3.662939224605091e-05,
+      "loss": 0.7237,
+      "step": 2370
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4270731388972814,
+      "learning_rate": 3.66261736738213e-05,
+      "loss": 0.7114,
+      "step": 2371
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4651550467655865,
+      "learning_rate": 3.66229537071746e-05,
+      "loss": 0.7342,
+      "step": 2372
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5164266923711124,
+      "learning_rate": 3.6619732346380834e-05,
+      "loss": 0.6616,
+      "step": 2373
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.49146383901588153,
+      "learning_rate": 3.6616509591710195e-05,
+      "loss": 0.7571,
+      "step": 2374
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5072715505516159,
+      "learning_rate": 3.661328544343295e-05,
+      "loss": 0.6902,
+      "step": 2375
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4136811160029453,
+      "learning_rate": 3.661005990181954e-05,
+      "loss": 0.6796,
+      "step": 2376
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5062173552303618,
+      "learning_rate": 3.6606832967140455e-05,
+      "loss": 0.708,
+      "step": 2377
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.46888279985618037,
+      "learning_rate": 3.660360463966634e-05,
+      "loss": 0.7101,
+      "step": 2378
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.519055115414987,
+      "learning_rate": 3.6600374919667966e-05,
+      "loss": 0.7096,
+      "step": 2379
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5034084573094453,
+      "learning_rate": 3.6597143807416186e-05,
+      "loss": 0.6608,
+      "step": 2380
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4239666817115916,
+      "learning_rate": 3.6593911303182005e-05,
+      "loss": 0.7267,
+      "step": 2381
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.44568476194639095,
+      "learning_rate": 3.659067740723652e-05,
+      "loss": 0.6832,
+      "step": 2382
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3874524314806226,
+      "learning_rate": 3.658744211985096e-05,
+      "loss": 0.6769,
+      "step": 2383
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4617933260269762,
+      "learning_rate": 3.6584205441296666e-05,
+      "loss": 0.7057,
+      "step": 2384
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5092027824635806,
+      "learning_rate": 3.658096737184508e-05,
+      "loss": 0.6986,
+      "step": 2385
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.47521856123760725,
+      "learning_rate": 3.6577727911767794e-05,
+      "loss": 0.7049,
+      "step": 2386
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4572921258237945,
+      "learning_rate": 3.6574487061336486e-05,
+      "loss": 0.7117,
+      "step": 2387
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4943166217077905,
+      "learning_rate": 3.657124482082297e-05,
+      "loss": 0.691,
+      "step": 2388
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.44878557058058416,
+      "learning_rate": 3.6568001190499156e-05,
+      "loss": 0.6894,
+      "step": 2389
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.46149269874027504,
+      "learning_rate": 3.656475617063709e-05,
+      "loss": 0.7078,
+      "step": 2390
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.41600855766851363,
+      "learning_rate": 3.656150976150894e-05,
+      "loss": 0.6787,
+      "step": 2391
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4475451981692427,
+      "learning_rate": 3.655826196338696e-05,
+      "loss": 0.665,
+      "step": 2392
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.45183359517531546,
+      "learning_rate": 3.655501277654355e-05,
+      "loss": 0.7224,
+      "step": 2393
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.47355455754932,
+      "learning_rate": 3.65517622012512e-05,
+      "loss": 0.7412,
+      "step": 2394
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.42600838434475197,
+      "learning_rate": 3.6548510237782557e-05,
+      "loss": 0.7339,
+      "step": 2395
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4807096052561072,
+      "learning_rate": 3.654525688641034e-05,
+      "loss": 0.6927,
+      "step": 2396
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.41324995725405417,
+      "learning_rate": 3.654200214740741e-05,
+      "loss": 0.6785,
+      "step": 2397
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.40998471393929725,
+      "learning_rate": 3.653874602104674e-05,
+      "loss": 0.7037,
+      "step": 2398
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4700226430237048,
+      "learning_rate": 3.653548850760142e-05,
+      "loss": 0.7079,
+      "step": 2399
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.44225517336790743,
+      "learning_rate": 3.653222960734464e-05,
+      "loss": 0.7305,
+      "step": 2400
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4387654528620805,
+      "learning_rate": 3.652896932054973e-05,
+      "loss": 0.7204,
+      "step": 2401
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.44913808845578085,
+      "learning_rate": 3.6525707647490126e-05,
+      "loss": 0.7097,
+      "step": 2402
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.40633062889714494,
+      "learning_rate": 3.652244458843938e-05,
+      "loss": 0.651,
+      "step": 2403
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4507332013944391,
+      "learning_rate": 3.651918014367116e-05,
+      "loss": 0.7672,
+      "step": 2404
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4793292419466086,
+      "learning_rate": 3.651591431345925e-05,
+      "loss": 0.7358,
+      "step": 2405
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 13.501079008449436,
+      "learning_rate": 3.6512647098077554e-05,
+      "loss": 7.5736,
+      "step": 2406
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4863564406841315,
+      "learning_rate": 3.650937849780009e-05,
+      "loss": 0.696,
+      "step": 2407
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.42943867539204683,
+      "learning_rate": 3.650610851290099e-05,
+      "loss": 0.6921,
+      "step": 2408
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5771763857921944,
+      "learning_rate": 3.6502837143654504e-05,
+      "loss": 0.709,
+      "step": 2409
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5534917432972564,
+      "learning_rate": 3.6499564390334995e-05,
+      "loss": 0.7,
+      "step": 2410
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5087274725834687,
+      "learning_rate": 3.6496290253216946e-05,
+      "loss": 0.7236,
+      "step": 2411
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5823002598665945,
+      "learning_rate": 3.649301473257495e-05,
+      "loss": 0.807,
+      "step": 2412
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.262118446052013,
+      "learning_rate": 3.648973782868374e-05,
+      "loss": 0.6812,
+      "step": 2413
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5068282278721904,
+      "learning_rate": 3.648645954181812e-05,
+      "loss": 0.716,
+      "step": 2414
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.544929493960707,
+      "learning_rate": 3.648317987225306e-05,
+      "loss": 0.6721,
+      "step": 2415
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 4.543267295718929,
+      "learning_rate": 3.647989882026361e-05,
+      "loss": 0.6879,
+      "step": 2416
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.49096619692446974,
+      "learning_rate": 3.647661638612494e-05,
+      "loss": 0.7052,
+      "step": 2417
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.501642849445768,
+      "learning_rate": 3.647333257011236e-05,
+      "loss": 0.7589,
+      "step": 2418
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4459185883456032,
+      "learning_rate": 3.647004737250127e-05,
+      "loss": 0.6828,
+      "step": 2419
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4756127088180957,
+      "learning_rate": 3.646676079356719e-05,
+      "loss": 0.6829,
+      "step": 2420
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 12.258052411356468,
+      "learning_rate": 3.646347283358578e-05,
+      "loss": 7.647,
+      "step": 2421
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5486285560160028,
+      "learning_rate": 3.646018349283278e-05,
+      "loss": 0.7423,
+      "step": 2422
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6543434697789823,
+      "learning_rate": 3.6456892771584075e-05,
+      "loss": 0.7173,
+      "step": 2423
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0141064518488965,
+      "learning_rate": 3.645360067011564e-05,
+      "loss": 0.7451,
+      "step": 2424
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9988253563049795,
+      "learning_rate": 3.645030718870359e-05,
+      "loss": 0.704,
+      "step": 2425
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.5977029490164694,
+      "learning_rate": 3.6447012327624156e-05,
+      "loss": 0.6916,
+      "step": 2426
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.405006465361147,
+      "learning_rate": 3.644371608715364e-05,
+      "loss": 0.6957,
+      "step": 2427
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7484438831356004,
+      "learning_rate": 3.644041846756853e-05,
+      "loss": 0.7443,
+      "step": 2428
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.21525785999206,
+      "learning_rate": 3.6437119469145375e-05,
+      "loss": 0.7291,
+      "step": 2429
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5957561759515096,
+      "learning_rate": 3.643381909216085e-05,
+      "loss": 0.7111,
+      "step": 2430
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5026422768016452,
+      "learning_rate": 3.6430517336891786e-05,
+      "loss": 0.7104,
+      "step": 2431
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.38627697145916523,
+      "learning_rate": 3.642721420361506e-05,
+      "loss": 0.694,
+      "step": 2432
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5258317712297531,
+      "learning_rate": 3.642390969260772e-05,
+      "loss": 0.7585,
+      "step": 2433
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4535645625383879,
+      "learning_rate": 3.642060380414691e-05,
+      "loss": 0.7215,
+      "step": 2434
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4842313814793173,
+      "learning_rate": 3.6417296538509894e-05,
+      "loss": 0.694,
+      "step": 2435
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.49669198420466615,
+      "learning_rate": 3.641398789597404e-05,
+      "loss": 0.6834,
+      "step": 2436
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4472064744370465,
+      "learning_rate": 3.641067787681685e-05,
+      "loss": 0.6912,
+      "step": 2437
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.45980650586898647,
+      "learning_rate": 3.6407366481315915e-05,
+      "loss": 0.7379,
+      "step": 2438
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4307870131473356,
+      "learning_rate": 3.6404053709748975e-05,
+      "loss": 0.6984,
+      "step": 2439
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.40978254702982597,
+      "learning_rate": 3.6400739562393867e-05,
+      "loss": 0.6605,
+      "step": 2440
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.470335203933101,
+      "learning_rate": 3.639742403952853e-05,
+      "loss": 0.7209,
+      "step": 2441
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4020780360447217,
+      "learning_rate": 3.6394107141431046e-05,
+      "loss": 0.6955,
+      "step": 2442
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4261836678624217,
+      "learning_rate": 3.639078886837959e-05,
+      "loss": 0.7149,
+      "step": 2443
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4479169890742894,
+      "learning_rate": 3.638746922065248e-05,
+      "loss": 0.7369,
+      "step": 2444
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.460643922072013,
+      "learning_rate": 3.6384148198528105e-05,
+      "loss": 0.6981,
+      "step": 2445
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.43755627144633324,
+      "learning_rate": 3.638082580228501e-05,
+      "loss": 0.7343,
+      "step": 2446
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4586046433057071,
+      "learning_rate": 3.6377502032201835e-05,
+      "loss": 0.7063,
+      "step": 2447
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.417281013275411,
+      "learning_rate": 3.637417688855735e-05,
+      "loss": 0.7213,
+      "step": 2448
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.46046176763138025,
+      "learning_rate": 3.6370850371630424e-05,
+      "loss": 0.7391,
+      "step": 2449
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.42938337426387096,
+      "learning_rate": 3.636752248170004e-05,
+      "loss": 0.7166,
+      "step": 2450
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.37644948992610067,
+      "learning_rate": 3.636419321904532e-05,
+      "loss": 0.6743,
+      "step": 2451
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4042457806245235,
+      "learning_rate": 3.636086258394548e-05,
+      "loss": 0.7002,
+      "step": 2452
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4202449410795264,
+      "learning_rate": 3.6357530576679845e-05,
+      "loss": 0.6471,
+      "step": 2453
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4089677648053497,
+      "learning_rate": 3.6354197197527884e-05,
+      "loss": 0.7032,
+      "step": 2454
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4336255435601836,
+      "learning_rate": 3.6350862446769155e-05,
+      "loss": 0.6966,
+      "step": 2455
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 12.253929025815527,
+      "learning_rate": 3.6347526324683336e-05,
+      "loss": 7.5858,
+      "step": 2456
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.44185621497742045,
+      "learning_rate": 3.6344188831550226e-05,
+      "loss": 0.7151,
+      "step": 2457
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.44977463147129604,
+      "learning_rate": 3.634084996764974e-05,
+      "loss": 0.7242,
+      "step": 2458
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.44250670425333993,
+      "learning_rate": 3.6337509733261894e-05,
+      "loss": 0.7218,
+      "step": 2459
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.0087041666099434,
+      "learning_rate": 3.633416812866685e-05,
+      "loss": 7.5111,
+      "step": 2460
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5042550680012232,
+      "learning_rate": 3.633082515414484e-05,
+      "loss": 0.7119,
+      "step": 2461
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.49499914231082753,
+      "learning_rate": 3.632748080997625e-05,
+      "loss": 0.7279,
+      "step": 2462
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.42989703456884476,
+      "learning_rate": 3.632413509644157e-05,
+      "loss": 0.6952,
+      "step": 2463
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.41446708182993697,
+      "learning_rate": 3.632078801382139e-05,
+      "loss": 0.683,
+      "step": 2464
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4739546638500601,
+      "learning_rate": 3.631743956239643e-05,
+      "loss": 0.6777,
+      "step": 2465
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5469259208075425,
+      "learning_rate": 3.631408974244752e-05,
+      "loss": 0.729,
+      "step": 2466
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.43086693141983934,
+      "learning_rate": 3.6310738554255604e-05,
+      "loss": 0.7217,
+      "step": 2467
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6058426725467373,
+      "learning_rate": 3.6307385998101744e-05,
+      "loss": 0.7176,
+      "step": 2468
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.41983700849230415,
+      "learning_rate": 3.630403207426712e-05,
+      "loss": 0.7148,
+      "step": 2469
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.38549598065116875,
+      "learning_rate": 3.630067678303301e-05,
+      "loss": 0.6991,
+      "step": 2470
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4253121956373694,
+      "learning_rate": 3.629732012468083e-05,
+      "loss": 0.6937,
+      "step": 2471
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3950200772924451,
+      "learning_rate": 3.629396209949208e-05,
+      "loss": 0.698,
+      "step": 2472
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.43761353948041987,
+      "learning_rate": 3.629060270774842e-05,
+      "loss": 0.702,
+      "step": 2473
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9131939059155973,
+      "learning_rate": 3.6287241949731585e-05,
+      "loss": 7.4475,
+      "step": 2474
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.42352131812601745,
+      "learning_rate": 3.628387982572343e-05,
+      "loss": 0.685,
+      "step": 2475
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4550669592443257,
+      "learning_rate": 3.6280516336005945e-05,
+      "loss": 0.6663,
+      "step": 2476
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.48412967665339535,
+      "learning_rate": 3.627715148086122e-05,
+      "loss": 0.7311,
+      "step": 2477
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.45394574459469317,
+      "learning_rate": 3.6273785260571455e-05,
+      "loss": 0.6936,
+      "step": 2478
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.42403972788715255,
+      "learning_rate": 3.6270417675418974e-05,
+      "loss": 0.698,
+      "step": 2479
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3939963208419623,
+      "learning_rate": 3.626704872568621e-05,
+      "loss": 0.7106,
+      "step": 2480
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1240361565188004,
+      "learning_rate": 3.626367841165571e-05,
+      "loss": 7.4244,
+      "step": 2481
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.48077100130388595,
+      "learning_rate": 3.626030673361015e-05,
+      "loss": 0.6921,
+      "step": 2482
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4305994145488111,
+      "learning_rate": 3.625693369183231e-05,
+      "loss": 0.6999,
+      "step": 2483
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.41803061658718677,
+      "learning_rate": 3.625355928660506e-05,
+      "loss": 0.6576,
+      "step": 2484
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.49558009611649617,
+      "learning_rate": 3.625018351821143e-05,
+      "loss": 0.7139,
+      "step": 2485
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.41628724432680764,
+      "learning_rate": 3.624680638693453e-05,
+      "loss": 0.6679,
+      "step": 2486
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4797914035395759,
+      "learning_rate": 3.62434278930576e-05,
+      "loss": 0.7388,
+      "step": 2487
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4078646731282426,
+      "learning_rate": 3.6240048036863995e-05,
+      "loss": 0.6973,
+      "step": 2488
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.35093153097630136,
+      "learning_rate": 3.623666681863717e-05,
+      "loss": 0.6407,
+      "step": 2489
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4418125601838795,
+      "learning_rate": 3.623328423866071e-05,
+      "loss": 0.7196,
+      "step": 2490
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.45622321398683585,
+      "learning_rate": 3.62299002972183e-05,
+      "loss": 0.7324,
+      "step": 2491
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.42816005509712945,
+      "learning_rate": 3.622651499459376e-05,
+      "loss": 0.7221,
+      "step": 2492
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3869611856947954,
+      "learning_rate": 3.6223128331071e-05,
+      "loss": 0.6568,
+      "step": 2493
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4279108657228218,
+      "learning_rate": 3.621974030693406e-05,
+      "loss": 0.7382,
+      "step": 2494
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4518801525763633,
+      "learning_rate": 3.621635092246709e-05,
+      "loss": 0.7506,
+      "step": 2495
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3954731176152587,
+      "learning_rate": 3.621296017795435e-05,
+      "loss": 0.7047,
+      "step": 2496
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.45608653781752706,
+      "learning_rate": 3.6209568073680224e-05,
+      "loss": 0.7052,
+      "step": 2497
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.45427925378393874,
+      "learning_rate": 3.6206174609929196e-05,
+      "loss": 0.715,
+      "step": 2498
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4582776994679536,
+      "learning_rate": 3.6202779786985875e-05,
+      "loss": 0.7347,
+      "step": 2499
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0066154509288383,
+      "learning_rate": 3.6199383605134986e-05,
+      "loss": 7.3808,
+      "step": 2500
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4341353360394312,
+      "learning_rate": 3.6195986064661356e-05,
+      "loss": 0.7368,
+      "step": 2501
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4482238694485038,
+      "learning_rate": 3.6192587165849926e-05,
+      "loss": 0.6946,
+      "step": 2502
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.46744264153010173,
+      "learning_rate": 3.6189186908985775e-05,
+      "loss": 0.7392,
+      "step": 2503
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.519107060078783,
+      "learning_rate": 3.6185785294354074e-05,
+      "loss": 0.7517,
+      "step": 2504
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.452275929052275,
+      "learning_rate": 3.61823823222401e-05,
+      "loss": 0.7209,
+      "step": 2505
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4921459003889346,
+      "learning_rate": 3.617897799292927e-05,
+      "loss": 0.7181,
+      "step": 2506
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4076337971260293,
+      "learning_rate": 3.617557230670709e-05,
+      "loss": 0.6707,
+      "step": 2507
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4680036162650948,
+      "learning_rate": 3.617216526385919e-05,
+      "loss": 0.7209,
+      "step": 2508
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5057690581003238,
+      "learning_rate": 3.6168756864671336e-05,
+      "loss": 0.6815,
+      "step": 2509
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.48842711579913617,
+      "learning_rate": 3.616534710942936e-05,
+      "loss": 0.6803,
+      "step": 2510
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4434862640385631,
+      "learning_rate": 3.616193599841924e-05,
+      "loss": 0.6957,
+      "step": 2511
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4184351328175589,
+      "learning_rate": 3.6158523531927074e-05,
+      "loss": 0.7538,
+      "step": 2512
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.417242504251461,
+      "learning_rate": 3.615510971023906e-05,
+      "loss": 0.7071,
+      "step": 2513
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.4151831015364562,
+      "learning_rate": 3.6151694533641496e-05,
+      "loss": 0.6893,
+      "step": 2514
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.43638166327828665,
+      "learning_rate": 3.614827800242082e-05,
+      "loss": 0.6739,
+      "step": 2515
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.42829679102458906,
+      "learning_rate": 3.614486011686357e-05,
+      "loss": 0.7183,
+      "step": 2516
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.45791823056301123,
+      "learning_rate": 3.61414408772564e-05,
+      "loss": 0.6554,
+      "step": 2517
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.45865187809232666,
+      "learning_rate": 3.6138020283886074e-05,
+      "loss": 0.7154,
+      "step": 2518
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3938815341428278,
+      "learning_rate": 3.6134598337039484e-05,
+      "loss": 0.6884,
+      "step": 2519
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4475765519414152,
+      "learning_rate": 3.613117503700361e-05,
+      "loss": 0.7325,
+      "step": 2520
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4160998664432251,
+      "learning_rate": 3.6127750384065575e-05,
+      "loss": 0.675,
+      "step": 2521
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4209526839474005,
+      "learning_rate": 3.6124324378512585e-05,
+      "loss": 0.6541,
+      "step": 2522
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.40394709526341294,
+      "learning_rate": 3.612089702063199e-05,
+      "loss": 0.6641,
+      "step": 2523
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4334859697538591,
+      "learning_rate": 3.611746831071123e-05,
+      "loss": 0.707,
+      "step": 2524
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41796520452577984,
+      "learning_rate": 3.6114038249037855e-05,
+      "loss": 0.7231,
+      "step": 2525
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3796110173294564,
+      "learning_rate": 3.611060683589957e-05,
+      "loss": 0.702,
+      "step": 2526
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.43176618800510036,
+      "learning_rate": 3.610717407158413e-05,
+      "loss": 0.677,
+      "step": 2527
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3824214833543111,
+      "learning_rate": 3.6103739956379464e-05,
+      "loss": 0.7052,
+      "step": 2528
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.40689816018803665,
+      "learning_rate": 3.610030449057358e-05,
+      "loss": 0.7534,
+      "step": 2529
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.39034346606016895,
+      "learning_rate": 3.609686767445459e-05,
+      "loss": 0.6372,
+      "step": 2530
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3913453213011736,
+      "learning_rate": 3.609342950831076e-05,
+      "loss": 0.6493,
+      "step": 2531
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.42529726083705666,
+      "learning_rate": 3.608998999243043e-05,
+      "loss": 0.7056,
+      "step": 2532
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.446842838820114,
+      "learning_rate": 3.6086549127102066e-05,
+      "loss": 0.6949,
+      "step": 2533
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9354091994600578,
+      "learning_rate": 3.608310691261426e-05,
+      "loss": 7.2329,
+      "step": 2534
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4534403463676132,
+      "learning_rate": 3.60796633492557e-05,
+      "loss": 0.7079,
+      "step": 2535
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.42755903660869804,
+      "learning_rate": 3.6076218437315184e-05,
+      "loss": 0.6951,
+      "step": 2536
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.43138976850408456,
+      "learning_rate": 3.607277217708165e-05,
+      "loss": 0.7365,
+      "step": 2537
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3987676211086679,
+      "learning_rate": 3.6069324568844135e-05,
+      "loss": 0.6752,
+      "step": 2538
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.42832290543231843,
+      "learning_rate": 3.6065875612891765e-05,
+      "loss": 0.7094,
+      "step": 2539
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4695301579377607,
+      "learning_rate": 3.606242530951381e-05,
+      "loss": 0.7087,
+      "step": 2540
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.43120122700456553,
+      "learning_rate": 3.6058973658999644e-05,
+      "loss": 0.7345,
+      "step": 2541
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.44923048853762443,
+      "learning_rate": 3.605552066163875e-05,
+      "loss": 0.7231,
+      "step": 2542
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.43238091485784685,
+      "learning_rate": 3.605206631772073e-05,
+      "loss": 0.7248,
+      "step": 2543
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41961981391120085,
+      "learning_rate": 3.6048610627535296e-05,
+      "loss": 0.7232,
+      "step": 2544
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3890946429416999,
+      "learning_rate": 3.6045153591372264e-05,
+      "loss": 0.6665,
+      "step": 2545
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3996105464507965,
+      "learning_rate": 3.6041695209521586e-05,
+      "loss": 0.7164,
+      "step": 2546
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7542864673301218,
+      "learning_rate": 3.60382354822733e-05,
+      "loss": 7.3962,
+      "step": 2547
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4299071948222621,
+      "learning_rate": 3.603477440991757e-05,
+      "loss": 0.6884,
+      "step": 2548
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.45099606343865684,
+      "learning_rate": 3.603131199274468e-05,
+      "loss": 0.7113,
+      "step": 2549
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4392655587064761,
+      "learning_rate": 3.602784823104501e-05,
+      "loss": 0.7143,
+      "step": 2550
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.44021695992172266,
+      "learning_rate": 3.602438312510908e-05,
+      "loss": 0.6683,
+      "step": 2551
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.421407670931619,
+      "learning_rate": 3.6020916675227476e-05,
+      "loss": 0.6985,
+      "step": 2552
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4849206399267615,
+      "learning_rate": 3.6017448881690936e-05,
+      "loss": 0.7615,
+      "step": 2553
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4368265511793137,
+      "learning_rate": 3.601397974479031e-05,
+      "loss": 0.7148,
+      "step": 2554
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.40054936773320093,
+      "learning_rate": 3.6010509264816545e-05,
+      "loss": 0.6749,
+      "step": 2555
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41579715774533105,
+      "learning_rate": 3.60070374420607e-05,
+      "loss": 0.6488,
+      "step": 2556
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4318639775768571,
+      "learning_rate": 3.600356427681395e-05,
+      "loss": 0.7009,
+      "step": 2557
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.43449661940717366,
+      "learning_rate": 3.60000897693676e-05,
+      "loss": 0.6802,
+      "step": 2558
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.40902595412028636,
+      "learning_rate": 3.599661392001304e-05,
+      "loss": 0.7012,
+      "step": 2559
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.42343256766300175,
+      "learning_rate": 3.599313672904179e-05,
+      "loss": 0.7192,
+      "step": 2560
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4135175592396193,
+      "learning_rate": 3.5989658196745476e-05,
+      "loss": 0.7016,
+      "step": 2561
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.43525851878478267,
+      "learning_rate": 3.5986178323415845e-05,
+      "loss": 0.7219,
+      "step": 2562
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3738872778935052,
+      "learning_rate": 3.5982697109344746e-05,
+      "loss": 0.6947,
+      "step": 2563
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.49774070522990627,
+      "learning_rate": 3.597921455482413e-05,
+      "loss": 0.7321,
+      "step": 2564
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4268693408723472,
+      "learning_rate": 3.5975730660146095e-05,
+      "loss": 0.7011,
+      "step": 2565
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5531257074062149,
+      "learning_rate": 3.597224542560283e-05,
+      "loss": 7.3376,
+      "step": 2566
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4387665442180965,
+      "learning_rate": 3.596875885148662e-05,
+      "loss": 0.7154,
+      "step": 2567
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3522198068886605,
+      "learning_rate": 3.59652709380899e-05,
+      "loss": 0.6287,
+      "step": 2568
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.392517486149683,
+      "learning_rate": 3.596178168570518e-05,
+      "loss": 0.6799,
+      "step": 2569
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4189644277475421,
+      "learning_rate": 3.595829109462511e-05,
+      "loss": 0.7214,
+      "step": 2570
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4358439137023056,
+      "learning_rate": 3.595479916514244e-05,
+      "loss": 0.736,
+      "step": 2571
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4774874229341309,
+      "learning_rate": 3.595130589755003e-05,
+      "loss": 0.7528,
+      "step": 2572
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4271812448651369,
+      "learning_rate": 3.594781129214086e-05,
+      "loss": 0.7434,
+      "step": 2573
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.44248340759069293,
+      "learning_rate": 3.5944315349208016e-05,
+      "loss": 0.7198,
+      "step": 2574
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.37847102664488863,
+      "learning_rate": 3.59408180690447e-05,
+      "loss": 0.709,
+      "step": 2575
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.46060187226457566,
+      "learning_rate": 3.5937319451944225e-05,
+      "loss": 0.6998,
+      "step": 2576
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3888604731215113,
+      "learning_rate": 3.5933819498200016e-05,
+      "loss": 0.6886,
+      "step": 2577
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.42127525605444843,
+      "learning_rate": 3.593031820810561e-05,
+      "loss": 0.6828,
+      "step": 2578
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4507630956250111,
+      "learning_rate": 3.592681558195465e-05,
+      "loss": 0.7517,
+      "step": 2579
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8920065215998622,
+      "learning_rate": 3.59233116200409e-05,
+      "loss": 7.5661,
+      "step": 2580
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4139698721236584,
+      "learning_rate": 3.591980632265824e-05,
+      "loss": 0.7245,
+      "step": 2581
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41802472735263246,
+      "learning_rate": 3.5916299690100643e-05,
+      "loss": 0.7072,
+      "step": 2582
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3639453002171928,
+      "learning_rate": 3.591279172266222e-05,
+      "loss": 0.6907,
+      "step": 2583
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41665358753571435,
+      "learning_rate": 3.5909282420637174e-05,
+      "loss": 0.7117,
+      "step": 2584
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.43399017061615175,
+      "learning_rate": 3.590577178431982e-05,
+      "loss": 0.7005,
+      "step": 2585
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4461999371261507,
+      "learning_rate": 3.59022598140046e-05,
+      "loss": 0.6982,
+      "step": 2586
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3759792026827843,
+      "learning_rate": 3.589874650998606e-05,
+      "loss": 0.6943,
+      "step": 2587
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4304702379284862,
+      "learning_rate": 3.589523187255884e-05,
+      "loss": 0.7042,
+      "step": 2588
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4199190363555218,
+      "learning_rate": 3.589171590201772e-05,
+      "loss": 0.6841,
+      "step": 2589
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4587339849435593,
+      "learning_rate": 3.5888198598657585e-05,
+      "loss": 0.7093,
+      "step": 2590
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5733205397529767,
+      "learning_rate": 3.588467996277342e-05,
+      "loss": 0.6996,
+      "step": 2591
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4380639728117856,
+      "learning_rate": 3.588115999466033e-05,
+      "loss": 0.7107,
+      "step": 2592
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.39851282568845975,
+      "learning_rate": 3.587763869461353e-05,
+      "loss": 0.6936,
+      "step": 2593
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.47275784641532104,
+      "learning_rate": 3.587411606292835e-05,
+      "loss": 0.733,
+      "step": 2594
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.40470291030338046,
+      "learning_rate": 3.5870592099900224e-05,
+      "loss": 0.6941,
+      "step": 2595
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6021115691453937,
+      "learning_rate": 3.586706680582471e-05,
+      "loss": 0.7403,
+      "step": 2596
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.44645290631816004,
+      "learning_rate": 3.586354018099746e-05,
+      "loss": 0.7333,
+      "step": 2597
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.45577662528366736,
+      "learning_rate": 3.5860012225714255e-05,
+      "loss": 0.7025,
+      "step": 2598
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4586102232409273,
+      "learning_rate": 3.585648294027098e-05,
+      "loss": 0.6976,
+      "step": 2599
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.469169334309171,
+      "learning_rate": 3.5852952324963634e-05,
+      "loss": 0.6779,
+      "step": 2600
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.421026218474303,
+      "learning_rate": 3.584942038008831e-05,
+      "loss": 0.7059,
+      "step": 2601
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.44366357515279,
+      "learning_rate": 3.584588710594125e-05,
+      "loss": 0.7001,
+      "step": 2602
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4766632625265187,
+      "learning_rate": 3.584235250281878e-05,
+      "loss": 0.693,
+      "step": 2603
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.47872817845428534,
+      "learning_rate": 3.5838816571017324e-05,
+      "loss": 0.7143,
+      "step": 2604
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4059438673716114,
+      "learning_rate": 3.583527931083346e-05,
+      "loss": 0.6593,
+      "step": 2605
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4171611215637362,
+      "learning_rate": 3.583174072256385e-05,
+      "loss": 0.7055,
+      "step": 2606
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4596490836589721,
+      "learning_rate": 3.582820080650526e-05,
+      "loss": 0.7467,
+      "step": 2607
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.38504340532128045,
+      "learning_rate": 3.5824659562954574e-05,
+      "loss": 0.7014,
+      "step": 2608
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5312533396909893,
+      "learning_rate": 3.582111699220881e-05,
+      "loss": 0.7047,
+      "step": 2609
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4085407887052443,
+      "learning_rate": 3.5817573094565076e-05,
+      "loss": 0.7163,
+      "step": 2610
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41764583113722614,
+      "learning_rate": 3.581402787032058e-05,
+      "loss": 0.6965,
+      "step": 2611
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41254880638686675,
+      "learning_rate": 3.581048131977267e-05,
+      "loss": 0.6791,
+      "step": 2612
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4131220210264849,
+      "learning_rate": 3.5806933443218787e-05,
+      "loss": 0.7363,
+      "step": 2613
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4573236526333459,
+      "learning_rate": 3.580338424095648e-05,
+      "loss": 0.6712,
+      "step": 2614
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4288225785234453,
+      "learning_rate": 3.579983371328343e-05,
+      "loss": 0.7102,
+      "step": 2615
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.44415839292974096,
+      "learning_rate": 3.5796281860497406e-05,
+      "loss": 0.7511,
+      "step": 2616
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.451974536552726,
+      "learning_rate": 3.5792728682896296e-05,
+      "loss": 0.6568,
+      "step": 2617
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4103263950469182,
+      "learning_rate": 3.57891741807781e-05,
+      "loss": 0.6888,
+      "step": 2618
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4430163633205285,
+      "learning_rate": 3.5785618354440945e-05,
+      "loss": 0.7199,
+      "step": 2619
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4984505665715321,
+      "learning_rate": 3.5782061204183036e-05,
+      "loss": 0.7317,
+      "step": 2620
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4467117019171592,
+      "learning_rate": 3.577850273030272e-05,
+      "loss": 0.74,
+      "step": 2621
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4619632846519289,
+      "learning_rate": 3.577494293309843e-05,
+      "loss": 0.7335,
+      "step": 2622
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3918574367275226,
+      "learning_rate": 3.577138181286873e-05,
+      "loss": 0.6816,
+      "step": 2623
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4296326722443219,
+      "learning_rate": 3.5767819369912284e-05,
+      "loss": 0.7069,
+      "step": 2624
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.48025324020357096,
+      "learning_rate": 3.576425560452787e-05,
+      "loss": 0.7304,
+      "step": 2625
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.36070208211303045,
+      "learning_rate": 3.576069051701437e-05,
+      "loss": 0.6607,
+      "step": 2626
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4307398846224419,
+      "learning_rate": 3.5757124107670806e-05,
+      "loss": 0.7022,
+      "step": 2627
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4203327121582497,
+      "learning_rate": 3.575355637679626e-05,
+      "loss": 0.6774,
+      "step": 2628
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.38581010611681293,
+      "learning_rate": 3.574998732468997e-05,
+      "loss": 0.6621,
+      "step": 2629
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.37876875114929437,
+      "learning_rate": 3.5746416951651265e-05,
+      "loss": 0.6718,
+      "step": 2630
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.43752769445628675,
+      "learning_rate": 3.574284525797958e-05,
+      "loss": 0.6935,
+      "step": 2631
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.44820579606218647,
+      "learning_rate": 3.5739272243974486e-05,
+      "loss": 0.6917,
+      "step": 2632
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.42589765875686814,
+      "learning_rate": 3.5735697909935636e-05,
+      "loss": 0.7069,
+      "step": 2633
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.40897188034628995,
+      "learning_rate": 3.57321222561628e-05,
+      "loss": 0.6368,
+      "step": 2634
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4044782912939606,
+      "learning_rate": 3.5728545282955874e-05,
+      "loss": 0.6832,
+      "step": 2635
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.38539698765297803,
+      "learning_rate": 3.5724966990614845e-05,
+      "loss": 0.6761,
+      "step": 2636
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.40333075326252005,
+      "learning_rate": 3.572138737943983e-05,
+      "loss": 0.6818,
+      "step": 2637
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.43835116251298745,
+      "learning_rate": 3.571780644973105e-05,
+      "loss": 0.7408,
+      "step": 2638
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4392108907871214,
+      "learning_rate": 3.571422420178881e-05,
+      "loss": 0.7287,
+      "step": 2639
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4280043127552962,
+      "learning_rate": 3.571064063591357e-05,
+      "loss": 0.6953,
+      "step": 2640
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.44385232160996585,
+      "learning_rate": 3.570705575240588e-05,
+      "loss": 0.6917,
+      "step": 2641
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4012220946555632,
+      "learning_rate": 3.5703469551566384e-05,
+      "loss": 0.6867,
+      "step": 2642
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.46448273739259305,
+      "learning_rate": 3.569988203369587e-05,
+      "loss": 0.6785,
+      "step": 2643
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4036889791358587,
+      "learning_rate": 3.56962931990952e-05,
+      "loss": 0.6847,
+      "step": 2644
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.44713143570576513,
+      "learning_rate": 3.569270304806539e-05,
+      "loss": 0.6814,
+      "step": 2645
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5049107693953069,
+      "learning_rate": 3.568911158090752e-05,
+      "loss": 0.7222,
+      "step": 2646
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4350617327219532,
+      "learning_rate": 3.568551879792281e-05,
+      "loss": 0.7231,
+      "step": 2647
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.40311339403602364,
+      "learning_rate": 3.5681924699412574e-05,
+      "loss": 0.6797,
+      "step": 2648
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.46520637694216116,
+      "learning_rate": 3.5678329285678266e-05,
+      "loss": 0.695,
+      "step": 2649
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3740711364233098,
+      "learning_rate": 3.56747325570214e-05,
+      "loss": 0.6985,
+      "step": 2650
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4092490772696387,
+      "learning_rate": 3.5671134513743654e-05,
+      "loss": 0.6835,
+      "step": 2651
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4123912555549862,
+      "learning_rate": 3.566753515614679e-05,
+      "loss": 0.6688,
+      "step": 2652
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.42878124964871334,
+      "learning_rate": 3.5663934484532664e-05,
+      "loss": 0.6965,
+      "step": 2653
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4165702394588525,
+      "learning_rate": 3.5660332499203274e-05,
+      "loss": 0.7159,
+      "step": 2654
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.47529421465233856,
+      "learning_rate": 3.565672920046071e-05,
+      "loss": 0.6949,
+      "step": 2655
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4880061472984907,
+      "learning_rate": 3.565312458860717e-05,
+      "loss": 0.7136,
+      "step": 2656
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4078452373090537,
+      "learning_rate": 3.564951866394498e-05,
+      "loss": 0.7258,
+      "step": 2657
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4260195644970103,
+      "learning_rate": 3.564591142677656e-05,
+      "loss": 0.6708,
+      "step": 2658
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4122876808992176,
+      "learning_rate": 3.564230287740445e-05,
+      "loss": 0.6842,
+      "step": 2659
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4386934619455345,
+      "learning_rate": 3.563869301613127e-05,
+      "loss": 0.7052,
+      "step": 2660
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.44394953706418916,
+      "learning_rate": 3.5635081843259803e-05,
+      "loss": 0.7188,
+      "step": 2661
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.41202175474910974,
+      "learning_rate": 3.563146935909291e-05,
+      "loss": 0.6821,
+      "step": 2662
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3639145864619272,
+      "learning_rate": 3.562785556393356e-05,
+      "loss": 0.6445,
+      "step": 2663
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4290164590389863,
+      "learning_rate": 3.562424045808483e-05,
+      "loss": 0.676,
+      "step": 2664
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4716889242631868,
+      "learning_rate": 3.562062404184993e-05,
+      "loss": 0.7173,
+      "step": 2665
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.406222204417364,
+      "learning_rate": 3.561700631553215e-05,
+      "loss": 0.7013,
+      "step": 2666
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.41673437259107526,
+      "learning_rate": 3.5613387279434914e-05,
+      "loss": 0.7273,
+      "step": 2667
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.38998880256749924,
+      "learning_rate": 3.5609766933861745e-05,
+      "loss": 0.6889,
+      "step": 2668
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5362337626875868,
+      "learning_rate": 3.560614527911627e-05,
+      "loss": 0.6928,
+      "step": 2669
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.39469236148215686,
+      "learning_rate": 3.560252231550224e-05,
+      "loss": 0.6719,
+      "step": 2670
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.39453215730779745,
+      "learning_rate": 3.55988980433235e-05,
+      "loss": 0.6715,
+      "step": 2671
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3800184898621566,
+      "learning_rate": 3.5595272462884035e-05,
+      "loss": 0.6428,
+      "step": 2672
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4678294946409289,
+      "learning_rate": 3.559164557448789e-05,
+      "loss": 0.707,
+      "step": 2673
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3931742947690449,
+      "learning_rate": 3.558801737843926e-05,
+      "loss": 0.7241,
+      "step": 2674
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.45239659669529225,
+      "learning_rate": 3.558438787504245e-05,
+      "loss": 0.7232,
+      "step": 2675
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.48706904684794544,
+      "learning_rate": 3.5580757064601847e-05,
+      "loss": 0.7449,
+      "step": 2676
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4014685072021696,
+      "learning_rate": 3.557712494742196e-05,
+      "loss": 0.6976,
+      "step": 2677
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.39163364058603795,
+      "learning_rate": 3.5573491523807414e-05,
+      "loss": 0.6712,
+      "step": 2678
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.42162626116125285,
+      "learning_rate": 3.556985679406295e-05,
+      "loss": 0.6884,
+      "step": 2679
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4025786852719168,
+      "learning_rate": 3.556622075849339e-05,
+      "loss": 0.6389,
+      "step": 2680
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4940243918496426,
+      "learning_rate": 3.5562583417403705e-05,
+      "loss": 0.7205,
+      "step": 2681
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.47564387704565725,
+      "learning_rate": 3.555894477109894e-05,
+      "loss": 0.7255,
+      "step": 2682
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4532343959224192,
+      "learning_rate": 3.5555304819884267e-05,
+      "loss": 0.6864,
+      "step": 2683
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3825087880907663,
+      "learning_rate": 3.555166356406496e-05,
+      "loss": 0.726,
+      "step": 2684
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.41506786434837206,
+      "learning_rate": 3.554802100394642e-05,
+      "loss": 0.6991,
+      "step": 2685
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9717295231965821,
+      "learning_rate": 3.554437713983413e-05,
+      "loss": 7.5083,
+      "step": 2686
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3754779566228985,
+      "learning_rate": 3.554073197203371e-05,
+      "loss": 0.6783,
+      "step": 2687
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4171746492631066,
+      "learning_rate": 3.5537085500850854e-05,
+      "loss": 0.6916,
+      "step": 2688
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4638701766719475,
+      "learning_rate": 3.553343772659141e-05,
+      "loss": 0.6957,
+      "step": 2689
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4403769217244899,
+      "learning_rate": 3.5529788649561304e-05,
+      "loss": 0.7437,
+      "step": 2690
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.46614255838962,
+      "learning_rate": 3.552613827006658e-05,
+      "loss": 0.6893,
+      "step": 2691
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4152787050310641,
+      "learning_rate": 3.5522486588413384e-05,
+      "loss": 0.6782,
+      "step": 2692
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4266237360284651,
+      "learning_rate": 3.5518833604907996e-05,
+      "loss": 0.6807,
+      "step": 2693
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.43698936981153275,
+      "learning_rate": 3.551517931985676e-05,
+      "loss": 0.713,
+      "step": 2694
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5252968309732735,
+      "learning_rate": 3.551152373356618e-05,
+      "loss": 0.7105,
+      "step": 2695
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5205913046480452,
+      "learning_rate": 3.5507866846342845e-05,
+      "loss": 0.7413,
+      "step": 2696
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.433369124892047,
+      "learning_rate": 3.5504208658493435e-05,
+      "loss": 0.7136,
+      "step": 2697
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3812497995600677,
+      "learning_rate": 3.550054917032477e-05,
+      "loss": 0.6757,
+      "step": 2698
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.42719803090135816,
+      "learning_rate": 3.549688838214377e-05,
+      "loss": 0.7014,
+      "step": 2699
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.37089807438056854,
+      "learning_rate": 3.5493226294257456e-05,
+      "loss": 0.6936,
+      "step": 2700
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.024280816422171,
+      "learning_rate": 3.5489562906972965e-05,
+      "loss": 7.3981,
+      "step": 2701
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4075157199659697,
+      "learning_rate": 3.5485898220597534e-05,
+      "loss": 0.6576,
+      "step": 2702
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.40555382798200745,
+      "learning_rate": 3.5482232235438525e-05,
+      "loss": 0.6772,
+      "step": 2703
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4576792902980343,
+      "learning_rate": 3.54785649518034e-05,
+      "loss": 0.7501,
+      "step": 2704
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3565607978646308,
+      "learning_rate": 3.5474896369999716e-05,
+      "loss": 0.6524,
+      "step": 2705
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.41306347880851796,
+      "learning_rate": 3.5471226490335174e-05,
+      "loss": 0.7006,
+      "step": 2706
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.45419833486338146,
+      "learning_rate": 3.546755531311754e-05,
+      "loss": 0.7052,
+      "step": 2707
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.39035836244884387,
+      "learning_rate": 3.546388283865473e-05,
+      "loss": 0.6861,
+      "step": 2708
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.43291527927452184,
+      "learning_rate": 3.546020906725474e-05,
+      "loss": 0.6907,
+      "step": 2709
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.39540040227550893,
+      "learning_rate": 3.545653399922569e-05,
+      "loss": 0.682,
+      "step": 2710
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4073312694860116,
+      "learning_rate": 3.54528576348758e-05,
+      "loss": 0.6961,
+      "step": 2711
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.41149359307076394,
+      "learning_rate": 3.544917997451339e-05,
+      "loss": 0.6916,
+      "step": 2712
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.40563910027799455,
+      "learning_rate": 3.544550101844693e-05,
+      "loss": 0.7142,
+      "step": 2713
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.45548003506778195,
+      "learning_rate": 3.5441820766984944e-05,
+      "loss": 0.7176,
+      "step": 2714
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.44966689868174525,
+      "learning_rate": 3.54381392204361e-05,
+      "loss": 0.6994,
+      "step": 2715
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4813040613680243,
+      "learning_rate": 3.5434456379109166e-05,
+      "loss": 0.7229,
+      "step": 2716
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.39072736132950314,
+      "learning_rate": 3.543077224331301e-05,
+      "loss": 0.6882,
+      "step": 2717
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4064829200573417,
+      "learning_rate": 3.542708681335663e-05,
+      "loss": 0.6783,
+      "step": 2718
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.43092940961702025,
+      "learning_rate": 3.54234000895491e-05,
+      "loss": 0.6918,
+      "step": 2719
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.39560941988144505,
+      "learning_rate": 3.541971207219964e-05,
+      "loss": 0.6849,
+      "step": 2720
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.36835366602823616,
+      "learning_rate": 3.541602276161755e-05,
+      "loss": 0.6822,
+      "step": 2721
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4041671997731848,
+      "learning_rate": 3.5412332158112246e-05,
+      "loss": 0.6841,
+      "step": 2722
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4257472511740162,
+      "learning_rate": 3.540864026199326e-05,
+      "loss": 0.759,
+      "step": 2723
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4093950612874814,
+      "learning_rate": 3.5404947073570225e-05,
+      "loss": 0.705,
+      "step": 2724
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.42409374291920837,
+      "learning_rate": 3.540125259315288e-05,
+      "loss": 0.723,
+      "step": 2725
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.38771623581180165,
+      "learning_rate": 3.539755682105108e-05,
+      "loss": 0.7016,
+      "step": 2726
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.46936382729106096,
+      "learning_rate": 3.539385975757479e-05,
+      "loss": 0.7023,
+      "step": 2727
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4171749361651386,
+      "learning_rate": 3.5390161403034064e-05,
+      "loss": 0.6901,
+      "step": 2728
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.6692448050737565,
+      "learning_rate": 3.53864617577391e-05,
+      "loss": 7.4232,
+      "step": 2729
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.41894210446343544,
+      "learning_rate": 3.5382760822000156e-05,
+      "loss": 0.6909,
+      "step": 2730
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4292306419844666,
+      "learning_rate": 3.5379058596127646e-05,
+      "loss": 0.6806,
+      "step": 2731
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4086198590706204,
+      "learning_rate": 3.537535508043207e-05,
+      "loss": 0.6922,
+      "step": 2732
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.41947950053551625,
+      "learning_rate": 3.5371650275224026e-05,
+      "loss": 0.6935,
+      "step": 2733
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.431892516924721,
+      "learning_rate": 3.5367944180814244e-05,
+      "loss": 0.6779,
+      "step": 2734
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4166269510383376,
+      "learning_rate": 3.536423679751353e-05,
+      "loss": 0.6512,
+      "step": 2735
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.43481003568698345,
+      "learning_rate": 3.536052812563284e-05,
+      "loss": 0.7158,
+      "step": 2736
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.44964852109208425,
+      "learning_rate": 3.535681816548321e-05,
+      "loss": 0.7156,
+      "step": 2737
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3820224256221874,
+      "learning_rate": 3.535310691737578e-05,
+      "loss": 0.6587,
+      "step": 2738
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3922546205306966,
+      "learning_rate": 3.5349394381621825e-05,
+      "loss": 0.7079,
+      "step": 2739
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.1568009821382828,
+      "learning_rate": 3.534568055853269e-05,
+      "loss": 7.3622,
+      "step": 2740
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4273944363854446,
+      "learning_rate": 3.5341965448419865e-05,
+      "loss": 0.7069,
+      "step": 2741
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4145375028454631,
+      "learning_rate": 3.5338249051594926e-05,
+      "loss": 0.7113,
+      "step": 2742
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.42207446747306115,
+      "learning_rate": 3.533453136836956e-05,
+      "loss": 0.6772,
+      "step": 2743
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5844194511941844,
+      "learning_rate": 3.5330812399055575e-05,
+      "loss": 0.7203,
+      "step": 2744
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4250090368929371,
+      "learning_rate": 3.5327092143964865e-05,
+      "loss": 0.6861,
+      "step": 2745
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3977516562101589,
+      "learning_rate": 3.532337060340944e-05,
+      "loss": 0.656,
+      "step": 2746
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3872998974981676,
+      "learning_rate": 3.531964777770143e-05,
+      "loss": 0.7072,
+      "step": 2747
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4592728962580888,
+      "learning_rate": 3.5315923667153075e-05,
+      "loss": 0.7187,
+      "step": 2748
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.47509266867631667,
+      "learning_rate": 3.5312198272076685e-05,
+      "loss": 0.7379,
+      "step": 2749
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6375190519553338,
+      "learning_rate": 3.530847159278472e-05,
+      "loss": 0.6801,
+      "step": 2750
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4278844698397615,
+      "learning_rate": 3.530474362958974e-05,
+      "loss": 0.6825,
+      "step": 2751
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.39376466079392203,
+      "learning_rate": 3.530101438280439e-05,
+      "loss": 0.6988,
+      "step": 2752
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.41159848747661865,
+      "learning_rate": 3.529728385274144e-05,
+      "loss": 0.656,
+      "step": 2753
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4394926342370403,
+      "learning_rate": 3.529355203971377e-05,
+      "loss": 0.6884,
+      "step": 2754
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4580680269169932,
+      "learning_rate": 3.528981894403436e-05,
+      "loss": 0.7178,
+      "step": 2755
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.41384353687923914,
+      "learning_rate": 3.52860845660163e-05,
+      "loss": 0.743,
+      "step": 2756
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.34920103875141684,
+      "learning_rate": 3.528234890597279e-05,
+      "loss": 0.6745,
+      "step": 2757
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.37848387844899156,
+      "learning_rate": 3.527861196421714e-05,
+      "loss": 0.6872,
+      "step": 2758
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.44036449731086347,
+      "learning_rate": 3.527487374106274e-05,
+      "loss": 0.7348,
+      "step": 2759
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.36927280850650557,
+      "learning_rate": 3.5271134236823136e-05,
+      "loss": 0.6958,
+      "step": 2760
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3843626977330459,
+      "learning_rate": 3.526739345181195e-05,
+      "loss": 0.6506,
+      "step": 2761
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1418502798108627,
+      "learning_rate": 3.5263651386342907e-05,
+      "loss": 7.3805,
+      "step": 2762
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.49488897325038816,
+      "learning_rate": 3.525990804072986e-05,
+      "loss": 0.6727,
+      "step": 2763
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3893648502049984,
+      "learning_rate": 3.525616341528675e-05,
+      "loss": 0.6688,
+      "step": 2764
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.45941666451164814,
+      "learning_rate": 3.525241751032765e-05,
+      "loss": 0.7073,
+      "step": 2765
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4116363953636379,
+      "learning_rate": 3.5248670326166704e-05,
+      "loss": 0.6682,
+      "step": 2766
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.37052372405678724,
+      "learning_rate": 3.5244921863118194e-05,
+      "loss": 0.7094,
+      "step": 2767
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4631806113202484,
+      "learning_rate": 3.52411721214965e-05,
+      "loss": 0.7364,
+      "step": 2768
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4461460583233569,
+      "learning_rate": 3.523742110161611e-05,
+      "loss": 0.6623,
+      "step": 2769
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4125713100440966,
+      "learning_rate": 3.523366880379161e-05,
+      "loss": 0.7081,
+      "step": 2770
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5392895864163849,
+      "learning_rate": 3.522991522833771e-05,
+      "loss": 0.689,
+      "step": 2771
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.43188252583201925,
+      "learning_rate": 3.522616037556921e-05,
+      "loss": 0.7301,
+      "step": 2772
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3894971709953191,
+      "learning_rate": 3.522240424580103e-05,
+      "loss": 0.6759,
+      "step": 2773
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3693928391715976,
+      "learning_rate": 3.5218646839348184e-05,
+      "loss": 0.6705,
+      "step": 2774
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4452124233708666,
+      "learning_rate": 3.521488815652581e-05,
+      "loss": 0.6734,
+      "step": 2775
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4120169230246022,
+      "learning_rate": 3.5211128197649144e-05,
+      "loss": 0.718,
+      "step": 2776
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.40401838115633815,
+      "learning_rate": 3.5207366963033525e-05,
+      "loss": 0.6759,
+      "step": 2777
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3902547157188235,
+      "learning_rate": 3.52036044529944e-05,
+      "loss": 0.6635,
+      "step": 2778
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.392703888813068,
+      "learning_rate": 3.519984066784733e-05,
+      "loss": 0.6575,
+      "step": 2779
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.40439922667028455,
+      "learning_rate": 3.519607560790799e-05,
+      "loss": 0.7011,
+      "step": 2780
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3622968498849398,
+      "learning_rate": 3.5192309273492136e-05,
+      "loss": 0.674,
+      "step": 2781
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.46967381998163826,
+      "learning_rate": 3.518854166491565e-05,
+      "loss": 0.7247,
+      "step": 2782
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.446097034145499,
+      "learning_rate": 3.518477278249452e-05,
+      "loss": 0.7361,
+      "step": 2783
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.38180877528753415,
+      "learning_rate": 3.518100262654483e-05,
+      "loss": 0.7035,
+      "step": 2784
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.372607316050817,
+      "learning_rate": 3.517723119738279e-05,
+      "loss": 0.6856,
+      "step": 2785
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.38118112049733605,
+      "learning_rate": 3.517345849532469e-05,
+      "loss": 0.7151,
+      "step": 2786
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.43979563347093026,
+      "learning_rate": 3.516968452068696e-05,
+      "loss": 0.731,
+      "step": 2787
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.365835651577427,
+      "learning_rate": 3.516590927378611e-05,
+      "loss": 0.6746,
+      "step": 2788
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3736187415897562,
+      "learning_rate": 3.516213275493876e-05,
+      "loss": 0.6474,
+      "step": 2789
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.44508549116821455,
+      "learning_rate": 3.515835496446165e-05,
+      "loss": 0.7419,
+      "step": 2790
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4478226062319774,
+      "learning_rate": 3.515457590267161e-05,
+      "loss": 0.6786,
+      "step": 2791
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.44334899217339707,
+      "learning_rate": 3.51507955698856e-05,
+      "loss": 0.7137,
+      "step": 2792
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.39075511306007216,
+      "learning_rate": 3.514701396642066e-05,
+      "loss": 0.6474,
+      "step": 2793
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4437633244017652,
+      "learning_rate": 3.5143231092593946e-05,
+      "loss": 0.6959,
+      "step": 2794
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4014370833001973,
+      "learning_rate": 3.513944694872273e-05,
+      "loss": 0.7095,
+      "step": 2795
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.34970968824201454,
+      "learning_rate": 3.51356615351244e-05,
+      "loss": 0.6954,
+      "step": 2796
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.45699685919140215,
+      "learning_rate": 3.5131874852116396e-05,
+      "loss": 0.7278,
+      "step": 2797
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.40912540088267024,
+      "learning_rate": 3.512808690001633e-05,
+      "loss": 0.6901,
+      "step": 2798
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4414476940279758,
+      "learning_rate": 3.512429767914188e-05,
+      "loss": 0.7062,
+      "step": 2799
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.8665576831832436,
+      "learning_rate": 3.512050718981085e-05,
+      "loss": 7.5988,
+      "step": 2800
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.38501066659805777,
+      "learning_rate": 3.5116715432341144e-05,
+      "loss": 0.6424,
+      "step": 2801
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.42509010224783733,
+      "learning_rate": 3.511292240705078e-05,
+      "loss": 0.6758,
+      "step": 2802
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4619254968897138,
+      "learning_rate": 3.5109128114257855e-05,
+      "loss": 0.6737,
+      "step": 2803
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4108607049407621,
+      "learning_rate": 3.510533255428061e-05,
+      "loss": 0.6871,
+      "step": 2804
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3838867257536594,
+      "learning_rate": 3.510153572743736e-05,
+      "loss": 0.6705,
+      "step": 2805
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.40108745872211954,
+      "learning_rate": 3.509773763404655e-05,
+      "loss": 0.6876,
+      "step": 2806
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.43107322728325165,
+      "learning_rate": 3.509393827442672e-05,
+      "loss": 0.7014,
+      "step": 2807
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4212285614600538,
+      "learning_rate": 3.509013764889651e-05,
+      "loss": 0.6888,
+      "step": 2808
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4565445821706989,
+      "learning_rate": 3.508633575777468e-05,
+      "loss": 0.6594,
+      "step": 2809
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.53597156684343,
+      "learning_rate": 3.508253260138009e-05,
+      "loss": 0.6728,
+      "step": 2810
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.35854833154968635,
+      "learning_rate": 3.507872818003171e-05,
+      "loss": 0.7162,
+      "step": 2811
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4286138150242452,
+      "learning_rate": 3.507492249404861e-05,
+      "loss": 0.6481,
+      "step": 2812
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4142621193135801,
+      "learning_rate": 3.507111554374996e-05,
+      "loss": 0.727,
+      "step": 2813
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4216057215574422,
+      "learning_rate": 3.506730732945506e-05,
+      "loss": 0.6856,
+      "step": 2814
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3980687913425856,
+      "learning_rate": 3.5063497851483285e-05,
+      "loss": 0.7283,
+      "step": 2815
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.47183869100327963,
+      "learning_rate": 3.505968711015413e-05,
+      "loss": 0.742,
+      "step": 2816
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.34683546277446986,
+      "learning_rate": 3.5055875105787215e-05,
+      "loss": 0.642,
+      "step": 2817
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4012605323241682,
+      "learning_rate": 3.5052061838702225e-05,
+      "loss": 0.712,
+      "step": 2818
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.46825979672594814,
+      "learning_rate": 3.5048247309218996e-05,
+      "loss": 0.7216,
+      "step": 2819
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 4.2024817710078155,
+      "learning_rate": 3.504443151765745e-05,
+      "loss": 7.56,
+      "step": 2820
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4301577435724903,
+      "learning_rate": 3.5040614464337576e-05,
+      "loss": 0.7267,
+      "step": 2821
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4211534017430849,
+      "learning_rate": 3.503679614957955e-05,
+      "loss": 0.6933,
+      "step": 2822
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.41532193272506257,
+      "learning_rate": 3.503297657370359e-05,
+      "loss": 0.6933,
+      "step": 2823
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3642867830130097,
+      "learning_rate": 3.5029155737030036e-05,
+      "loss": 0.6901,
+      "step": 2824
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4340774322646227,
+      "learning_rate": 3.502533363987934e-05,
+      "loss": 0.7004,
+      "step": 2825
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.42984052554302554,
+      "learning_rate": 3.502151028257206e-05,
+      "loss": 0.6627,
+      "step": 2826
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.41385190539597466,
+      "learning_rate": 3.5017685665428854e-05,
+      "loss": 0.6781,
+      "step": 2827
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.39025471702129017,
+      "learning_rate": 3.501385978877048e-05,
+      "loss": 0.7149,
+      "step": 2828
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.44094013167578433,
+      "learning_rate": 3.5010032652917823e-05,
+      "loss": 0.7249,
+      "step": 2829
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.42129302127619844,
+      "learning_rate": 3.500620425819186e-05,
+      "loss": 0.7242,
+      "step": 2830
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.397055495216645,
+      "learning_rate": 3.5002374604913666e-05,
+      "loss": 0.6927,
+      "step": 2831
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4046193354397218,
+      "learning_rate": 3.499854369340443e-05,
+      "loss": 0.6805,
+      "step": 2832
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.41047686214959644,
+      "learning_rate": 3.499471152398545e-05,
+      "loss": 0.6784,
+      "step": 2833
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.42469698566673747,
+      "learning_rate": 3.499087809697813e-05,
+      "loss": 0.7237,
+      "step": 2834
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4514219404273984,
+      "learning_rate": 3.4987043412703956e-05,
+      "loss": 0.7306,
+      "step": 2835
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.391445392918237,
+      "learning_rate": 3.498320747148455e-05,
+      "loss": 0.6627,
+      "step": 2836
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.42634461012864033,
+      "learning_rate": 3.497937027364164e-05,
+      "loss": 0.6906,
+      "step": 2837
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.36180042698682413,
+      "learning_rate": 3.4975531819497027e-05,
+      "loss": 0.6638,
+      "step": 2838
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4327347137088507,
+      "learning_rate": 3.4971692109372656e-05,
+      "loss": 0.7211,
+      "step": 2839
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.39068862321343756,
+      "learning_rate": 3.496785114359054e-05,
+      "loss": 0.654,
+      "step": 2840
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.40325296852244,
+      "learning_rate": 3.496400892247283e-05,
+      "loss": 0.7196,
+      "step": 2841
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3660461159450299,
+      "learning_rate": 3.496016544634176e-05,
+      "loss": 0.6856,
+      "step": 2842
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3673878884171619,
+      "learning_rate": 3.4956320715519675e-05,
+      "loss": 0.6749,
+      "step": 2843
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.43671001738700393,
+      "learning_rate": 3.495247473032905e-05,
+      "loss": 0.7042,
+      "step": 2844
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4272921134282617,
+      "learning_rate": 3.494862749109241e-05,
+      "loss": 0.7029,
+      "step": 2845
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3778417385790537,
+      "learning_rate": 3.494477899813245e-05,
+      "loss": 0.6895,
+      "step": 2846
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.39762516459494684,
+      "learning_rate": 3.494092925177192e-05,
+      "loss": 0.6931,
+      "step": 2847
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.43545248080055865,
+      "learning_rate": 3.49370782523337e-05,
+      "loss": 0.7187,
+      "step": 2848
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.45159248531168844,
+      "learning_rate": 3.493322600014076e-05,
+      "loss": 0.7009,
+      "step": 2849
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4135378572825343,
+      "learning_rate": 3.492937249551619e-05,
+      "loss": 0.7153,
+      "step": 2850
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.46229041853554376,
+      "learning_rate": 3.492551773878318e-05,
+      "loss": 0.71,
+      "step": 2851
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.41072731202717594,
+      "learning_rate": 3.4921661730265025e-05,
+      "loss": 0.7192,
+      "step": 2852
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4285590413965995,
+      "learning_rate": 3.4917804470285124e-05,
+      "loss": 0.7144,
+      "step": 2853
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4179232012605743,
+      "learning_rate": 3.491394595916697e-05,
+      "loss": 0.7043,
+      "step": 2854
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.41776432032113314,
+      "learning_rate": 3.4910086197234195e-05,
+      "loss": 0.7247,
+      "step": 2855
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.37330944892655005,
+      "learning_rate": 3.490622518481049e-05,
+      "loss": 0.6916,
+      "step": 2856
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4391705330264212,
+      "learning_rate": 3.4902362922219673e-05,
+      "loss": 0.7088,
+      "step": 2857
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.34625539331796434,
+      "learning_rate": 3.4898499409785685e-05,
+      "loss": 0.6859,
+      "step": 2858
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4308977227093952,
+      "learning_rate": 3.489463464783254e-05,
+      "loss": 0.6994,
+      "step": 2859
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.40927494592984287,
+      "learning_rate": 3.489076863668437e-05,
+      "loss": 0.7015,
+      "step": 2860
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4117989950589254,
+      "learning_rate": 3.4886901376665424e-05,
+      "loss": 0.6616,
+      "step": 2861
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.41370037475154,
+      "learning_rate": 3.488303286810004e-05,
+      "loss": 0.6839,
+      "step": 2862
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.35496964493331645,
+      "learning_rate": 3.487916311131266e-05,
+      "loss": 0.6735,
+      "step": 2863
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4003806056171781,
+      "learning_rate": 3.487529210662784e-05,
+      "loss": 0.6796,
+      "step": 2864
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.45518229577575375,
+      "learning_rate": 3.487141985437024e-05,
+      "loss": 0.7642,
+      "step": 2865
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4296179407319972,
+      "learning_rate": 3.486754635486461e-05,
+      "loss": 0.6942,
+      "step": 2866
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.38433095490861685,
+      "learning_rate": 3.4863671608435825e-05,
+      "loss": 0.6654,
+      "step": 2867
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.439968744570548,
+      "learning_rate": 3.4859795615408855e-05,
+      "loss": 0.6392,
+      "step": 2868
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4542668387746079,
+      "learning_rate": 3.4855918376108773e-05,
+      "loss": 0.7199,
+      "step": 2869
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3924832407446957,
+      "learning_rate": 3.485203989086076e-05,
+      "loss": 0.6814,
+      "step": 2870
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4150552162878009,
+      "learning_rate": 3.484816015999009e-05,
+      "loss": 0.6826,
+      "step": 2871
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.41353638693657424,
+      "learning_rate": 3.484427918382217e-05,
+      "loss": 0.7324,
+      "step": 2872
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4090563360667662,
+      "learning_rate": 3.484039696268248e-05,
+      "loss": 0.7059,
+      "step": 2873
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.39474924462497124,
+      "learning_rate": 3.483651349689661e-05,
+      "loss": 0.6542,
+      "step": 2874
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4988193850456227,
+      "learning_rate": 3.4832628786790276e-05,
+      "loss": 0.6902,
+      "step": 2875
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.43778380305893805,
+      "learning_rate": 3.482874283268929e-05,
+      "loss": 0.7188,
+      "step": 2876
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4286258583653468,
+      "learning_rate": 3.4824855634919535e-05,
+      "loss": 0.6797,
+      "step": 2877
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.3167214600161197,
+      "learning_rate": 3.482096719380705e-05,
+      "loss": 7.4834,
+      "step": 2878
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4197694093644619,
+      "learning_rate": 3.481707750967794e-05,
+      "loss": 0.6973,
+      "step": 2879
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5005469011431511,
+      "learning_rate": 3.481318658285844e-05,
+      "loss": 0.7066,
+      "step": 2880
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.36244961462012815,
+      "learning_rate": 3.480929441367487e-05,
+      "loss": 0.6794,
+      "step": 2881
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.39478876946704083,
+      "learning_rate": 3.480540100245366e-05,
+      "loss": 0.7082,
+      "step": 2882
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.36172866155711925,
+      "learning_rate": 3.480150634952135e-05,
+      "loss": 0.6587,
+      "step": 2883
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.42805194245565437,
+      "learning_rate": 3.479761045520458e-05,
+      "loss": 0.6901,
+      "step": 2884
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4060132044494527,
+      "learning_rate": 3.479371331983009e-05,
+      "loss": 0.6541,
+      "step": 2885
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.34794827780243465,
+      "learning_rate": 3.478981494372473e-05,
+      "loss": 0.665,
+      "step": 2886
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4433198201023092,
+      "learning_rate": 3.4785915327215444e-05,
+      "loss": 0.7101,
+      "step": 2887
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.420207227666944,
+      "learning_rate": 3.478201447062931e-05,
+      "loss": 0.7133,
+      "step": 2888
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.37953203448649786,
+      "learning_rate": 3.477811237429347e-05,
+      "loss": 0.6956,
+      "step": 2889
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3992742273907896,
+      "learning_rate": 3.477420903853519e-05,
+      "loss": 0.7261,
+      "step": 2890
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3784757052324262,
+      "learning_rate": 3.477030446368185e-05,
+      "loss": 0.6524,
+      "step": 2891
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4067970480774829,
+      "learning_rate": 3.47663986500609e-05,
+      "loss": 0.7128,
+      "step": 2892
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.40117853745272947,
+      "learning_rate": 3.4762491597999935e-05,
+      "loss": 0.6727,
+      "step": 2893
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.37920339190189134,
+      "learning_rate": 3.475858330782662e-05,
+      "loss": 0.6809,
+      "step": 2894
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.43879145685360366,
+      "learning_rate": 3.475467377986876e-05,
+      "loss": 0.6733,
+      "step": 2895
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3784906571606364,
+      "learning_rate": 3.4750763014454216e-05,
+      "loss": 0.69,
+      "step": 2896
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3747233328140673,
+      "learning_rate": 3.4746851011911e-05,
+      "loss": 0.6608,
+      "step": 2897
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.41857132606570224,
+      "learning_rate": 3.47429377725672e-05,
+      "loss": 0.7242,
+      "step": 2898
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.38291594551459934,
+      "learning_rate": 3.473902329675102e-05,
+      "loss": 0.7113,
+      "step": 2899
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.39931244803939125,
+      "learning_rate": 3.4735107584790745e-05,
+      "loss": 0.6921,
+      "step": 2900
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.39222979673831565,
+      "learning_rate": 3.47311906370148e-05,
+      "loss": 0.6889,
+      "step": 2901
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3898506349830417,
+      "learning_rate": 3.472727245375167e-05,
+      "loss": 0.6721,
+      "step": 2902
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 3.9909371440569954,
+      "learning_rate": 3.472335303533001e-05,
+      "loss": 7.4316,
+      "step": 2903
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4414527600497929,
+      "learning_rate": 3.4719432382078496e-05,
+      "loss": 0.6952,
+      "step": 2904
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4276447215847779,
+      "learning_rate": 3.4715510494325965e-05,
+      "loss": 0.7203,
+      "step": 2905
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4555033219193227,
+      "learning_rate": 3.471158737240135e-05,
+      "loss": 0.7492,
+      "step": 2906
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.44771861413099456,
+      "learning_rate": 3.4707663016633666e-05,
+      "loss": 0.6873,
+      "step": 2907
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.43666739656877307,
+      "learning_rate": 3.470373742735205e-05,
+      "loss": 0.6847,
+      "step": 2908
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4536523672597925,
+      "learning_rate": 3.4699810604885726e-05,
+      "loss": 0.7095,
+      "step": 2909
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4699120600286391,
+      "learning_rate": 3.469588254956404e-05,
+      "loss": 0.7303,
+      "step": 2910
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4277238879563874,
+      "learning_rate": 3.469195326171644e-05,
+      "loss": 0.6956,
+      "step": 2911
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.420393339742015,
+      "learning_rate": 3.468802274167246e-05,
+      "loss": 0.7167,
+      "step": 2912
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.37844935520727346,
+      "learning_rate": 3.468409098976175e-05,
+      "loss": 0.6708,
+      "step": 2913
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.48747919967432657,
+      "learning_rate": 3.468015800631406e-05,
+      "loss": 0.7002,
+      "step": 2914
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.43777991117933585,
+      "learning_rate": 3.467622379165925e-05,
+      "loss": 0.6712,
+      "step": 2915
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4248544061491322,
+      "learning_rate": 3.467228834612727e-05,
+      "loss": 0.7164,
+      "step": 2916
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4118708282625414,
+      "learning_rate": 3.466835167004819e-05,
+      "loss": 0.7159,
+      "step": 2917
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3991276116645618,
+      "learning_rate": 3.466441376375217e-05,
+      "loss": 0.7018,
+      "step": 2918
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4363453798235504,
+      "learning_rate": 3.466047462756948e-05,
+      "loss": 0.7504,
+      "step": 2919
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4008015154872547,
+      "learning_rate": 3.465653426183049e-05,
+      "loss": 0.6477,
+      "step": 2920
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.45165104071781514,
+      "learning_rate": 3.465259266686567e-05,
+      "loss": 0.6811,
+      "step": 2921
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.413942977477078,
+      "learning_rate": 3.4648649843005597e-05,
+      "loss": 0.6806,
+      "step": 2922
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3745126952857293,
+      "learning_rate": 3.464470579058096e-05,
+      "loss": 0.6728,
+      "step": 2923
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4559462224676761,
+      "learning_rate": 3.464076050992252e-05,
+      "loss": 0.6905,
+      "step": 2924
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.42111874274011296,
+      "learning_rate": 3.463681400136119e-05,
+      "loss": 0.7056,
+      "step": 2925
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4542755854413041,
+      "learning_rate": 3.463286626522794e-05,
+      "loss": 0.7166,
+      "step": 2926
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3923837349511645,
+      "learning_rate": 3.462891730185387e-05,
+      "loss": 0.6783,
+      "step": 2927
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4667536706485057,
+      "learning_rate": 3.4624967111570175e-05,
+      "loss": 0.696,
+      "step": 2928
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3689385247309114,
+      "learning_rate": 3.462101569470815e-05,
+      "loss": 0.685,
+      "step": 2929
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.34795778531828464,
+      "learning_rate": 3.461706305159919e-05,
+      "loss": 0.6522,
+      "step": 2930
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4419274459733643,
+      "learning_rate": 3.461310918257482e-05,
+      "loss": 0.676,
+      "step": 2931
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.42974201895370046,
+      "learning_rate": 3.460915408796662e-05,
+      "loss": 0.6973,
+      "step": 2932
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4484933990888038,
+      "learning_rate": 3.4605197768106306e-05,
+      "loss": 0.69,
+      "step": 2933
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.38702195785900084,
+      "learning_rate": 3.4601240223325704e-05,
+      "loss": 0.6373,
+      "step": 2934
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4394172976756947,
+      "learning_rate": 3.459728145395671e-05,
+      "loss": 0.7107,
+      "step": 2935
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.45318498760341197,
+      "learning_rate": 3.459332146033136e-05,
+      "loss": 0.6864,
+      "step": 2936
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.41208493315034433,
+      "learning_rate": 3.4589360242781755e-05,
+      "loss": 0.6768,
+      "step": 2937
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.37944526819420393,
+      "learning_rate": 3.458539780164013e-05,
+      "loss": 0.6799,
+      "step": 2938
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3957393908555829,
+      "learning_rate": 3.4581434137238806e-05,
+      "loss": 0.6639,
+      "step": 2939
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4380511283727663,
+      "learning_rate": 3.457746924991021e-05,
+      "loss": 0.7112,
+      "step": 2940
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4499411385873113,
+      "learning_rate": 3.457350313998687e-05,
+      "loss": 0.7186,
+      "step": 2941
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3744032248361135,
+      "learning_rate": 3.456953580780142e-05,
+      "loss": 0.7258,
+      "step": 2942
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4239694651934847,
+      "learning_rate": 3.456556725368661e-05,
+      "loss": 0.6929,
+      "step": 2943
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.48189348267108617,
+      "learning_rate": 3.4561597477975254e-05,
+      "loss": 0.6834,
+      "step": 2944
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3827147576400864,
+      "learning_rate": 3.455762648100031e-05,
+      "loss": 0.6288,
+      "step": 2945
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4947279745233755,
+      "learning_rate": 3.4553654263094805e-05,
+      "loss": 0.6995,
+      "step": 2946
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.421858023820289,
+      "learning_rate": 3.45496808245919e-05,
+      "loss": 0.698,
+      "step": 2947
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3634996152024769,
+      "learning_rate": 3.454570616582483e-05,
+      "loss": 0.6581,
+      "step": 2948
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4450954469040707,
+      "learning_rate": 3.4541730287126954e-05,
+      "loss": 0.7012,
+      "step": 2949
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4313105125943785,
+      "learning_rate": 3.453775318883173e-05,
+      "loss": 0.6944,
+      "step": 2950
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3788216278897195,
+      "learning_rate": 3.453377487127269e-05,
+      "loss": 0.6683,
+      "step": 2951
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.35702897937792566,
+      "learning_rate": 3.4529795334783513e-05,
+      "loss": 0.6565,
+      "step": 2952
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.459591585839877,
+      "learning_rate": 3.4525814579697944e-05,
+      "loss": 0.6942,
+      "step": 2953
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5382257571749481,
+      "learning_rate": 3.452183260634985e-05,
+      "loss": 0.7625,
+      "step": 2954
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4139191678870221,
+      "learning_rate": 3.451784941507319e-05,
+      "loss": 0.6994,
+      "step": 2955
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.35381078310146213,
+      "learning_rate": 3.4513865006202035e-05,
+      "loss": 0.7045,
+      "step": 2956
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.37367494141456636,
+      "learning_rate": 3.450987938007056e-05,
+      "loss": 0.6947,
+      "step": 2957
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.45939308595126643,
+      "learning_rate": 3.450589253701301e-05,
+      "loss": 0.7741,
+      "step": 2958
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4372262258848035,
+      "learning_rate": 3.4501904477363774e-05,
+      "loss": 0.7188,
+      "step": 2959
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4120789068446824,
+      "learning_rate": 3.449791520145733e-05,
+      "loss": 0.6841,
+      "step": 2960
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4807109951843884,
+      "learning_rate": 3.449392470962824e-05,
+      "loss": 0.6964,
+      "step": 2961
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.39055020034406346,
+      "learning_rate": 3.44899330022112e-05,
+      "loss": 0.69,
+      "step": 2962
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.48316388658732223,
+      "learning_rate": 3.4485940079540965e-05,
+      "loss": 0.7298,
+      "step": 2963
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.43338283788823256,
+      "learning_rate": 3.448194594195244e-05,
+      "loss": 0.7484,
+      "step": 2964
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.38290680458542437,
+      "learning_rate": 3.447795058978059e-05,
+      "loss": 0.6855,
+      "step": 2965
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4445602572499133,
+      "learning_rate": 3.4473954023360515e-05,
+      "loss": 0.6568,
+      "step": 2966
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.40374033102529283,
+      "learning_rate": 3.44699562430274e-05,
+      "loss": 0.7076,
+      "step": 2967
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4097459831248915,
+      "learning_rate": 3.4465957249116524e-05,
+      "loss": 0.7033,
+      "step": 2968
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41141950767872537,
+      "learning_rate": 3.4461957041963285e-05,
+      "loss": 0.6849,
+      "step": 2969
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.39706520114612454,
+      "learning_rate": 3.4457955621903175e-05,
+      "loss": 0.6792,
+      "step": 2970
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4885373183324069,
+      "learning_rate": 3.445395298927179e-05,
+      "loss": 0.6607,
+      "step": 2971
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41820981034918403,
+      "learning_rate": 3.444994914440482e-05,
+      "loss": 0.6987,
+      "step": 2972
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4397780383657602,
+      "learning_rate": 3.444594408763806e-05,
+      "loss": 0.6634,
+      "step": 2973
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3776558660470575,
+      "learning_rate": 3.444193781930743e-05,
+      "loss": 0.6583,
+      "step": 2974
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.39587196701125626,
+      "learning_rate": 3.443793033974891e-05,
+      "loss": 0.7043,
+      "step": 2975
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.43511712435534844,
+      "learning_rate": 3.443392164929861e-05,
+      "loss": 0.6907,
+      "step": 2976
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3831500556909923,
+      "learning_rate": 3.442991174829273e-05,
+      "loss": 0.7115,
+      "step": 2977
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.38246718837627436,
+      "learning_rate": 3.442590063706758e-05,
+      "loss": 0.7059,
+      "step": 2978
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.43508568542748904,
+      "learning_rate": 3.442188831595956e-05,
+      "loss": 0.7086,
+      "step": 2979
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4855175102502384,
+      "learning_rate": 3.4417874785305194e-05,
+      "loss": 0.7468,
+      "step": 2980
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4180891374430836,
+      "learning_rate": 3.441386004544108e-05,
+      "loss": 0.7105,
+      "step": 2981
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.43904138831895906,
+      "learning_rate": 3.440984409670393e-05,
+      "loss": 0.6836,
+      "step": 2982
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4174004280859614,
+      "learning_rate": 3.440582693943056e-05,
+      "loss": 0.7047,
+      "step": 2983
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.40975278663773157,
+      "learning_rate": 3.440180857395788e-05,
+      "loss": 0.7213,
+      "step": 2984
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4736175381362938,
+      "learning_rate": 3.439778900062291e-05,
+      "loss": 0.715,
+      "step": 2985
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3980613538530371,
+      "learning_rate": 3.439376821976276e-05,
+      "loss": 0.6858,
+      "step": 2986
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3794802148247474,
+      "learning_rate": 3.438974623171466e-05,
+      "loss": 0.6957,
+      "step": 2987
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3860423953827915,
+      "learning_rate": 3.4385723036815916e-05,
+      "loss": 0.6771,
+      "step": 2988
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4366658944161586,
+      "learning_rate": 3.438169863540395e-05,
+      "loss": 0.6936,
+      "step": 2989
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.37240713011192816,
+      "learning_rate": 3.4377673027816305e-05,
+      "loss": 0.6974,
+      "step": 2990
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.39798939020929414,
+      "learning_rate": 3.4373646214390576e-05,
+      "loss": 0.704,
+      "step": 2991
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.42916995664078994,
+      "learning_rate": 3.4369618195464494e-05,
+      "loss": 0.7104,
+      "step": 2992
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41406020647710307,
+      "learning_rate": 3.436558897137589e-05,
+      "loss": 0.6955,
+      "step": 2993
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7670325738727773,
+      "learning_rate": 3.436155854246269e-05,
+      "loss": 0.6966,
+      "step": 2994
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3863692651900546,
+      "learning_rate": 3.435752690906293e-05,
+      "loss": 0.6962,
+      "step": 2995
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.357074110421138,
+      "learning_rate": 3.435349407151471e-05,
+      "loss": 0.687,
+      "step": 2996
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.45596497770082545,
+      "learning_rate": 3.434946003015628e-05,
+      "loss": 0.7175,
+      "step": 2997
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41495702366681947,
+      "learning_rate": 3.434542478532598e-05,
+      "loss": 0.7483,
+      "step": 2998
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4284394394007488,
+      "learning_rate": 3.4341388337362215e-05,
+      "loss": 0.7041,
+      "step": 2999
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4406348778061033,
+      "learning_rate": 3.4337350686603536e-05,
+      "loss": 0.7082,
+      "step": 3000
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.42305344546297835,
+      "learning_rate": 3.433331183338857e-05,
+      "loss": 0.7228,
+      "step": 3001
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4729635810755663,
+      "learning_rate": 3.432927177805605e-05,
+      "loss": 0.6996,
+      "step": 3002
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.43464341508495574,
+      "learning_rate": 3.432523052094481e-05,
+      "loss": 0.7227,
+      "step": 3003
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4143430436388867,
+      "learning_rate": 3.432118806239379e-05,
+      "loss": 0.6758,
+      "step": 3004
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4959846974893361,
+      "learning_rate": 3.431714440274202e-05,
+      "loss": 0.7255,
+      "step": 3005
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4081752839674109,
+      "learning_rate": 3.431309954232864e-05,
+      "loss": 0.6931,
+      "step": 3006
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.42727319078879206,
+      "learning_rate": 3.430905348149288e-05,
+      "loss": 0.6953,
+      "step": 3007
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4037393098813674,
+      "learning_rate": 3.4305006220574096e-05,
+      "loss": 0.6885,
+      "step": 3008
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.40697630799552253,
+      "learning_rate": 3.430095775991172e-05,
+      "loss": 0.6942,
+      "step": 3009
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.42520736586845503,
+      "learning_rate": 3.429690809984528e-05,
+      "loss": 0.6804,
+      "step": 3010
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3450849228297821,
+      "learning_rate": 3.4292857240714425e-05,
+      "loss": 0.7074,
+      "step": 3011
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4290960166108636,
+      "learning_rate": 3.428880518285889e-05,
+      "loss": 0.6784,
+      "step": 3012
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4129664179137373,
+      "learning_rate": 3.428475192661853e-05,
+      "loss": 0.6823,
+      "step": 3013
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41281025052674286,
+      "learning_rate": 3.428069747233327e-05,
+      "loss": 0.7001,
+      "step": 3014
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.38194417351226057,
+      "learning_rate": 3.427664182034317e-05,
+      "loss": 0.7076,
+      "step": 3015
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.46250277858013655,
+      "learning_rate": 3.4272584970988356e-05,
+      "loss": 0.705,
+      "step": 3016
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.39490715775499474,
+      "learning_rate": 3.4268526924609066e-05,
+      "loss": 0.6938,
+      "step": 3017
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4227914318191905,
+      "learning_rate": 3.4264467681545674e-05,
+      "loss": 0.6844,
+      "step": 3018
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4042439895805792,
+      "learning_rate": 3.426040724213859e-05,
+      "loss": 0.6468,
+      "step": 3019
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.47381906617689923,
+      "learning_rate": 3.425634560672838e-05,
+      "loss": 0.7232,
+      "step": 3020
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4238743450429394,
+      "learning_rate": 3.4252282775655685e-05,
+      "loss": 0.6806,
+      "step": 3021
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4561181117749713,
+      "learning_rate": 3.424821874926123e-05,
+      "loss": 0.6952,
+      "step": 3022
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.42967765722770623,
+      "learning_rate": 3.424415352788589e-05,
+      "loss": 0.7107,
+      "step": 3023
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.409180666210175,
+      "learning_rate": 3.42400871118706e-05,
+      "loss": 0.6723,
+      "step": 3024
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4342192249486126,
+      "learning_rate": 3.423601950155639e-05,
+      "loss": 0.6783,
+      "step": 3025
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4381381700258796,
+      "learning_rate": 3.4231950697284424e-05,
+      "loss": 0.6901,
+      "step": 3026
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.36458702146591,
+      "learning_rate": 3.4227880699395935e-05,
+      "loss": 0.6625,
+      "step": 3027
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4192341373684975,
+      "learning_rate": 3.4223809508232285e-05,
+      "loss": 0.7315,
+      "step": 3028
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3451209126110774,
+      "learning_rate": 3.42197371241349e-05,
+      "loss": 0.6741,
+      "step": 3029
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.40759266406144173,
+      "learning_rate": 3.421566354744534e-05,
+      "loss": 0.698,
+      "step": 3030
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41082735545908267,
+      "learning_rate": 3.421158877850525e-05,
+      "loss": 0.6777,
+      "step": 3031
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4611728201439485,
+      "learning_rate": 3.420751281765637e-05,
+      "loss": 0.7175,
+      "step": 3032
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.40461168597324965,
+      "learning_rate": 3.420343566524054e-05,
+      "loss": 0.7003,
+      "step": 3033
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.38866496095085035,
+      "learning_rate": 3.419935732159972e-05,
+      "loss": 0.653,
+      "step": 3034
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.39773399827069417,
+      "learning_rate": 3.419527778707595e-05,
+      "loss": 0.6857,
+      "step": 3035
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.378838390037203,
+      "learning_rate": 3.4191197062011374e-05,
+      "loss": 0.6677,
+      "step": 3036
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.39901742017349867,
+      "learning_rate": 3.418711514674824e-05,
+      "loss": 0.6805,
+      "step": 3037
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41922001015481747,
+      "learning_rate": 3.418303204162889e-05,
+      "loss": 0.7211,
+      "step": 3038
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.42749737049502423,
+      "learning_rate": 3.417894774699578e-05,
+      "loss": 0.7065,
+      "step": 3039
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4043856663107311,
+      "learning_rate": 3.4174862263191446e-05,
+      "loss": 0.6978,
+      "step": 3040
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4159678495603867,
+      "learning_rate": 3.417077559055852e-05,
+      "loss": 0.6869,
+      "step": 3041
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.421893988165877,
+      "learning_rate": 3.4166687729439774e-05,
+      "loss": 0.7529,
+      "step": 3042
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4267150717984289,
+      "learning_rate": 3.4162598680178026e-05,
+      "loss": 0.6955,
+      "step": 3043
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3621652214488553,
+      "learning_rate": 3.415850844311624e-05,
+      "loss": 0.6491,
+      "step": 3044
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3952852037122343,
+      "learning_rate": 3.415441701859744e-05,
+      "loss": 0.7044,
+      "step": 3045
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4397491138917153,
+      "learning_rate": 3.415032440696478e-05,
+      "loss": 0.7196,
+      "step": 3046
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4381401747762043,
+      "learning_rate": 3.414623060856151e-05,
+      "loss": 0.6379,
+      "step": 3047
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4262377907054368,
+      "learning_rate": 3.4142135623730954e-05,
+      "loss": 0.719,
+      "step": 3048
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.44474616415343543,
+      "learning_rate": 3.413803945281657e-05,
+      "loss": 0.6931,
+      "step": 3049
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.44725663969574925,
+      "learning_rate": 3.413394209616188e-05,
+      "loss": 0.7312,
+      "step": 3050
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4038599044441596,
+      "learning_rate": 3.412984355411055e-05,
+      "loss": 0.6616,
+      "step": 3051
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4721215895520942,
+      "learning_rate": 3.41257438270063e-05,
+      "loss": 0.6858,
+      "step": 3052
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3737142430174592,
+      "learning_rate": 3.4121642915192974e-05,
+      "loss": 0.657,
+      "step": 3053
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41721147940619857,
+      "learning_rate": 3.411754081901451e-05,
+      "loss": 0.7245,
+      "step": 3054
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3672754916354457,
+      "learning_rate": 3.411343753881495e-05,
+      "loss": 0.6772,
+      "step": 3055
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.43359541362533777,
+      "learning_rate": 3.410933307493843e-05,
+      "loss": 0.7357,
+      "step": 3056
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4246416418246993,
+      "learning_rate": 3.4105227427729176e-05,
+      "loss": 0.7165,
+      "step": 3057
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3819371990731674,
+      "learning_rate": 3.410112059753154e-05,
+      "loss": 0.6916,
+      "step": 3058
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3513173429293682,
+      "learning_rate": 3.409701258468995e-05,
+      "loss": 0.6485,
+      "step": 3059
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.40018683466880745,
+      "learning_rate": 3.4092903389548936e-05,
+      "loss": 0.6789,
+      "step": 3060
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.37925969545272104,
+      "learning_rate": 3.408879301245313e-05,
+      "loss": 0.6868,
+      "step": 3061
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4171476473690013,
+      "learning_rate": 3.408468145374728e-05,
+      "loss": 0.6816,
+      "step": 3062
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4250224965896315,
+      "learning_rate": 3.4080568713776206e-05,
+      "loss": 0.6947,
+      "step": 3063
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4272548936186096,
+      "learning_rate": 3.407645479288483e-05,
+      "loss": 0.7365,
+      "step": 3064
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.46617814039391914,
+      "learning_rate": 3.407233969141819e-05,
+      "loss": 0.6903,
+      "step": 3065
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4373573966742151,
+      "learning_rate": 3.406822340972142e-05,
+      "loss": 0.7688,
+      "step": 3066
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4257592614530124,
+      "learning_rate": 3.406410594813975e-05,
+      "loss": 0.6786,
+      "step": 3067
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6135071416584441,
+      "learning_rate": 3.405998730701849e-05,
+      "loss": 0.6877,
+      "step": 3068
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3964969164820439,
+      "learning_rate": 3.4055867486703075e-05,
+      "loss": 0.6459,
+      "step": 3069
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4055615354317854,
+      "learning_rate": 3.4051746487539036e-05,
+      "loss": 0.6695,
+      "step": 3070
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4554305168526291,
+      "learning_rate": 3.404762430987198e-05,
+      "loss": 0.7496,
+      "step": 3071
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4260223231198322,
+      "learning_rate": 3.4043500954047646e-05,
+      "loss": 0.7006,
+      "step": 3072
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4434287603898639,
+      "learning_rate": 3.403937642041184e-05,
+      "loss": 0.6875,
+      "step": 3073
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5077775328985751,
+      "learning_rate": 3.403525070931049e-05,
+      "loss": 0.6931,
+      "step": 3074
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.38380841473208327,
+      "learning_rate": 3.403112382108962e-05,
+      "loss": 0.7052,
+      "step": 3075
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.1040355051331328,
+      "learning_rate": 3.402699575609534e-05,
+      "loss": 0.7756,
+      "step": 3076
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4126417690779714,
+      "learning_rate": 3.4022866514673864e-05,
+      "loss": 0.717,
+      "step": 3077
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5673255886161849,
+      "learning_rate": 3.40187360971715e-05,
+      "loss": 0.7158,
+      "step": 3078
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4362538657763912,
+      "learning_rate": 3.401460450393468e-05,
+      "loss": 0.726,
+      "step": 3079
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4034886161696716,
+      "learning_rate": 3.40104717353099e-05,
+      "loss": 0.6551,
+      "step": 3080
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3927617609427996,
+      "learning_rate": 3.400633779164378e-05,
+      "loss": 0.6479,
+      "step": 3081
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4045909089619454,
+      "learning_rate": 3.400220267328303e-05,
+      "loss": 0.6819,
+      "step": 3082
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.44990940197017043,
+      "learning_rate": 3.399806638057444e-05,
+      "loss": 0.7464,
+      "step": 3083
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3576210047077827,
+      "learning_rate": 3.3993928913864934e-05,
+      "loss": 0.6945,
+      "step": 3084
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3844588873645849,
+      "learning_rate": 3.398979027350151e-05,
+      "loss": 0.6728,
+      "step": 3085
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.37819093913158014,
+      "learning_rate": 3.398565045983127e-05,
+      "loss": 0.662,
+      "step": 3086
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3931804903584944,
+      "learning_rate": 3.398150947320142e-05,
+      "loss": 0.6921,
+      "step": 3087
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.411304965983631,
+      "learning_rate": 3.397736731395925e-05,
+      "loss": 0.6968,
+      "step": 3088
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4090866701080508,
+      "learning_rate": 3.397322398245217e-05,
+      "loss": 0.6597,
+      "step": 3089
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.43534967661760365,
+      "learning_rate": 3.396907947902767e-05,
+      "loss": 0.6952,
+      "step": 3090
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.39250406115874936,
+      "learning_rate": 3.396493380403334e-05,
+      "loss": 0.673,
+      "step": 3091
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.43392359221814436,
+      "learning_rate": 3.396078695781688e-05,
+      "loss": 0.6833,
+      "step": 3092
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.443855142860063,
+      "learning_rate": 3.395663894072607e-05,
+      "loss": 0.7182,
+      "step": 3093
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.42208724400863107,
+      "learning_rate": 3.395248975310882e-05,
+      "loss": 0.6585,
+      "step": 3094
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3690593521326722,
+      "learning_rate": 3.39483393953131e-05,
+      "loss": 0.7161,
+      "step": 3095
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4039404176418742,
+      "learning_rate": 3.3944187867687e-05,
+      "loss": 0.661,
+      "step": 3096
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4633333870980786,
+      "learning_rate": 3.39400351705787e-05,
+      "loss": 0.6787,
+      "step": 3097
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3761968467134338,
+      "learning_rate": 3.3935881304336495e-05,
+      "loss": 0.6822,
+      "step": 3098
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.36002361900886226,
+      "learning_rate": 3.393172626930874e-05,
+      "loss": 0.6471,
+      "step": 3099
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4536653751946035,
+      "learning_rate": 3.3927570065843945e-05,
+      "loss": 0.6946,
+      "step": 3100
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3933293130168169,
+      "learning_rate": 3.392341269429066e-05,
+      "loss": 0.6555,
+      "step": 3101
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.43669511289218427,
+      "learning_rate": 3.391925415499756e-05,
+      "loss": 0.7166,
+      "step": 3102
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4400763294342398,
+      "learning_rate": 3.391509444831344e-05,
+      "loss": 0.6812,
+      "step": 3103
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.36007435876071486,
+      "learning_rate": 3.391093357458715e-05,
+      "loss": 0.6507,
+      "step": 3104
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.37664987294027524,
+      "learning_rate": 3.390677153416765e-05,
+      "loss": 0.6669,
+      "step": 3105
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4463267907747458,
+      "learning_rate": 3.390260832740403e-05,
+      "loss": 0.6972,
+      "step": 3106
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.49443751906884104,
+      "learning_rate": 3.389844395464543e-05,
+      "loss": 0.749,
+      "step": 3107
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.468949292438785,
+      "learning_rate": 3.389427841624113e-05,
+      "loss": 0.7076,
+      "step": 3108
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4350125047277639,
+      "learning_rate": 3.389011171254047e-05,
+      "loss": 0.7053,
+      "step": 3109
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 5.4765024024826845,
+      "learning_rate": 3.388594384389292e-05,
+      "loss": 7.3825,
+      "step": 3110
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4026502636467353,
+      "learning_rate": 3.388177481064803e-05,
+      "loss": 0.702,
+      "step": 3111
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4460169750290837,
+      "learning_rate": 3.3877604613155455e-05,
+      "loss": 0.6823,
+      "step": 3112
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.43895575054628166,
+      "learning_rate": 3.3873433251764935e-05,
+      "loss": 0.7381,
+      "step": 3113
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.37617914501696464,
+      "learning_rate": 3.386926072682632e-05,
+      "loss": 0.6431,
+      "step": 3114
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.40347749991420595,
+      "learning_rate": 3.386508703868957e-05,
+      "loss": 0.6757,
+      "step": 3115
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.46787777073524467,
+      "learning_rate": 3.38609121877047e-05,
+      "loss": 0.7125,
+      "step": 3116
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.34982820102020956,
+      "learning_rate": 3.385673617422187e-05,
+      "loss": 0.6446,
+      "step": 3117
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.46672644718546735,
+      "learning_rate": 3.385255899859132e-05,
+      "loss": 0.7026,
+      "step": 3118
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.416713102457746,
+      "learning_rate": 3.384838066116337e-05,
+      "loss": 0.7036,
+      "step": 3119
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.34398531327776466,
+      "learning_rate": 3.3844201162288456e-05,
+      "loss": 0.6233,
+      "step": 3120
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3606078299456595,
+      "learning_rate": 3.3840020502317124e-05,
+      "loss": 0.6474,
+      "step": 3121
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3887035902907244,
+      "learning_rate": 3.3835838681599975e-05,
+      "loss": 0.682,
+      "step": 3122
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3893056329733332,
+      "learning_rate": 3.383165570048774e-05,
+      "loss": 0.6533,
+      "step": 3123
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.402256564895185,
+      "learning_rate": 3.3827471559331256e-05,
+      "loss": 0.6849,
+      "step": 3124
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.414930044167645,
+      "learning_rate": 3.382328625848143e-05,
+      "loss": 0.7356,
+      "step": 3125
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.356628354260074,
+      "learning_rate": 3.381909979828927e-05,
+      "loss": 0.6656,
+      "step": 3126
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.38967390775626637,
+      "learning_rate": 3.381491217910591e-05,
+      "loss": 0.6715,
+      "step": 3127
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3490516915843096,
+      "learning_rate": 3.3810723401282544e-05,
+      "loss": 0.6658,
+      "step": 3128
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.44395415906014213,
+      "learning_rate": 3.380653346517049e-05,
+      "loss": 0.6798,
+      "step": 3129
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3598065616229047,
+      "learning_rate": 3.380234237112115e-05,
+      "loss": 0.6777,
+      "step": 3130
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.42431116267424956,
+      "learning_rate": 3.379815011948602e-05,
+      "loss": 0.6521,
+      "step": 3131
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3873995022458552,
+      "learning_rate": 3.37939567106167e-05,
+      "loss": 0.6674,
+      "step": 3132
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3495974954512865,
+      "learning_rate": 3.378976214486489e-05,
+      "loss": 0.6901,
+      "step": 3133
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3936226253154194,
+      "learning_rate": 3.378556642258239e-05,
+      "loss": 0.6578,
+      "step": 3134
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3989950336848445,
+      "learning_rate": 3.378136954412108e-05,
+      "loss": 0.6598,
+      "step": 3135
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3933363518372692,
+      "learning_rate": 3.377717150983295e-05,
+      "loss": 0.6872,
+      "step": 3136
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3862337924979248,
+      "learning_rate": 3.377297232007008e-05,
+      "loss": 0.7,
+      "step": 3137
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9055146901214397,
+      "learning_rate": 3.376877197518466e-05,
+      "loss": 7.4397,
+      "step": 3138
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5274720111271096,
+      "learning_rate": 3.376457047552896e-05,
+      "loss": 0.7149,
+      "step": 3139
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8368570815289531,
+      "learning_rate": 3.3760367821455364e-05,
+      "loss": 0.6822,
+      "step": 3140
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.581636503413984,
+      "learning_rate": 3.375616401331633e-05,
+      "loss": 0.7324,
+      "step": 3141
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4130192518483716,
+      "learning_rate": 3.3751959051464434e-05,
+      "loss": 0.6871,
+      "step": 3142
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.48365593992591904,
+      "learning_rate": 3.374775293625234e-05,
+      "loss": 0.7423,
+      "step": 3143
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4565963218852305,
+      "learning_rate": 3.374354566803281e-05,
+      "loss": 0.7333,
+      "step": 3144
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.38010320391102553,
+      "learning_rate": 3.373933724715871e-05,
+      "loss": 0.6596,
+      "step": 3145
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4441680878108839,
+      "learning_rate": 3.373512767398299e-05,
+      "loss": 0.7009,
+      "step": 3146
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3914524042988782,
+      "learning_rate": 3.3730916948858684e-05,
+      "loss": 0.712,
+      "step": 3147
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4289806006272709,
+      "learning_rate": 3.372670507213896e-05,
+      "loss": 0.7434,
+      "step": 3148
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3998529638094462,
+      "learning_rate": 3.372249204417707e-05,
+      "loss": 0.6816,
+      "step": 3149
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3854901739050163,
+      "learning_rate": 3.371827786532634e-05,
+      "loss": 0.679,
+      "step": 3150
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4096571522053148,
+      "learning_rate": 3.371406253594021e-05,
+      "loss": 0.6943,
+      "step": 3151
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3908838353891811,
+      "learning_rate": 3.3709846056372224e-05,
+      "loss": 0.685,
+      "step": 3152
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.46084792915511674,
+      "learning_rate": 3.3705628426976e-05,
+      "loss": 0.7359,
+      "step": 3153
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.45930828082867975,
+      "learning_rate": 3.370140964810527e-05,
+      "loss": 0.6734,
+      "step": 3154
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4203751083667515,
+      "learning_rate": 3.369718972011387e-05,
+      "loss": 0.6557,
+      "step": 3155
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4640218245458999,
+      "learning_rate": 3.3692968643355706e-05,
+      "loss": 0.7398,
+      "step": 3156
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.43916234363431095,
+      "learning_rate": 3.3688746418184795e-05,
+      "loss": 0.7251,
+      "step": 3157
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.437043196911583,
+      "learning_rate": 3.368452304495525e-05,
+      "loss": 0.6958,
+      "step": 3158
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.38560279661721253,
+      "learning_rate": 3.3680298524021295e-05,
+      "loss": 0.6477,
+      "step": 3159
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.41815477998400574,
+      "learning_rate": 3.367607285573722e-05,
+      "loss": 0.6961,
+      "step": 3160
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4742909637682478,
+      "learning_rate": 3.367184604045743e-05,
+      "loss": 0.7326,
+      "step": 3161
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4144395923570391,
+      "learning_rate": 3.3667618078536426e-05,
+      "loss": 0.7263,
+      "step": 3162
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.442073347522807,
+      "learning_rate": 3.36633889703288e-05,
+      "loss": 0.7093,
+      "step": 3163
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.38107525704816136,
+      "learning_rate": 3.365915871618923e-05,
+      "loss": 0.7018,
+      "step": 3164
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4053791630773425,
+      "learning_rate": 3.365492731647253e-05,
+      "loss": 0.6657,
+      "step": 3165
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4175754052826737,
+      "learning_rate": 3.365069477153356e-05,
+      "loss": 0.6296,
+      "step": 3166
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4256919038111365,
+      "learning_rate": 3.364646108172731e-05,
+      "loss": 0.6591,
+      "step": 3167
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.352206781743714,
+      "learning_rate": 3.364222624740885e-05,
+      "loss": 0.6123,
+      "step": 3168
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3753751927079323,
+      "learning_rate": 3.3637990268933345e-05,
+      "loss": 0.6621,
+      "step": 3169
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4211403644788685,
+      "learning_rate": 3.363375314665607e-05,
+      "loss": 0.6974,
+      "step": 3170
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.42272959782237435,
+      "learning_rate": 3.362951488093238e-05,
+      "loss": 0.6864,
+      "step": 3171
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3688204238326773,
+      "learning_rate": 3.362527547211774e-05,
+      "loss": 0.6806,
+      "step": 3172
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.38998025176642576,
+      "learning_rate": 3.362103492056771e-05,
+      "loss": 0.6859,
+      "step": 3173
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.034166758777849,
+      "learning_rate": 3.361679322663792e-05,
+      "loss": 7.3779,
+      "step": 3174
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4202667354528535,
+      "learning_rate": 3.361255039068413e-05,
+      "loss": 0.6824,
+      "step": 3175
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.43082684984189135,
+      "learning_rate": 3.3608306413062174e-05,
+      "loss": 0.7308,
+      "step": 3176
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.44586455148224247,
+      "learning_rate": 3.3604061294128e-05,
+      "loss": 0.7251,
+      "step": 3177
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4487878560579509,
+      "learning_rate": 3.359981503423764e-05,
+      "loss": 0.684,
+      "step": 3178
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5889370934240186,
+      "learning_rate": 3.359556763374721e-05,
+      "loss": 7.3296,
+      "step": 3179
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.43646802688072805,
+      "learning_rate": 3.3591319093012935e-05,
+      "loss": 0.7216,
+      "step": 3180
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4236354309118148,
+      "learning_rate": 3.3587069412391154e-05,
+      "loss": 0.7229,
+      "step": 3181
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.42735745380408696,
+      "learning_rate": 3.358281859223826e-05,
+      "loss": 0.6778,
+      "step": 3182
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4575980243791074,
+      "learning_rate": 3.357856663291078e-05,
+      "loss": 0.7269,
+      "step": 3183
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4455208571375163,
+      "learning_rate": 3.357431353476532e-05,
+      "loss": 0.7069,
+      "step": 3184
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.44124011882970865,
+      "learning_rate": 3.357005929815857e-05,
+      "loss": 0.7036,
+      "step": 3185
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.39027493606697,
+      "learning_rate": 3.356580392344734e-05,
+      "loss": 0.6359,
+      "step": 3186
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4482028036029781,
+      "learning_rate": 3.356154741098852e-05,
+      "loss": 0.7422,
+      "step": 3187
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.40819568423853697,
+      "learning_rate": 3.355728976113909e-05,
+      "loss": 0.6778,
+      "step": 3188
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.40956831040641806,
+      "learning_rate": 3.355303097425615e-05,
+      "loss": 0.6265,
+      "step": 3189
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3974142154529256,
+      "learning_rate": 3.354877105069686e-05,
+      "loss": 0.6711,
+      "step": 3190
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5321289643834926,
+      "learning_rate": 3.354450999081851e-05,
+      "loss": 0.6763,
+      "step": 3191
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.36579808086203686,
+      "learning_rate": 3.354024779497846e-05,
+      "loss": 0.6526,
+      "step": 3192
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3713191699107512,
+      "learning_rate": 3.353598446353418e-05,
+      "loss": 0.6863,
+      "step": 3193
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.44095957255821816,
+      "learning_rate": 3.353171999684324e-05,
+      "loss": 0.6787,
+      "step": 3194
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.38222725803626983,
+      "learning_rate": 3.3527454395263284e-05,
+      "loss": 0.6864,
+      "step": 3195
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.4029966082822105,
+      "learning_rate": 3.3523187659152057e-05,
+      "loss": 0.7219,
+      "step": 3196
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3377718521801338,
+      "learning_rate": 3.351891978886742e-05,
+      "loss": 0.6602,
+      "step": 3197
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.40536264104787534,
+      "learning_rate": 3.351465078476731e-05,
+      "loss": 0.6799,
+      "step": 3198
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.37050769976049674,
+      "learning_rate": 3.3510380647209746e-05,
+      "loss": 0.6596,
+      "step": 3199
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4058668684280955,
+      "learning_rate": 3.3506109376552887e-05,
+      "loss": 0.7057,
+      "step": 3200
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4131394860619081,
+      "learning_rate": 3.350183697315494e-05,
+      "loss": 0.6587,
+      "step": 3201
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.43758998928371834,
+      "learning_rate": 3.3497563437374225e-05,
+      "loss": 0.6967,
+      "step": 3202
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.42000788733713723,
+      "learning_rate": 3.3493288769569176e-05,
+      "loss": 0.6893,
+      "step": 3203
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.869612479274852,
+      "learning_rate": 3.348901297009829e-05,
+      "loss": 7.2677,
+      "step": 3204
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.43542050560786655,
+      "learning_rate": 3.348473603932018e-05,
+      "loss": 0.6729,
+      "step": 3205
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 4.593464994244523,
+      "learning_rate": 3.3480457977593537e-05,
+      "loss": 0.6929,
+      "step": 3206
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4070607029378866,
+      "learning_rate": 3.3476178785277174e-05,
+      "loss": 0.6431,
+      "step": 3207
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4221120510179955,
+      "learning_rate": 3.347189846272996e-05,
+      "loss": 0.6666,
+      "step": 3208
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.43086497759693687,
+      "learning_rate": 3.3467617010310896e-05,
+      "loss": 0.7498,
+      "step": 3209
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5267531055454531,
+      "learning_rate": 3.346333442837906e-05,
+      "loss": 0.6802,
+      "step": 3210
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.0647816641125674,
+      "learning_rate": 3.3459050717293625e-05,
+      "loss": 0.7164,
+      "step": 3211
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7224958296727374,
+      "learning_rate": 3.345476587741386e-05,
+      "loss": 0.6829,
+      "step": 3212
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5099585786250783,
+      "learning_rate": 3.3450479909099134e-05,
+      "loss": 0.6519,
+      "step": 3213
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5231724621510861,
+      "learning_rate": 3.34461928127089e-05,
+      "loss": 0.6904,
+      "step": 3214
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6952640103496351,
+      "learning_rate": 3.3441904588602714e-05,
+      "loss": 0.7168,
+      "step": 3215
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3942833565374841,
+      "learning_rate": 3.343761523714022e-05,
+      "loss": 0.6848,
+      "step": 3216
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.443309328116339,
+      "learning_rate": 3.343332475868116e-05,
+      "loss": 0.7099,
+      "step": 3217
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3613495235434658,
+      "learning_rate": 3.342903315358538e-05,
+      "loss": 0.6538,
+      "step": 3218
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4449866209264452,
+      "learning_rate": 3.342474042221282e-05,
+      "loss": 0.69,
+      "step": 3219
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4347680132432836,
+      "learning_rate": 3.3420446564923476e-05,
+      "loss": 0.741,
+      "step": 3220
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.36534606916740814,
+      "learning_rate": 3.34161515820775e-05,
+      "loss": 0.6639,
+      "step": 3221
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4821083885623824,
+      "learning_rate": 3.341185547403509e-05,
+      "loss": 0.7035,
+      "step": 3222
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.41771071809302873,
+      "learning_rate": 3.340755824115655e-05,
+      "loss": 0.6917,
+      "step": 3223
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.41943897335892477,
+      "learning_rate": 3.34032598838023e-05,
+      "loss": 0.6553,
+      "step": 3224
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.43036158386635054,
+      "learning_rate": 3.339896040233283e-05,
+      "loss": 0.6881,
+      "step": 3225
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3756636468896259,
+      "learning_rate": 3.339465979710873e-05,
+      "loss": 0.6892,
+      "step": 3226
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3991730833692644,
+      "learning_rate": 3.3390358068490694e-05,
+      "loss": 0.6857,
+      "step": 3227
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5557770944162338,
+      "learning_rate": 3.338605521683949e-05,
+      "loss": 0.6906,
+      "step": 3228
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.46672714643647045,
+      "learning_rate": 3.3381751242516014e-05,
+      "loss": 0.7273,
+      "step": 3229
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4787792064551006,
+      "learning_rate": 3.3377446145881214e-05,
+      "loss": 0.7094,
+      "step": 3230
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.377548865717099,
+      "learning_rate": 3.3373139927296164e-05,
+      "loss": 0.6872,
+      "step": 3231
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4720093276872643,
+      "learning_rate": 3.336883258712202e-05,
+      "loss": 0.6982,
+      "step": 3232
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.44207743891513573,
+      "learning_rate": 3.336452412572003e-05,
+      "loss": 0.6313,
+      "step": 3233
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4664512069479879,
+      "learning_rate": 3.3360214543451546e-05,
+      "loss": 0.7048,
+      "step": 3234
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.44017814005806133,
+      "learning_rate": 3.3355903840677996e-05,
+      "loss": 0.7416,
+      "step": 3235
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.36163631847984384,
+      "learning_rate": 3.3351592017760925e-05,
+      "loss": 0.6875,
+      "step": 3236
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.488277529811871,
+      "learning_rate": 3.334727907506195e-05,
+      "loss": 0.6463,
+      "step": 3237
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.43612869430833406,
+      "learning_rate": 3.3342965012942806e-05,
+      "loss": 0.6933,
+      "step": 3238
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3751596924169448,
+      "learning_rate": 3.33386498317653e-05,
+      "loss": 0.6604,
+      "step": 3239
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.35490347669431177,
+      "learning_rate": 3.333433353189134e-05,
+      "loss": 0.6361,
+      "step": 3240
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.39539867849557114,
+      "learning_rate": 3.333001611368293e-05,
+      "loss": 0.6613,
+      "step": 3241
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.41916383872189367,
+      "learning_rate": 3.332569757750217e-05,
+      "loss": 0.702,
+      "step": 3242
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.36666003318062745,
+      "learning_rate": 3.332137792371125e-05,
+      "loss": 0.6758,
+      "step": 3243
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.414297623570834,
+      "learning_rate": 3.3317057152672446e-05,
+      "loss": 0.7011,
+      "step": 3244
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4355372573829986,
+      "learning_rate": 3.3312735264748136e-05,
+      "loss": 0.6973,
+      "step": 3245
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5132676918323834,
+      "learning_rate": 3.330841226030081e-05,
+      "loss": 0.7215,
+      "step": 3246
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3973011345112023,
+      "learning_rate": 3.330408813969301e-05,
+      "loss": 0.7131,
+      "step": 3247
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.35876280532395854,
+      "learning_rate": 3.329976290328741e-05,
+      "loss": 0.672,
+      "step": 3248
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4793628644969398,
+      "learning_rate": 3.329543655144676e-05,
+      "loss": 0.7229,
+      "step": 3249
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4454866215891234,
+      "learning_rate": 3.32911090845339e-05,
+      "loss": 0.7072,
+      "step": 3250
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3613813217566518,
+      "learning_rate": 3.328678050291177e-05,
+      "loss": 0.6447,
+      "step": 3251
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3944720951418567,
+      "learning_rate": 3.32824508069434e-05,
+      "loss": 0.6681,
+      "step": 3252
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3687416731213321,
+      "learning_rate": 3.327811999699193e-05,
+      "loss": 0.6769,
+      "step": 3253
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4224054113148723,
+      "learning_rate": 3.327378807342057e-05,
+      "loss": 0.7053,
+      "step": 3254
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4134701536460793,
+      "learning_rate": 3.3269455036592644e-05,
+      "loss": 0.709,
+      "step": 3255
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.44908476932495706,
+      "learning_rate": 3.326512088687154e-05,
+      "loss": 0.696,
+      "step": 3256
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3472416472613683,
+      "learning_rate": 3.326078562462077e-05,
+      "loss": 0.6394,
+      "step": 3257
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.362526744298043,
+      "learning_rate": 3.325644925020393e-05,
+      "loss": 0.6533,
+      "step": 3258
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4010090507399522,
+      "learning_rate": 3.325211176398469e-05,
+      "loss": 0.6746,
+      "step": 3259
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4093174149083709,
+      "learning_rate": 3.324777316632685e-05,
+      "loss": 0.6598,
+      "step": 3260
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.39006738400280383,
+      "learning_rate": 3.324343345759427e-05,
+      "loss": 0.6717,
+      "step": 3261
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3976726835284563,
+      "learning_rate": 3.323909263815091e-05,
+      "loss": 0.7517,
+      "step": 3262
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4551705709847696,
+      "learning_rate": 3.323475070836085e-05,
+      "loss": 0.7168,
+      "step": 3263
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4389658681681229,
+      "learning_rate": 3.323040766858824e-05,
+      "loss": 0.705,
+      "step": 3264
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3837930863922495,
+      "learning_rate": 3.32260635191973e-05,
+      "loss": 0.6721,
+      "step": 3265
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.34142581741709277,
+      "learning_rate": 3.322171826055239e-05,
+      "loss": 0.6716,
+      "step": 3266
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.364055225167357,
+      "learning_rate": 3.3217371893017945e-05,
+      "loss": 0.6811,
+      "step": 3267
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.4697788403263616,
+      "learning_rate": 3.3213024416958474e-05,
+      "loss": 7.3261,
+      "step": 3268
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4158789945989467,
+      "learning_rate": 3.320867583273861e-05,
+      "loss": 0.677,
+      "step": 3269
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3507811784425293,
+      "learning_rate": 3.320432614072305e-05,
+      "loss": 0.6457,
+      "step": 3270
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4258491789551792,
+      "learning_rate": 3.319997534127661e-05,
+      "loss": 0.6788,
+      "step": 3271
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.466702779306577,
+      "learning_rate": 3.319562343476417e-05,
+      "loss": 0.6831,
+      "step": 3272
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9716516838578783,
+      "learning_rate": 3.3191270421550734e-05,
+      "loss": 7.2934,
+      "step": 3273
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4371814992134926,
+      "learning_rate": 3.318691630200138e-05,
+      "loss": 0.6692,
+      "step": 3274
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4414163904342723,
+      "learning_rate": 3.318256107648128e-05,
+      "loss": 0.6813,
+      "step": 3275
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4240309334537368,
+      "learning_rate": 3.3178204745355706e-05,
+      "loss": 0.6844,
+      "step": 3276
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3396225426281739,
+      "learning_rate": 3.317384730899001e-05,
+      "loss": 0.6705,
+      "step": 3277
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4211844431742362,
+      "learning_rate": 3.3169488767749656e-05,
+      "loss": 0.6933,
+      "step": 3278
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4020593734334987,
+      "learning_rate": 3.316512912200017e-05,
+      "loss": 0.6908,
+      "step": 3279
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4190421599942454,
+      "learning_rate": 3.316076837210722e-05,
+      "loss": 0.6944,
+      "step": 3280
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4078537848428112,
+      "learning_rate": 3.315640651843651e-05,
+      "loss": 0.6985,
+      "step": 3281
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.42476035500718023,
+      "learning_rate": 3.315204356135389e-05,
+      "loss": 0.682,
+      "step": 3282
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4445841924893984,
+      "learning_rate": 3.3147679501225245e-05,
+      "loss": 0.6999,
+      "step": 3283
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.41493142867847027,
+      "learning_rate": 3.31433143384166e-05,
+      "loss": 0.6686,
+      "step": 3284
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.39822522516120307,
+      "learning_rate": 3.313894807329406e-05,
+      "loss": 0.6457,
+      "step": 3285
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4436078360870724,
+      "learning_rate": 3.313458070622381e-05,
+      "loss": 0.6717,
+      "step": 3286
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9666696610747325,
+      "learning_rate": 3.313021223757214e-05,
+      "loss": 7.3967,
+      "step": 3287
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.537556421842459,
+      "learning_rate": 3.312584266770543e-05,
+      "loss": 0.7346,
+      "step": 3288
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.39364994611389337,
+      "learning_rate": 3.3121471996990134e-05,
+      "loss": 0.6207,
+      "step": 3289
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4000145408849178,
+      "learning_rate": 3.311710022579284e-05,
+      "loss": 0.6821,
+      "step": 3290
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4276467193529271,
+      "learning_rate": 3.311272735448019e-05,
+      "loss": 0.6876,
+      "step": 3291
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4012872443352911,
+      "learning_rate": 3.310835338341893e-05,
+      "loss": 0.696,
+      "step": 3292
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5117657054901414,
+      "learning_rate": 3.310397831297591e-05,
+      "loss": 0.7128,
+      "step": 3293
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.37854932901761595,
+      "learning_rate": 3.3099602143518054e-05,
+      "loss": 0.6967,
+      "step": 3294
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.33311925334164383,
+      "learning_rate": 3.309522487541238e-05,
+      "loss": 0.6467,
+      "step": 3295
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4265943331205213,
+      "learning_rate": 3.3090846509026006e-05,
+      "loss": 0.6703,
+      "step": 3296
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.40166614421833724,
+      "learning_rate": 3.308646704472616e-05,
+      "loss": 0.6746,
+      "step": 3297
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5962134324373096,
+      "learning_rate": 3.3082086482880115e-05,
+      "loss": 0.7184,
+      "step": 3298
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.36201584682573046,
+      "learning_rate": 3.307770482385528e-05,
+      "loss": 0.6766,
+      "step": 3299
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40651082981713604,
+      "learning_rate": 3.307332206801913e-05,
+      "loss": 0.6957,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.35707556669737156,
+      "learning_rate": 3.306893821573926e-05,
+      "loss": 0.6638,
+      "step": 3301
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40727718794193185,
+      "learning_rate": 3.306455326738332e-05,
+      "loss": 0.6884,
+      "step": 3302
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.46255505134557984,
+      "learning_rate": 3.3060167223319066e-05,
+      "loss": 0.7531,
+      "step": 3303
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4506971143115295,
+      "learning_rate": 3.305578008391437e-05,
+      "loss": 0.7376,
+      "step": 3304
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4386677781584626,
+      "learning_rate": 3.3051391849537165e-05,
+      "loss": 0.7293,
+      "step": 3305
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.41942590216032005,
+      "learning_rate": 3.304700252055548e-05,
+      "loss": 0.6691,
+      "step": 3306
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6633820415465619,
+      "learning_rate": 3.304261209733745e-05,
+      "loss": 7.3006,
+      "step": 3307
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.33456350256129613,
+      "learning_rate": 3.303822058025131e-05,
+      "loss": 0.6641,
+      "step": 3308
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.45014389111759856,
+      "learning_rate": 3.3033827969665336e-05,
+      "loss": 0.6711,
+      "step": 3309
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.45184052977125216,
+      "learning_rate": 3.302943426594797e-05,
+      "loss": 0.6564,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.42606935268389173,
+      "learning_rate": 3.3025039469467675e-05,
+      "loss": 0.7005,
+      "step": 3311
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4318280099893924,
+      "learning_rate": 3.302064358059305e-05,
+      "loss": 0.7033,
+      "step": 3312
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.997783143589582,
+      "learning_rate": 3.301624659969278e-05,
+      "loss": 7.4412,
+      "step": 3313
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4285717065566139,
+      "learning_rate": 3.3011848527135626e-05,
+      "loss": 0.7113,
+      "step": 3314
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4150527819480433,
+      "learning_rate": 3.3007449363290445e-05,
+      "loss": 0.6739,
+      "step": 3315
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4313438806842938,
+      "learning_rate": 3.30030491085262e-05,
+      "loss": 0.6739,
+      "step": 3316
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.45960110088330886,
+      "learning_rate": 3.299864776321193e-05,
+      "loss": 0.7087,
+      "step": 3317
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40630492688596265,
+      "learning_rate": 3.299424532771677e-05,
+      "loss": 0.7159,
+      "step": 3318
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3837583459271341,
+      "learning_rate": 3.298984180240994e-05,
+      "loss": 0.6912,
+      "step": 3319
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.470010839731149,
+      "learning_rate": 3.298543718766078e-05,
+      "loss": 0.7814,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.45612616397768113,
+      "learning_rate": 3.298103148383867e-05,
+      "loss": 0.6882,
+      "step": 3321
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4458286865096734,
+      "learning_rate": 3.2976624691313134e-05,
+      "loss": 0.6763,
+      "step": 3322
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.43267517523160504,
+      "learning_rate": 3.297221681045376e-05,
+      "loss": 0.7016,
+      "step": 3323
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4265990962939914,
+      "learning_rate": 3.2967807841630224e-05,
+      "loss": 0.7059,
+      "step": 3324
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4232333183984162,
+      "learning_rate": 3.2963397785212304e-05,
+      "loss": 0.6916,
+      "step": 3325
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5987894417350135,
+      "learning_rate": 3.2958986641569875e-05,
+      "loss": 7.3523,
+      "step": 3326
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40467779604720006,
+      "learning_rate": 3.295457441107288e-05,
+      "loss": 0.6538,
+      "step": 3327
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4719021794169592,
+      "learning_rate": 3.2950161094091376e-05,
+      "loss": 0.7061,
+      "step": 3328
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4331279648535401,
+      "learning_rate": 3.2945746690995504e-05,
+      "loss": 0.7011,
+      "step": 3329
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4234415130988891,
+      "learning_rate": 3.2941331202155486e-05,
+      "loss": 0.6919,
+      "step": 3330
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1038342932378507,
+      "learning_rate": 3.293691462794165e-05,
+      "loss": 7.3396,
+      "step": 3331
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5413684128213841,
+      "learning_rate": 3.2932496968724415e-05,
+      "loss": 0.6417,
+      "step": 3332
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.39794254057566664,
+      "learning_rate": 3.292807822487428e-05,
+      "loss": 0.6697,
+      "step": 3333
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4990551113356324,
+      "learning_rate": 3.292365839676182e-05,
+      "loss": 0.6985,
+      "step": 3334
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4579351581543523,
+      "learning_rate": 3.291923748475776e-05,
+      "loss": 0.7083,
+      "step": 3335
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4381800956773415,
+      "learning_rate": 3.291481548923283e-05,
+      "loss": 0.6775,
+      "step": 3336
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 13.047119504984334,
+      "learning_rate": 3.291039241055794e-05,
+      "loss": 0.7289,
+      "step": 3337
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4057175529725346,
+      "learning_rate": 3.290596824910403e-05,
+      "loss": 0.7177,
+      "step": 3338
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4283783418289371,
+      "learning_rate": 3.290154300524214e-05,
+      "loss": 0.7195,
+      "step": 3339
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.46756230681142336,
+      "learning_rate": 3.289711667934343e-05,
+      "loss": 0.677,
+      "step": 3340
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.508675465233441,
+      "learning_rate": 3.289268927177911e-05,
+      "loss": 0.7184,
+      "step": 3341
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.398332074663459,
+      "learning_rate": 3.2888260782920514e-05,
+      "loss": 0.6952,
+      "step": 3342
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6796346866168832,
+      "learning_rate": 3.288383121313905e-05,
+      "loss": 0.7126,
+      "step": 3343
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8054386608248651,
+      "learning_rate": 3.2879400562806226e-05,
+      "loss": 0.6899,
+      "step": 3344
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.43503469453578886,
+      "learning_rate": 3.287496883229363e-05,
+      "loss": 0.6401,
+      "step": 3345
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5236992570886224,
+      "learning_rate": 3.287053602197295e-05,
+      "loss": 0.7199,
+      "step": 3346
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.9811049453492421,
+      "learning_rate": 3.286610213221596e-05,
+      "loss": 0.7518,
+      "step": 3347
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5118244247948733,
+      "learning_rate": 3.2861667163394506e-05,
+      "loss": 0.6817,
+      "step": 3348
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.42605905597911115,
+      "learning_rate": 3.285723111588058e-05,
+      "loss": 0.7122,
+      "step": 3349
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.47885965742161346,
+      "learning_rate": 3.28527939900462e-05,
+      "loss": 0.7374,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4334731877866253,
+      "learning_rate": 3.28483557862635e-05,
+      "loss": 0.6872,
+      "step": 3351
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.45620166864696093,
+      "learning_rate": 3.284391650490473e-05,
+      "loss": 0.6772,
+      "step": 3352
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.49318854128624184,
+      "learning_rate": 3.283947614634219e-05,
+      "loss": 0.7068,
+      "step": 3353
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5353516718075316,
+      "learning_rate": 3.2835034710948295e-05,
+      "loss": 0.7013,
+      "step": 3354
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40128572403540863,
+      "learning_rate": 3.283059219909553e-05,
+      "loss": 0.7018,
+      "step": 3355
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4635344151530903,
+      "learning_rate": 3.28261486111565e-05,
+      "loss": 0.7109,
+      "step": 3356
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.43185535866291413,
+      "learning_rate": 3.282170394750388e-05,
+      "loss": 0.6917,
+      "step": 3357
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.459682867364674,
+      "learning_rate": 3.281725820851043e-05,
+      "loss": 0.7164,
+      "step": 3358
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4568120783086786,
+      "learning_rate": 3.281281139454901e-05,
+      "loss": 0.6823,
+      "step": 3359
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4114499454432751,
+      "learning_rate": 3.2808363505992574e-05,
+      "loss": 0.6743,
+      "step": 3360
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.43661779017278507,
+      "learning_rate": 3.2803914543214167e-05,
+      "loss": 0.6969,
+      "step": 3361
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4436640970122818,
+      "learning_rate": 3.27994645065869e-05,
+      "loss": 0.7275,
+      "step": 3362
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.45948896474547296,
+      "learning_rate": 3.279501339648401e-05,
+      "loss": 0.7041,
+      "step": 3363
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3909720687312892,
+      "learning_rate": 3.2790561213278794e-05,
+      "loss": 0.6447,
+      "step": 3364
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3852898563263604,
+      "learning_rate": 3.278610795734466e-05,
+      "loss": 0.6589,
+      "step": 3365
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4163510219487547,
+      "learning_rate": 3.278165362905509e-05,
+      "loss": 0.7353,
+      "step": 3366
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3869302324185589,
+      "learning_rate": 3.277719822878367e-05,
+      "loss": 0.6859,
+      "step": 3367
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.43784312206779835,
+      "learning_rate": 3.277274175690406e-05,
+      "loss": 0.6587,
+      "step": 3368
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4617301569736102,
+      "learning_rate": 3.276828421379003e-05,
+      "loss": 0.7294,
+      "step": 3369
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.46620770834001346,
+      "learning_rate": 3.276382559981542e-05,
+      "loss": 0.6946,
+      "step": 3370
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4720044714142762,
+      "learning_rate": 3.275936591535418e-05,
+      "loss": 0.7046,
+      "step": 3371
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3736191595838042,
+      "learning_rate": 3.275490516078032e-05,
+      "loss": 0.6818,
+      "step": 3372
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.36868848095241635,
+      "learning_rate": 3.275044333646797e-05,
+      "loss": 0.7099,
+      "step": 3373
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.42135461831461657,
+      "learning_rate": 3.274598044279134e-05,
+      "loss": 0.6996,
+      "step": 3374
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.432028011777095,
+      "learning_rate": 3.274151648012471e-05,
+      "loss": 0.7261,
+      "step": 3375
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40959779210354497,
+      "learning_rate": 3.2737051448842496e-05,
+      "loss": 0.672,
+      "step": 3376
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4276010337946351,
+      "learning_rate": 3.273258534931916e-05,
+      "loss": 0.6976,
+      "step": 3377
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4005876130049651,
+      "learning_rate": 3.272811818192926e-05,
+      "loss": 0.6804,
+      "step": 3378
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.44863977982279674,
+      "learning_rate": 3.272364994704746e-05,
+      "loss": 0.6656,
+      "step": 3379
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.45534301737167265,
+      "learning_rate": 3.271918064504851e-05,
+      "loss": 0.7391,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.47053316474376455,
+      "learning_rate": 3.271471027630724e-05,
+      "loss": 0.6812,
+      "step": 3381
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4075214577416628,
+      "learning_rate": 3.2710238841198565e-05,
+      "loss": 0.7051,
+      "step": 3382
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4595999105253186,
+      "learning_rate": 3.270576634009752e-05,
+      "loss": 0.6972,
+      "step": 3383
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3961351534781274,
+      "learning_rate": 3.270129277337919e-05,
+      "loss": 0.6411,
+      "step": 3384
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3952617854858447,
+      "learning_rate": 3.269681814141878e-05,
+      "loss": 0.7105,
+      "step": 3385
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.44802727261221537,
+      "learning_rate": 3.269234244459156e-05,
+      "loss": 0.693,
+      "step": 3386
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.42253036353447093,
+      "learning_rate": 3.268786568327291e-05,
+      "loss": 0.7401,
+      "step": 3387
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40866944114915427,
+      "learning_rate": 3.268338785783829e-05,
+      "loss": 0.6881,
+      "step": 3388
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.39231125962053026,
+      "learning_rate": 3.267890896866325e-05,
+      "loss": 0.6558,
+      "step": 3389
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3989970609691178,
+      "learning_rate": 3.267442901612342e-05,
+      "loss": 0.6774,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.37755903347109737,
+      "learning_rate": 3.2669948000594544e-05,
+      "loss": 0.695,
+      "step": 3391
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3717400614753837,
+      "learning_rate": 3.266546592245243e-05,
+      "loss": 0.6322,
+      "step": 3392
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4183351565933458,
+      "learning_rate": 3.2660982782072976e-05,
+      "loss": 0.6926,
+      "step": 3393
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.42348639341187266,
+      "learning_rate": 3.265649857983219e-05,
+      "loss": 0.654,
+      "step": 3394
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.43288313565948117,
+      "learning_rate": 3.265201331610616e-05,
+      "loss": 0.6905,
+      "step": 3395
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40410165287406313,
+      "learning_rate": 3.2647526991271044e-05,
+      "loss": 0.6703,
+      "step": 3396
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4038609367292325,
+      "learning_rate": 3.2643039605703116e-05,
+      "loss": 0.6831,
+      "step": 3397
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 43.68434257265063,
+      "learning_rate": 3.263855115977873e-05,
+      "loss": 7.6792,
+      "step": 3398
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.433598158565937,
+      "learning_rate": 3.263406165387432e-05,
+      "loss": 0.6506,
+      "step": 3399
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5059241955501159,
+      "learning_rate": 3.262957108836641e-05,
+      "loss": 0.7284,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4715221536136489,
+      "learning_rate": 3.262507946363163e-05,
+      "loss": 0.6551,
+      "step": 3401
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.754792035309904,
+      "learning_rate": 3.2620586780046694e-05,
+      "loss": 0.6584,
+      "step": 3402
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6898611935055594,
+      "learning_rate": 3.261609303798837e-05,
+      "loss": 0.6799,
+      "step": 3403
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.1326334847035613,
+      "learning_rate": 3.261159823783357e-05,
+      "loss": 0.7259,
+      "step": 3404
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.122463439589082,
+      "learning_rate": 3.260710237995925e-05,
+      "loss": 0.7387,
+      "step": 3405
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9009884537287078,
+      "learning_rate": 3.260260546474248e-05,
+      "loss": 0.7252,
+      "step": 3406
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.5986035724754686,
+      "learning_rate": 3.2598107492560414e-05,
+      "loss": 7.9903,
+      "step": 3407
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5750270050496191,
+      "learning_rate": 3.259360846379028e-05,
+      "loss": 0.7203,
+      "step": 3408
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4499273844159964,
+      "learning_rate": 3.258910837880942e-05,
+      "loss": 0.6876,
+      "step": 3409
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5029444672565047,
+      "learning_rate": 3.2584607237995244e-05,
+      "loss": 0.6751,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 9.071816983139753,
+      "learning_rate": 3.2580105041725247e-05,
+      "loss": 0.7237,
+      "step": 3411
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.0306511214406293,
+      "learning_rate": 3.2575601790377044e-05,
+      "loss": 0.6972,
+      "step": 3412
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6730528649000869,
+      "learning_rate": 3.2571097484328305e-05,
+      "loss": 0.7428,
+      "step": 3413
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.1956520040483338,
+      "learning_rate": 3.25665921239568e-05,
+      "loss": 0.7212,
+      "step": 3414
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8928069171793668,
+      "learning_rate": 3.256208570964039e-05,
+      "loss": 0.7197,
+      "step": 3415
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8593395681758451,
+      "learning_rate": 3.255757824175703e-05,
+      "loss": 0.7362,
+      "step": 3416
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.526892322543331,
+      "learning_rate": 3.2553069720684736e-05,
+      "loss": 0.743,
+      "step": 3417
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5204567754193093,
+      "learning_rate": 3.254856014680165e-05,
+      "loss": 0.6743,
+      "step": 3418
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.47219837505479906,
+      "learning_rate": 3.2544049520485984e-05,
+      "loss": 0.6696,
+      "step": 3419
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5885208627225343,
+      "learning_rate": 3.2539537842116026e-05,
+      "loss": 0.7303,
+      "step": 3420
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4372007058847126,
+      "learning_rate": 3.253502511207018e-05,
+      "loss": 0.6586,
+      "step": 3421
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4182084409751857,
+      "learning_rate": 3.253051133072691e-05,
+      "loss": 0.6654,
+      "step": 3422
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.40912261163896674,
+      "learning_rate": 3.25259964984648e-05,
+      "loss": 0.7279,
+      "step": 3423
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.43551250764647215,
+      "learning_rate": 3.252148061566248e-05,
+      "loss": 0.6621,
+      "step": 3424
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.47758476354683943,
+      "learning_rate": 3.25169636826987e-05,
+      "loss": 0.696,
+      "step": 3425
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.46855754937583216,
+      "learning_rate": 3.2512445699952304e-05,
+      "loss": 0.7386,
+      "step": 3426
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.46447387437349225,
+      "learning_rate": 3.2507926667802184e-05,
+      "loss": 0.6838,
+      "step": 3427
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.47937833384725465,
+      "learning_rate": 3.250340658662737e-05,
+      "loss": 0.7056,
+      "step": 3428
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.543890539586936,
+      "learning_rate": 3.249888545680695e-05,
+      "loss": 7.8246,
+      "step": 3429
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4173370637954535,
+      "learning_rate": 3.249436327872009e-05,
+      "loss": 0.6599,
+      "step": 3430
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4706602810707094,
+      "learning_rate": 3.2489840052746074e-05,
+      "loss": 0.7527,
+      "step": 3431
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4893785695246159,
+      "learning_rate": 3.248531577926427e-05,
+      "loss": 0.6812,
+      "step": 3432
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4525324701467162,
+      "learning_rate": 3.24807904586541e-05,
+      "loss": 0.6644,
+      "step": 3433
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.46354977117316015,
+      "learning_rate": 3.24762640912951e-05,
+      "loss": 0.7537,
+      "step": 3434
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.38975427052487865,
+      "learning_rate": 3.247173667756691e-05,
+      "loss": 0.6713,
+      "step": 3435
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3656521038153689,
+      "learning_rate": 3.246720821784922e-05,
+      "loss": 0.6784,
+      "step": 3436
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.48153489363633306,
+      "learning_rate": 3.246267871252184e-05,
+      "loss": 0.6834,
+      "step": 3437
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.39330199439404717,
+      "learning_rate": 3.245814816196464e-05,
+      "loss": 0.6605,
+      "step": 3438
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3402546379130407,
+      "learning_rate": 3.2453616566557605e-05,
+      "loss": 0.6728,
+      "step": 3439
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4046168748644486,
+      "learning_rate": 3.2449083926680784e-05,
+      "loss": 0.7091,
+      "step": 3440
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3576374821376085,
+      "learning_rate": 3.2444550242714324e-05,
+      "loss": 0.6849,
+      "step": 3441
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3857237221280106,
+      "learning_rate": 3.2440015515038474e-05,
+      "loss": 0.6781,
+      "step": 3442
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4024539459324827,
+      "learning_rate": 3.2435479744033545e-05,
+      "loss": 0.6818,
+      "step": 3443
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.40402198565436703,
+      "learning_rate": 3.243094293007995e-05,
+      "loss": 0.6752,
+      "step": 3444
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.40266276758699276,
+      "learning_rate": 3.2426405073558176e-05,
+      "loss": 0.7451,
+      "step": 3445
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6317232241677818,
+      "learning_rate": 3.2421866174848826e-05,
+      "loss": 0.7261,
+      "step": 3446
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4477150280384154,
+      "learning_rate": 3.2417326234332556e-05,
+      "loss": 0.7067,
+      "step": 3447
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3665171946888892,
+      "learning_rate": 3.241278525239013e-05,
+      "loss": 0.7123,
+      "step": 3448
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.41255225058966893,
+      "learning_rate": 3.24082432294024e-05,
+      "loss": 0.7029,
+      "step": 3449
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3730320303649979,
+      "learning_rate": 3.240370016575029e-05,
+      "loss": 0.6758,
+      "step": 3450
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3545931596585988,
+      "learning_rate": 3.239915606181483e-05,
+      "loss": 0.6921,
+      "step": 3451
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3815356101259525,
+      "learning_rate": 3.2394610917977124e-05,
+      "loss": 0.7071,
+      "step": 3452
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3784471952212804,
+      "learning_rate": 3.239006473461837e-05,
+      "loss": 0.717,
+      "step": 3453
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4552467240724608,
+      "learning_rate": 3.238551751211986e-05,
+      "loss": 0.7417,
+      "step": 3454
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4790182749182709,
+      "learning_rate": 3.238096925086294e-05,
+      "loss": 0.695,
+      "step": 3455
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4969041695523207,
+      "learning_rate": 3.237641995122908e-05,
+      "loss": 0.7156,
+      "step": 3456
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4619055031835884,
+      "learning_rate": 3.237186961359983e-05,
+      "loss": 0.7429,
+      "step": 3457
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.42838915886839385,
+      "learning_rate": 3.2367318238356827e-05,
+      "loss": 0.707,
+      "step": 3458
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3789467090952308,
+      "learning_rate": 3.236276582588177e-05,
+      "loss": 0.653,
+      "step": 3459
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.39771839138444975,
+      "learning_rate": 3.235821237655647e-05,
+      "loss": 0.716,
+      "step": 3460
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3866423999176203,
+      "learning_rate": 3.235365789076283e-05,
+      "loss": 0.6629,
+      "step": 3461
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.45017319367888686,
+      "learning_rate": 3.234910236888282e-05,
+      "loss": 0.7005,
+      "step": 3462
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3799719630641762,
+      "learning_rate": 3.234454581129851e-05,
+      "loss": 0.6712,
+      "step": 3463
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.409953226672179,
+      "learning_rate": 3.233998821839205e-05,
+      "loss": 0.6474,
+      "step": 3464
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.39999085700519554,
+      "learning_rate": 3.2335429590545673e-05,
+      "loss": 0.7416,
+      "step": 3465
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4115404289883185,
+      "learning_rate": 3.2330869928141726e-05,
+      "loss": 0.7081,
+      "step": 3466
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4402251385900859,
+      "learning_rate": 3.2326309231562606e-05,
+      "loss": 0.7141,
+      "step": 3467
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.42158564539390603,
+      "learning_rate": 3.232174750119081e-05,
+      "loss": 0.6631,
+      "step": 3468
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.38302568269772136,
+      "learning_rate": 3.2317184737408946e-05,
+      "loss": 0.689,
+      "step": 3469
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4042497632635781,
+      "learning_rate": 3.2312620940599663e-05,
+      "loss": 0.6599,
+      "step": 3470
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.433629835797824,
+      "learning_rate": 3.2308056111145735e-05,
+      "loss": 0.7336,
+      "step": 3471
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.47766206021657853,
+      "learning_rate": 3.2303490249430006e-05,
+      "loss": 0.7549,
+      "step": 3472
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3885396011458876,
+      "learning_rate": 3.22989233558354e-05,
+      "loss": 0.647,
+      "step": 3473
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3767561687760372,
+      "learning_rate": 3.2294355430744955e-05,
+      "loss": 0.6733,
+      "step": 3474
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.39302195506638915,
+      "learning_rate": 3.228978647454177e-05,
+      "loss": 0.6747,
+      "step": 3475
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4145915059035423,
+      "learning_rate": 3.228521648760903e-05,
+      "loss": 0.6911,
+      "step": 3476
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4380266008352994,
+      "learning_rate": 3.228064547033002e-05,
+      "loss": 0.7263,
+      "step": 3477
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4022322795424343,
+      "learning_rate": 3.2276073423088106e-05,
+      "loss": 0.6723,
+      "step": 3478
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3781621064900686,
+      "learning_rate": 3.227150034626674e-05,
+      "loss": 0.6557,
+      "step": 3479
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5766581283815799,
+      "learning_rate": 3.226692624024946e-05,
+      "loss": 0.7333,
+      "step": 3480
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3876658773389152,
+      "learning_rate": 3.2262351105419895e-05,
+      "loss": 0.6858,
+      "step": 3481
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3721593420588353,
+      "learning_rate": 3.225777494216175e-05,
+      "loss": 0.7232,
+      "step": 3482
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3774059612136674,
+      "learning_rate": 3.225319775085882e-05,
+      "loss": 0.6818,
+      "step": 3483
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.34886536465342016,
+      "learning_rate": 3.2248619531895e-05,
+      "loss": 0.6553,
+      "step": 3484
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3931578355540503,
+      "learning_rate": 3.2244040285654246e-05,
+      "loss": 0.6602,
+      "step": 3485
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.46101488912814337,
+      "learning_rate": 3.223946001252062e-05,
+      "loss": 0.6917,
+      "step": 3486
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.439549881788191,
+      "learning_rate": 3.223487871287827e-05,
+      "loss": 0.7311,
+      "step": 3487
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.429207188142648,
+      "learning_rate": 3.2230296387111415e-05,
+      "loss": 0.6581,
+      "step": 3488
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.43986375290576935,
+      "learning_rate": 3.222571303560437e-05,
+      "loss": 0.7004,
+      "step": 3489
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.36437003800460593,
+      "learning_rate": 3.222112865874154e-05,
+      "loss": 0.6998,
+      "step": 3490
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4482995315728772,
+      "learning_rate": 3.221654325690741e-05,
+      "loss": 0.7032,
+      "step": 3491
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.38215699778866047,
+      "learning_rate": 3.221195683048655e-05,
+      "loss": 0.6711,
+      "step": 3492
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3752084261810108,
+      "learning_rate": 3.2207369379863615e-05,
+      "loss": 0.6613,
+      "step": 3493
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 3.592820582116873,
+      "learning_rate": 3.220278090542336e-05,
+      "loss": 7.8654,
+      "step": 3494
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.37293156908309966,
+      "learning_rate": 3.21981914075506e-05,
+      "loss": 0.6799,
+      "step": 3495
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3740117769168215,
+      "learning_rate": 3.219360088663026e-05,
+      "loss": 0.657,
+      "step": 3496
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3892828458487245,
+      "learning_rate": 3.2189009343047345e-05,
+      "loss": 0.6702,
+      "step": 3497
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4003675797960291,
+      "learning_rate": 3.2184416777186936e-05,
+      "loss": 0.6984,
+      "step": 3498
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.41862893857706274,
+      "learning_rate": 3.2179823189434205e-05,
+      "loss": 0.691,
+      "step": 3499
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.38600181186821997,
+      "learning_rate": 3.217522858017442e-05,
+      "loss": 0.709,
+      "step": 3500
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.39238998630420663,
+      "learning_rate": 3.2170632949792905e-05,
+      "loss": 0.6942,
+      "step": 3501
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4150741357536465,
+      "learning_rate": 3.2166036298675115e-05,
+      "loss": 0.6888,
+      "step": 3502
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.39242333007484753,
+      "learning_rate": 3.216143862720654e-05,
+      "loss": 0.6712,
+      "step": 3503
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.43230114444153145,
+      "learning_rate": 3.2156839935772805e-05,
+      "loss": 0.664,
+      "step": 3504
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3971164255604865,
+      "learning_rate": 3.2152240224759586e-05,
+      "loss": 0.6926,
+      "step": 3505
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3685321172123293,
+      "learning_rate": 3.214763949455265e-05,
+      "loss": 0.7151,
+      "step": 3506
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3886083292119125,
+      "learning_rate": 3.2143037745537856e-05,
+      "loss": 0.6653,
+      "step": 3507
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.41296791443979103,
+      "learning_rate": 3.213843497810117e-05,
+      "loss": 0.6574,
+      "step": 3508
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4228785781251861,
+      "learning_rate": 3.2133831192628586e-05,
+      "loss": 0.6876,
+      "step": 3509
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5586422968218164,
+      "learning_rate": 3.212922638950624e-05,
+      "loss": 0.6782,
+      "step": 3510
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4020716400626449,
+      "learning_rate": 3.212462056912032e-05,
+      "loss": 0.7071,
+      "step": 3511
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3895938141448023,
+      "learning_rate": 3.212001373185712e-05,
+      "loss": 0.656,
+      "step": 3512
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.43792885126641007,
+      "learning_rate": 3.2115405878103005e-05,
+      "loss": 0.7333,
+      "step": 3513
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 4.537008315135866,
+      "learning_rate": 3.211079700824443e-05,
+      "loss": 7.5965,
+      "step": 3514
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4228931558041017,
+      "learning_rate": 3.2106187122667934e-05,
+      "loss": 0.7232,
+      "step": 3515
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4032517095243693,
+      "learning_rate": 3.210157622176015e-05,
+      "loss": 0.6669,
+      "step": 3516
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.0308600395662768,
+      "learning_rate": 3.2096964305907774e-05,
+      "loss": 0.7006,
+      "step": 3517
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.40629408848136805,
+      "learning_rate": 3.2092351375497624e-05,
+      "loss": 0.6459,
+      "step": 3518
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.41057120473966474,
+      "learning_rate": 3.208773743091656e-05,
+      "loss": 0.6882,
+      "step": 3519
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3723782955508978,
+      "learning_rate": 3.2083122472551555e-05,
+      "loss": 0.6795,
+      "step": 3520
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4126299306937764,
+      "learning_rate": 3.207850650078967e-05,
+      "loss": 0.6881,
+      "step": 3521
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.43315311766486275,
+      "learning_rate": 3.207388951601802e-05,
+      "loss": 0.6636,
+      "step": 3522
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.41842686829282844,
+      "learning_rate": 3.206927151862385e-05,
+      "loss": 0.6778,
+      "step": 3523
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.41358155595130147,
+      "learning_rate": 3.206465250899444e-05,
+      "loss": 0.6419,
+      "step": 3524
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4750036823580102,
+      "learning_rate": 3.20600324875172e-05,
+      "loss": 0.6916,
+      "step": 3525
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4279102293237525,
+      "learning_rate": 3.2055411454579606e-05,
+      "loss": 0.7143,
+      "step": 3526
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.7939392156064358,
+      "learning_rate": 3.205078941056921e-05,
+      "loss": 7.7362,
+      "step": 3527
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4191760060392699,
+      "learning_rate": 3.204616635587365e-05,
+      "loss": 0.6659,
+      "step": 3528
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4135849596345436,
+      "learning_rate": 3.204154229088067e-05,
+      "loss": 0.695,
+      "step": 3529
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.34821892317802056,
+      "learning_rate": 3.203691721597809e-05,
+      "loss": 0.6881,
+      "step": 3530
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.38082197397327594,
+      "learning_rate": 3.203229113155379e-05,
+      "loss": 0.6704,
+      "step": 3531
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40470131381668334,
+      "learning_rate": 3.202766403799577e-05,
+      "loss": 0.6814,
+      "step": 3532
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3857569361835805,
+      "learning_rate": 3.2023035935692086e-05,
+      "loss": 0.6829,
+      "step": 3533
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.423965973816473,
+      "learning_rate": 3.201840682503091e-05,
+      "loss": 0.673,
+      "step": 3534
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4199226094343778,
+      "learning_rate": 3.2013776706400456e-05,
+      "loss": 0.702,
+      "step": 3535
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.43063588638100714,
+      "learning_rate": 3.200914558018907e-05,
+      "loss": 0.6765,
+      "step": 3536
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.41216934137608774,
+      "learning_rate": 3.2004513446785143e-05,
+      "loss": 0.683,
+      "step": 3537
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4569040222315553,
+      "learning_rate": 3.199988030657716e-05,
+      "loss": 0.7403,
+      "step": 3538
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.45065977569920407,
+      "learning_rate": 3.199524615995372e-05,
+      "loss": 0.6459,
+      "step": 3539
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40557153214874986,
+      "learning_rate": 3.199061100730347e-05,
+      "loss": 0.684,
+      "step": 3540
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.43305788544914897,
+      "learning_rate": 3.198597484901516e-05,
+      "loss": 0.6652,
+      "step": 3541
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.377463658172263,
+      "learning_rate": 3.19813376854776e-05,
+      "loss": 0.6702,
+      "step": 3542
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.43616497556551126,
+      "learning_rate": 3.197669951707973e-05,
+      "loss": 0.6676,
+      "step": 3543
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.37426946796126404,
+      "learning_rate": 3.1972060344210536e-05,
+      "loss": 0.685,
+      "step": 3544
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3969874788075403,
+      "learning_rate": 3.196742016725909e-05,
+      "loss": 0.6822,
+      "step": 3545
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.41104121672075267,
+      "learning_rate": 3.1962778986614576e-05,
+      "loss": 0.6636,
+      "step": 3546
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4263931438949608,
+      "learning_rate": 3.195813680266624e-05,
+      "loss": 0.7246,
+      "step": 3547
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.45833025305794434,
+      "learning_rate": 3.19534936158034e-05,
+      "loss": 0.6929,
+      "step": 3548
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4434681488659824,
+      "learning_rate": 3.19488494264155e-05,
+      "loss": 0.711,
+      "step": 3549
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3928809437591957,
+      "learning_rate": 3.194420423489202e-05,
+      "loss": 0.6685,
+      "step": 3550
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4283964200378458,
+      "learning_rate": 3.193955804162256e-05,
+      "loss": 0.7187,
+      "step": 3551
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3968705143136308,
+      "learning_rate": 3.1934910846996774e-05,
+      "loss": 0.6639,
+      "step": 3552
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3897555965689157,
+      "learning_rate": 3.1930262651404445e-05,
+      "loss": 0.6699,
+      "step": 3553
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4422838599956842,
+      "learning_rate": 3.192561345523539e-05,
+      "loss": 0.681,
+      "step": 3554
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4412845906420469,
+      "learning_rate": 3.1920963258879534e-05,
+      "loss": 0.7128,
+      "step": 3555
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.44964936328522737,
+      "learning_rate": 3.1916312062726894e-05,
+      "loss": 0.6991,
+      "step": 3556
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40158643682591666,
+      "learning_rate": 3.1911659867167546e-05,
+      "loss": 0.6723,
+      "step": 3557
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6423816454604029,
+      "learning_rate": 3.190700667259167e-05,
+      "loss": 0.6966,
+      "step": 3558
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.38336615130327223,
+      "learning_rate": 3.190235247938953e-05,
+      "loss": 0.6533,
+      "step": 3559
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3997312631318247,
+      "learning_rate": 3.1897697287951456e-05,
+      "loss": 0.6859,
+      "step": 3560
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.37682075262255693,
+      "learning_rate": 3.189304109866788e-05,
+      "loss": 0.6158,
+      "step": 3561
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4441841308420618,
+      "learning_rate": 3.188838391192931e-05,
+      "loss": 0.6838,
+      "step": 3562
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.39979208773153996,
+      "learning_rate": 3.188372572812634e-05,
+      "loss": 0.7071,
+      "step": 3563
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.36589808138578195,
+      "learning_rate": 3.1879066547649645e-05,
+      "loss": 0.6645,
+      "step": 3564
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.43108207142233,
+      "learning_rate": 3.187440637088999e-05,
+      "loss": 0.7165,
+      "step": 3565
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.3901243463412982,
+      "learning_rate": 3.18697451982382e-05,
+      "loss": 7.5696,
+      "step": 3566
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3936361198203024,
+      "learning_rate": 3.186508303008523e-05,
+      "loss": 0.6977,
+      "step": 3567
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.37850627675161963,
+      "learning_rate": 3.186041986682207e-05,
+      "loss": 0.7051,
+      "step": 3568
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3913785339116885,
+      "learning_rate": 3.185575570883982e-05,
+      "loss": 0.6626,
+      "step": 3569
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4396087744388906,
+      "learning_rate": 3.185109055652966e-05,
+      "loss": 0.6991,
+      "step": 3570
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.43431992247405654,
+      "learning_rate": 3.184642441028284e-05,
+      "loss": 0.6989,
+      "step": 3571
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4025861531750412,
+      "learning_rate": 3.1841757270490715e-05,
+      "loss": 0.6921,
+      "step": 3572
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40734323674872536,
+      "learning_rate": 3.183708913754471e-05,
+      "loss": 0.6867,
+      "step": 3573
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40782226820741824,
+      "learning_rate": 3.183242001183633e-05,
+      "loss": 0.6674,
+      "step": 3574
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3870934088634292,
+      "learning_rate": 3.1827749893757176e-05,
+      "loss": 0.6482,
+      "step": 3575
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.39992311934927566,
+      "learning_rate": 3.182307878369892e-05,
+      "loss": 0.702,
+      "step": 3576
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4444251930396615,
+      "learning_rate": 3.1818406682053335e-05,
+      "loss": 0.7017,
+      "step": 3577
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3661425352752172,
+      "learning_rate": 3.181373358921225e-05,
+      "loss": 0.7185,
+      "step": 3578
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.36530888387946203,
+      "learning_rate": 3.180905950556759e-05,
+      "loss": 0.672,
+      "step": 3579
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.38499449760995186,
+      "learning_rate": 3.180438443151138e-05,
+      "loss": 0.6812,
+      "step": 3580
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40712356170993264,
+      "learning_rate": 3.1799708367435705e-05,
+      "loss": 0.6401,
+      "step": 3581
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4491777791811982,
+      "learning_rate": 3.179503131373273e-05,
+      "loss": 0.6682,
+      "step": 3582
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.47654202125541595,
+      "learning_rate": 3.179035327079473e-05,
+      "loss": 0.7338,
+      "step": 3583
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3760289885450845,
+      "learning_rate": 3.1785674239014034e-05,
+      "loss": 0.6609,
+      "step": 3584
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4620722742177812,
+      "learning_rate": 3.1780994218783084e-05,
+      "loss": 0.6992,
+      "step": 3585
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4248942709110434,
+      "learning_rate": 3.177631321049437e-05,
+      "loss": 0.6953,
+      "step": 3586
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.42139711802961144,
+      "learning_rate": 3.177163121454048e-05,
+      "loss": 0.735,
+      "step": 3587
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5147393342727629,
+      "learning_rate": 3.1766948231314114e-05,
+      "loss": 7.4637,
+      "step": 3588
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40397183147746074,
+      "learning_rate": 3.1762264261208e-05,
+      "loss": 0.6969,
+      "step": 3589
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.44841915592808873,
+      "learning_rate": 3.1757579304614994e-05,
+      "loss": 0.7144,
+      "step": 3590
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4053068241992543,
+      "learning_rate": 3.1752893361928e-05,
+      "loss": 0.6655,
+      "step": 3591
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3825111231409221,
+      "learning_rate": 3.174820643354004e-05,
+      "loss": 0.6718,
+      "step": 3592
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3940958941973277,
+      "learning_rate": 3.17435185198442e-05,
+      "loss": 0.6997,
+      "step": 3593
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3815548219443808,
+      "learning_rate": 3.173882962123364e-05,
+      "loss": 0.6948,
+      "step": 3594
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.42922032561038653,
+      "learning_rate": 3.1734139738101615e-05,
+      "loss": 0.6744,
+      "step": 3595
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3751584804850926,
+      "learning_rate": 3.1729448870841464e-05,
+      "loss": 0.7267,
+      "step": 3596
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.41509865188624645,
+      "learning_rate": 3.1724757019846606e-05,
+      "loss": 0.682,
+      "step": 3597
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.41037578700158045,
+      "learning_rate": 3.1720064185510525e-05,
+      "loss": 0.695,
+      "step": 3598
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4203982443094395,
+      "learning_rate": 3.171537036822682e-05,
+      "loss": 0.7132,
+      "step": 3599
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4057813781225989,
+      "learning_rate": 3.171067556838915e-05,
+      "loss": 0.7076,
+      "step": 3600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4161840480289687,
+      "learning_rate": 3.170597978639127e-05,
+      "loss": 0.7341,
+      "step": 3601
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4211557025288301,
+      "learning_rate": 3.1701283022626995e-05,
+      "loss": 0.7097,
+      "step": 3602
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.41938297662563206,
+      "learning_rate": 3.169658527749025e-05,
+      "loss": 0.7139,
+      "step": 3603
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.37229060977321754,
+      "learning_rate": 3.169188655137502e-05,
+      "loss": 0.6755,
+      "step": 3604
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4749288528751265,
+      "learning_rate": 3.168718684467539e-05,
+      "loss": 0.7458,
+      "step": 3605
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3537897484849401,
+      "learning_rate": 3.168248615778551e-05,
+      "loss": 0.6612,
+      "step": 3606
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.44614310254574246,
+      "learning_rate": 3.167778449109962e-05,
+      "loss": 0.6901,
+      "step": 3607
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.43522121955739546,
+      "learning_rate": 3.167308184501206e-05,
+      "loss": 0.6979,
+      "step": 3608
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4008370981590401,
+      "learning_rate": 3.1668378219917214e-05,
+      "loss": 0.6871,
+      "step": 3609
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.41428627652018213,
+      "learning_rate": 3.1663673616209584e-05,
+      "loss": 0.7077,
+      "step": 3610
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.36365127826541177,
+      "learning_rate": 3.165896803428372e-05,
+      "loss": 0.6932,
+      "step": 3611
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.39367664842938194,
+      "learning_rate": 3.1654261474534304e-05,
+      "loss": 0.6852,
+      "step": 3612
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4243477250703443,
+      "learning_rate": 3.164955393735605e-05,
+      "loss": 0.663,
+      "step": 3613
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.45098756641958704,
+      "learning_rate": 3.164484542314377e-05,
+      "loss": 0.7051,
+      "step": 3614
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.34493511816900724,
+      "learning_rate": 3.1640135932292375e-05,
+      "loss": 0.6619,
+      "step": 3615
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3774438934857586,
+      "learning_rate": 3.1635425465196824e-05,
+      "loss": 0.6609,
+      "step": 3616
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.37175255095166726,
+      "learning_rate": 3.1630714022252206e-05,
+      "loss": 0.7042,
+      "step": 3617
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3872729412540659,
+      "learning_rate": 3.1626001603853645e-05,
+      "loss": 0.6896,
+      "step": 3618
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.38733634598089556,
+      "learning_rate": 3.1621288210396366e-05,
+      "loss": 0.6841,
+      "step": 3619
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.38860015924008806,
+      "learning_rate": 3.161657384227568e-05,
+      "loss": 0.6824,
+      "step": 3620
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.48026967506972196,
+      "learning_rate": 3.161185849988698e-05,
+      "loss": 0.689,
+      "step": 3621
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40922094706238527,
+      "learning_rate": 3.160714218362573e-05,
+      "loss": 0.6878,
+      "step": 3622
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4325806481458068,
+      "learning_rate": 3.160242489388748e-05,
+      "loss": 0.7225,
+      "step": 3623
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4072944742955121,
+      "learning_rate": 3.1597706631067864e-05,
+      "loss": 0.7065,
+      "step": 3624
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3580322061316145,
+      "learning_rate": 3.159298739556261e-05,
+      "loss": 0.6318,
+      "step": 3625
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.42769703394507247,
+      "learning_rate": 3.15882671877675e-05,
+      "loss": 0.6584,
+      "step": 3626
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4443978115892922,
+      "learning_rate": 3.158354600807841e-05,
+      "loss": 0.6796,
+      "step": 3627
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40892407499986655,
+      "learning_rate": 3.1578823856891306e-05,
+      "loss": 0.7107,
+      "step": 3628
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40431465163484,
+      "learning_rate": 3.157410073460223e-05,
+      "loss": 0.6686,
+      "step": 3629
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40895994727792184,
+      "learning_rate": 3.156937664160731e-05,
+      "loss": 0.6958,
+      "step": 3630
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3780335842973326,
+      "learning_rate": 3.156465157830273e-05,
+      "loss": 0.7204,
+      "step": 3631
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3771641993221167,
+      "learning_rate": 3.155992554508481e-05,
+      "loss": 0.6743,
+      "step": 3632
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.41160534930996096,
+      "learning_rate": 3.155519854234988e-05,
+      "loss": 0.6855,
+      "step": 3633
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4376036233065397,
+      "learning_rate": 3.155047057049441e-05,
+      "loss": 0.6638,
+      "step": 3634
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3761774932876888,
+      "learning_rate": 3.154574162991491e-05,
+      "loss": 0.662,
+      "step": 3635
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.39756791624262094,
+      "learning_rate": 3.154101172100801e-05,
+      "loss": 0.6913,
+      "step": 3636
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.37256732762315503,
+      "learning_rate": 3.15362808441704e-05,
+      "loss": 0.6597,
+      "step": 3637
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.44659350201293274,
+      "learning_rate": 3.153154899979884e-05,
+      "loss": 0.6743,
+      "step": 3638
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3951918186988703,
+      "learning_rate": 3.152681618829021e-05,
+      "loss": 0.7412,
+      "step": 3639
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3935343113738017,
+      "learning_rate": 3.152208241004141e-05,
+      "loss": 0.7227,
+      "step": 3640
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3665644749307235,
+      "learning_rate": 3.151734766544948e-05,
+      "loss": 0.663,
+      "step": 3641
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3647670925260735,
+      "learning_rate": 3.151261195491151e-05,
+      "loss": 0.6925,
+      "step": 3642
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8108271641576205,
+      "learning_rate": 3.1507875278824674e-05,
+      "loss": 0.6722,
+      "step": 3643
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4522974396174534,
+      "learning_rate": 3.150313763758624e-05,
+      "loss": 0.721,
+      "step": 3644
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.39093297710928,
+      "learning_rate": 3.1498399031593537e-05,
+      "loss": 0.7239,
+      "step": 3645
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3618396982005158,
+      "learning_rate": 3.1493659461244e-05,
+      "loss": 0.713,
+      "step": 3646
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3944633218112719,
+      "learning_rate": 3.148891892693513e-05,
+      "loss": 0.7243,
+      "step": 3647
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3980126269065166,
+      "learning_rate": 3.1484177429064495e-05,
+      "loss": 0.7066,
+      "step": 3648
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.37020386824715396,
+      "learning_rate": 3.1479434968029776e-05,
+      "loss": 0.6981,
+      "step": 3649
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4060065517230576,
+      "learning_rate": 3.147469154422871e-05,
+      "loss": 0.6982,
+      "step": 3650
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3729852611894034,
+      "learning_rate": 3.146994715805912e-05,
+      "loss": 0.6864,
+      "step": 3651
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4562752236533333,
+      "learning_rate": 3.146520180991891e-05,
+      "loss": 0.6569,
+      "step": 3652
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3977523247688886,
+      "learning_rate": 3.146045550020608e-05,
+      "loss": 0.6894,
+      "step": 3653
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5037898747231241,
+      "learning_rate": 3.145570822931868e-05,
+      "loss": 0.7605,
+      "step": 3654
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3874936934635398,
+      "learning_rate": 3.145095999765487e-05,
+      "loss": 0.6849,
+      "step": 3655
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.36809993066883573,
+      "learning_rate": 3.144621080561287e-05,
+      "loss": 0.7183,
+      "step": 3656
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.40848644220367875,
+      "learning_rate": 3.1441460653590996e-05,
+      "loss": 0.6826,
+      "step": 3657
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3998482241811529,
+      "learning_rate": 3.143670954198763e-05,
+      "loss": 0.7552,
+      "step": 3658
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.43060042819115957,
+      "learning_rate": 3.143195747120125e-05,
+      "loss": 0.7142,
+      "step": 3659
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.40675399069436974,
+      "learning_rate": 3.142720444163041e-05,
+      "loss": 0.6671,
+      "step": 3660
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4073776270296788,
+      "learning_rate": 3.142245045367373e-05,
+      "loss": 0.6744,
+      "step": 3661
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.43187678592997086,
+      "learning_rate": 3.141769550772993e-05,
+      "loss": 0.6787,
+      "step": 3662
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3638053765018059,
+      "learning_rate": 3.141293960419778e-05,
+      "loss": 0.6778,
+      "step": 3663
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.41178474838410395,
+      "learning_rate": 3.140818274347618e-05,
+      "loss": 0.7034,
+      "step": 3664
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3750430650535957,
+      "learning_rate": 3.1403424925964076e-05,
+      "loss": 0.6537,
+      "step": 3665
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.38677001853392395,
+      "learning_rate": 3.139866615206049e-05,
+      "loss": 0.6838,
+      "step": 3666
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.39660874290354,
+      "learning_rate": 3.1393906422164535e-05,
+      "loss": 0.6873,
+      "step": 3667
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3828099160056501,
+      "learning_rate": 3.1389145736675414e-05,
+      "loss": 0.6742,
+      "step": 3668
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.42506489520214286,
+      "learning_rate": 3.138438409599239e-05,
+      "loss": 0.6691,
+      "step": 3669
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3753627522173836,
+      "learning_rate": 3.137962150051483e-05,
+      "loss": 0.6825,
+      "step": 3670
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.43394725824237795,
+      "learning_rate": 3.137485795064215e-05,
+      "loss": 0.708,
+      "step": 3671
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.40179248053885164,
+      "learning_rate": 3.137009344677388e-05,
+      "loss": 0.6948,
+      "step": 3672
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4263547901975787,
+      "learning_rate": 3.136532798930959e-05,
+      "loss": 0.6985,
+      "step": 3673
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.31910620764532427,
+      "learning_rate": 3.136056157864898e-05,
+      "loss": 0.631,
+      "step": 3674
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.44740370060893103,
+      "learning_rate": 3.135579421519178e-05,
+      "loss": 0.6902,
+      "step": 3675
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4250219327978557,
+      "learning_rate": 3.135102589933784e-05,
+      "loss": 0.671,
+      "step": 3676
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.47060041425083143,
+      "learning_rate": 3.1346256631487065e-05,
+      "loss": 0.7103,
+      "step": 3677
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4514467723226947,
+      "learning_rate": 3.134148641203945e-05,
+      "loss": 0.6695,
+      "step": 3678
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4291218639280644,
+      "learning_rate": 3.133671524139507e-05,
+      "loss": 0.6829,
+      "step": 3679
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4056947265448923,
+      "learning_rate": 3.133194311995407e-05,
+      "loss": 0.6755,
+      "step": 3680
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.41711318468128106,
+      "learning_rate": 3.132717004811669e-05,
+      "loss": 0.6713,
+      "step": 3681
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.45097028704613046,
+      "learning_rate": 3.132239602628323e-05,
+      "loss": 0.7317,
+      "step": 3682
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0256478572078542,
+      "learning_rate": 3.1317621054854095e-05,
+      "loss": 0.7216,
+      "step": 3683
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3963963183230602,
+      "learning_rate": 3.1312845134229744e-05,
+      "loss": 0.7102,
+      "step": 3684
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.36534982949127065,
+      "learning_rate": 3.130806826481074e-05,
+      "loss": 0.6726,
+      "step": 3685
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0591410106001837,
+      "learning_rate": 3.130329044699771e-05,
+      "loss": 7.3835,
+      "step": 3686
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4407236196660112,
+      "learning_rate": 3.129851168119135e-05,
+      "loss": 0.7107,
+      "step": 3687
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9582242916806867,
+      "learning_rate": 3.1293731967792476e-05,
+      "loss": 7.3978,
+      "step": 3688
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.46571917255073214,
+      "learning_rate": 3.1288951307201934e-05,
+      "loss": 0.7139,
+      "step": 3689
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.43649681329074996,
+      "learning_rate": 3.128416969982068e-05,
+      "loss": 0.6956,
+      "step": 3690
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3967804082545036,
+      "learning_rate": 3.127938714604973e-05,
+      "loss": 0.6637,
+      "step": 3691
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.438839476413246,
+      "learning_rate": 3.127460364629021e-05,
+      "loss": 0.7111,
+      "step": 3692
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3640289640878451,
+      "learning_rate": 3.126981920094331e-05,
+      "loss": 0.6825,
+      "step": 3693
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.40657038459636896,
+      "learning_rate": 3.126503381041027e-05,
+      "loss": 0.6835,
+      "step": 3694
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4072432481810613,
+      "learning_rate": 3.126024747509245e-05,
+      "loss": 0.6846,
+      "step": 3695
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.42087241734263464,
+      "learning_rate": 3.125546019539128e-05,
+      "loss": 0.6943,
+      "step": 3696
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.43190764413539895,
+      "learning_rate": 3.125067197170825e-05,
+      "loss": 0.6578,
+      "step": 3697
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.41265019636582506,
+      "learning_rate": 3.124588280444495e-05,
+      "loss": 0.663,
+      "step": 3698
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4208655370297958,
+      "learning_rate": 3.124109269400305e-05,
+      "loss": 0.6841,
+      "step": 3699
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4727476052826699,
+      "learning_rate": 3.123630164078427e-05,
+      "loss": 0.6751,
+      "step": 3700
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.37762795329242266,
+      "learning_rate": 3.123150964519045e-05,
+      "loss": 0.615,
+      "step": 3701
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.42249690076257235,
+      "learning_rate": 3.1226716707623475e-05,
+      "loss": 0.7157,
+      "step": 3702
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4059101526321291,
+      "learning_rate": 3.122192282848533e-05,
+      "loss": 0.6948,
+      "step": 3703
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3781969521957994,
+      "learning_rate": 3.121712800817807e-05,
+      "loss": 0.6784,
+      "step": 3704
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4335286510345025,
+      "learning_rate": 3.1212332247103824e-05,
+      "loss": 0.7028,
+      "step": 3705
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.32118924676620314,
+      "learning_rate": 3.120753554566482e-05,
+      "loss": 0.6541,
+      "step": 3706
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3518689805300693,
+      "learning_rate": 3.1202737904263345e-05,
+      "loss": 0.6539,
+      "step": 3707
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4091567370761371,
+      "learning_rate": 3.119793932330177e-05,
+      "loss": 0.7305,
+      "step": 3708
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3510583074717147,
+      "learning_rate": 3.119313980318255e-05,
+      "loss": 0.6466,
+      "step": 3709
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.352464259271521,
+      "learning_rate": 3.1188339344308215e-05,
+      "loss": 0.6147,
+      "step": 3710
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3795810367151487,
+      "learning_rate": 3.118353794708137e-05,
+      "loss": 0.6546,
+      "step": 3711
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3520539962936865,
+      "learning_rate": 3.11787356119047e-05,
+      "loss": 0.6422,
+      "step": 3712
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.39797588180058635,
+      "learning_rate": 3.117393233918097e-05,
+      "loss": 0.6864,
+      "step": 3713
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4350623682257194,
+      "learning_rate": 3.116912812931304e-05,
+      "loss": 0.6833,
+      "step": 3714
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 4.870109131301757,
+      "learning_rate": 3.1164322982703816e-05,
+      "loss": 7.5117,
+      "step": 3715
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.388379720113741,
+      "learning_rate": 3.115951689975631e-05,
+      "loss": 0.6887,
+      "step": 3716
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4028039327271029,
+      "learning_rate": 3.11547098808736e-05,
+      "loss": 0.7019,
+      "step": 3717
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3722493277474058,
+      "learning_rate": 3.114990192645884e-05,
+      "loss": 0.6779,
+      "step": 3718
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.429109271021515,
+      "learning_rate": 3.114509303691527e-05,
+      "loss": 0.7323,
+      "step": 3719
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.41107050996657607,
+      "learning_rate": 3.114028321264621e-05,
+      "loss": 0.6883,
+      "step": 3720
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4384020151435536,
+      "learning_rate": 3.113547245405505e-05,
+      "loss": 0.7295,
+      "step": 3721
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3440940034165995,
+      "learning_rate": 3.113066076154527e-05,
+      "loss": 0.6475,
+      "step": 3722
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5130812505660282,
+      "learning_rate": 3.1125848135520404e-05,
+      "loss": 0.7175,
+      "step": 3723
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4909830963883784,
+      "learning_rate": 3.1121034576384096e-05,
+      "loss": 0.6772,
+      "step": 3724
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3878052765098295,
+      "learning_rate": 3.111622008454005e-05,
+      "loss": 0.6407,
+      "step": 3725
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.35436583529032284,
+      "learning_rate": 3.111140466039205e-05,
+      "loss": 0.6578,
+      "step": 3726
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3636661252275432,
+      "learning_rate": 3.110658830434396e-05,
+      "loss": 0.6736,
+      "step": 3727
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3970639727534583,
+      "learning_rate": 3.1101771016799714e-05,
+      "loss": 0.698,
+      "step": 3728
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3781042842682548,
+      "learning_rate": 3.1096952798163354e-05,
+      "loss": 0.6558,
+      "step": 3729
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.43344765274938246,
+      "learning_rate": 3.109213364883896e-05,
+      "loss": 0.6871,
+      "step": 3730
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4336949134827448,
+      "learning_rate": 3.108731356923071e-05,
+      "loss": 0.7009,
+      "step": 3731
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.234469282773947,
+      "learning_rate": 3.108249255974286e-05,
+      "loss": 7.3764,
+      "step": 3732
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4419105881414959,
+      "learning_rate": 3.107767062077975e-05,
+      "loss": 0.733,
+      "step": 3733
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.697703013944658,
+      "learning_rate": 3.107284775274577e-05,
+      "loss": 7.3453,
+      "step": 3734
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.42210382618457337,
+      "learning_rate": 3.1068023956045435e-05,
+      "loss": 0.6895,
+      "step": 3735
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3518003210350136,
+      "learning_rate": 3.1063199231083295e-05,
+      "loss": 0.6265,
+      "step": 3736
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.43805294476147455,
+      "learning_rate": 3.105837357826399e-05,
+      "loss": 0.6582,
+      "step": 3737
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.44485404793764094,
+      "learning_rate": 3.105354699799226e-05,
+      "loss": 0.7048,
+      "step": 3738
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4713009960264513,
+      "learning_rate": 3.104871949067288e-05,
+      "loss": 0.733,
+      "step": 3739
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.40873371241566747,
+      "learning_rate": 3.104389105671075e-05,
+      "loss": 0.6637,
+      "step": 3740
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1591363699472614,
+      "learning_rate": 3.1039061696510814e-05,
+      "loss": 7.4125,
+      "step": 3741
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.43822974226276046,
+      "learning_rate": 3.1034231410478095e-05,
+      "loss": 0.657,
+      "step": 3742
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.44476959206738353,
+      "learning_rate": 3.102940019901772e-05,
+      "loss": 0.637,
+      "step": 3743
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4081383599224253,
+      "learning_rate": 3.102456806253488e-05,
+      "loss": 0.7139,
+      "step": 3744
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3968042047588391,
+      "learning_rate": 3.101973500143482e-05,
+      "loss": 0.6743,
+      "step": 3745
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.37017653463265066,
+      "learning_rate": 3.101490101612289e-05,
+      "loss": 0.6799,
+      "step": 3746
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4144281582810889,
+      "learning_rate": 3.1010066107004524e-05,
+      "loss": 0.7344,
+      "step": 3747
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.43111763849448465,
+      "learning_rate": 3.10052302744852e-05,
+      "loss": 0.6931,
+      "step": 3748
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4399770708116548,
+      "learning_rate": 3.100039351897051e-05,
+      "loss": 0.7111,
+      "step": 3749
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3965895255695004,
+      "learning_rate": 3.09955558408661e-05,
+      "loss": 0.7115,
+      "step": 3750
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4015476357573573,
+      "learning_rate": 3.09907172405777e-05,
+      "loss": 0.6942,
+      "step": 3751
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.40626392928675176,
+      "learning_rate": 3.098587771851112e-05,
+      "loss": 0.7376,
+      "step": 3752
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3768947774198348,
+      "learning_rate": 3.0981037275072244e-05,
+      "loss": 0.6778,
+      "step": 3753
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.31254764903843174,
+      "learning_rate": 3.097619591066703e-05,
+      "loss": 0.6327,
+      "step": 3754
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3817329685589158,
+      "learning_rate": 3.097135362570153e-05,
+      "loss": 0.6601,
+      "step": 3755
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3719580362338664,
+      "learning_rate": 3.096651042058184e-05,
+      "loss": 0.6808,
+      "step": 3756
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.41422957707672015,
+      "learning_rate": 3.096166629571417e-05,
+      "loss": 0.6746,
+      "step": 3757
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.41824662392316564,
+      "learning_rate": 3.095682125150479e-05,
+      "loss": 0.6825,
+      "step": 3758
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.40148250095504234,
+      "learning_rate": 3.095197528836004e-05,
+      "loss": 0.6581,
+      "step": 3759
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4362538114868192,
+      "learning_rate": 3.0947128406686346e-05,
+      "loss": 0.7042,
+      "step": 3760
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4469606868190711,
+      "learning_rate": 3.094228060689022e-05,
+      "loss": 0.7152,
+      "step": 3761
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.47095410531290127,
+      "learning_rate": 3.093743188937823e-05,
+      "loss": 0.7125,
+      "step": 3762
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3395714971539994,
+      "learning_rate": 3.0932582254557036e-05,
+      "loss": 0.6403,
+      "step": 3763
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3361931884834336,
+      "learning_rate": 3.092773170283337e-05,
+      "loss": 0.6855,
+      "step": 3764
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4076035649141612,
+      "learning_rate": 3.092288023461405e-05,
+      "loss": 0.6954,
+      "step": 3765
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4459051153187583,
+      "learning_rate": 3.0918027850305955e-05,
+      "loss": 0.7287,
+      "step": 3766
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3902162596928823,
+      "learning_rate": 3.091317455031605e-05,
+      "loss": 0.6642,
+      "step": 3767
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3971322208709067,
+      "learning_rate": 3.0908320335051375e-05,
+      "loss": 0.7067,
+      "step": 3768
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.42266651961334334,
+      "learning_rate": 3.0903465204919046e-05,
+      "loss": 0.7225,
+      "step": 3769
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.42388250258644417,
+      "learning_rate": 3.089860916032625e-05,
+      "loss": 0.6918,
+      "step": 3770
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3967862928297576,
+      "learning_rate": 3.0893752201680274e-05,
+      "loss": 0.6863,
+      "step": 3771
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4477021588167692,
+      "learning_rate": 3.088889432938846e-05,
+      "loss": 0.6715,
+      "step": 3772
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4165951386561592,
+      "learning_rate": 3.0884035543858226e-05,
+      "loss": 0.6825,
+      "step": 3773
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3439581616351642,
+      "learning_rate": 3.087917584549708e-05,
+      "loss": 0.6453,
+      "step": 3774
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.379739676352572,
+      "learning_rate": 3.0874315234712584e-05,
+      "loss": 0.6805,
+      "step": 3775
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4064723281507436,
+      "learning_rate": 3.086945371191241e-05,
+      "loss": 0.6582,
+      "step": 3776
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.359712110306628,
+      "learning_rate": 3.086459127750428e-05,
+      "loss": 0.6731,
+      "step": 3777
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.362354578337748,
+      "learning_rate": 3.0859727931896e-05,
+      "loss": 0.6556,
+      "step": 3778
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3586956637115038,
+      "learning_rate": 3.0854863675495454e-05,
+      "loss": 0.6893,
+      "step": 3779
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.209458021278887,
+      "learning_rate": 3.084999850871059e-05,
+      "loss": 7.532,
+      "step": 3780
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4180374220465269,
+      "learning_rate": 3.084513243194947e-05,
+      "loss": 0.6834,
+      "step": 3781
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.40445067086961645,
+      "learning_rate": 3.0840265445620186e-05,
+      "loss": 0.7236,
+      "step": 3782
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4269261235963448,
+      "learning_rate": 3.083539755013093e-05,
+      "loss": 0.7353,
+      "step": 3783
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.47177094905129996,
+      "learning_rate": 3.083052874588996e-05,
+      "loss": 0.6901,
+      "step": 3784
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3783634515313233,
+      "learning_rate": 3.0825659033305633e-05,
+      "loss": 0.6878,
+      "step": 3785
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.35622922173536903,
+      "learning_rate": 3.082078841278636e-05,
+      "loss": 0.7096,
+      "step": 3786
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3754961968199897,
+      "learning_rate": 3.0815916884740626e-05,
+      "loss": 0.6689,
+      "step": 3787
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3624823323574984,
+      "learning_rate": 3.0811044449577e-05,
+      "loss": 0.6767,
+      "step": 3788
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.48653358369719923,
+      "learning_rate": 3.080617110770414e-05,
+      "loss": 0.6588,
+      "step": 3789
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.34664403418521267,
+      "learning_rate": 3.080129685953075e-05,
+      "loss": 0.638,
+      "step": 3790
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3564389735449296,
+      "learning_rate": 3.0796421705465654e-05,
+      "loss": 0.6817,
+      "step": 3791
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.38886647042273664,
+      "learning_rate": 3.079154564591769e-05,
+      "loss": 0.6964,
+      "step": 3792
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7888419267696767,
+      "learning_rate": 3.078666868129584e-05,
+      "loss": 7.4757,
+      "step": 3793
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4125408577033184,
+      "learning_rate": 3.078179081200911e-05,
+      "loss": 0.696,
+      "step": 3794
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4065725870057624,
+      "learning_rate": 3.0776912038466605e-05,
+      "loss": 0.6614,
+      "step": 3795
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3944031254099989,
+      "learning_rate": 3.07720323610775e-05,
+      "loss": 0.6863,
+      "step": 3796
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.44505685015768953,
+      "learning_rate": 3.076715178025105e-05,
+      "loss": 0.6987,
+      "step": 3797
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3507101443125289,
+      "learning_rate": 3.076227029639659e-05,
+      "loss": 0.7039,
+      "step": 3798
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.391448564311815,
+      "learning_rate": 3.0757387909923506e-05,
+      "loss": 0.6662,
+      "step": 3799
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4200134399648791,
+      "learning_rate": 3.0752504621241304e-05,
+      "loss": 0.671,
+      "step": 3800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4091321070807136,
+      "learning_rate": 3.074762043075952e-05,
+      "loss": 0.7175,
+      "step": 3801
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.36190310079787585,
+      "learning_rate": 3.074273533888779e-05,
+      "loss": 0.6652,
+      "step": 3802
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3738668039560617,
+      "learning_rate": 3.073784934603581e-05,
+      "loss": 0.6531,
+      "step": 3803
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.38654546167292375,
+      "learning_rate": 3.0732962452613385e-05,
+      "loss": 0.6732,
+      "step": 3804
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.40875454765993097,
+      "learning_rate": 3.072807465903036e-05,
+      "loss": 0.7066,
+      "step": 3805
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.43760227994151013,
+      "learning_rate": 3.072318596569667e-05,
+      "loss": 0.7053,
+      "step": 3806
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.39788502861541575,
+      "learning_rate": 3.071829637302233e-05,
+      "loss": 0.7129,
+      "step": 3807
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.41700780568026347,
+      "learning_rate": 3.07134058814174e-05,
+      "loss": 0.7063,
+      "step": 3808
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7103169827414029,
+      "learning_rate": 3.070851449129207e-05,
+      "loss": 7.3706,
+      "step": 3809
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.42300218049731464,
+      "learning_rate": 3.070362220305656e-05,
+      "loss": 0.6903,
+      "step": 3810
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3992330911970842,
+      "learning_rate": 3.0698729017121176e-05,
+      "loss": 0.6911,
+      "step": 3811
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4091298519967557,
+      "learning_rate": 3.0693834933896326e-05,
+      "loss": 0.6703,
+      "step": 3812
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.43658854936415803,
+      "learning_rate": 3.068893995379244e-05,
+      "loss": 0.6985,
+      "step": 3813
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3684050138669257,
+      "learning_rate": 3.068404407722008e-05,
+      "loss": 0.6917,
+      "step": 3814
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.44565095871559074,
+      "learning_rate": 3.0679147304589833e-05,
+      "loss": 0.7385,
+      "step": 3815
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3764822963545976,
+      "learning_rate": 3.0674249636312406e-05,
+      "loss": 0.6598,
+      "step": 3816
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4111623947828609,
+      "learning_rate": 3.066935107279855e-05,
+      "loss": 0.7005,
+      "step": 3817
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.39470503385178113,
+      "learning_rate": 3.0664451614459117e-05,
+      "loss": 0.7503,
+      "step": 3818
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.40434492255275967,
+      "learning_rate": 3.065955126170499e-05,
+      "loss": 0.7059,
+      "step": 3819
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3952113738727594,
+      "learning_rate": 3.0654650014947183e-05,
+      "loss": 0.7053,
+      "step": 3820
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3569501770378708,
+      "learning_rate": 3.064974787459675e-05,
+      "loss": 0.6561,
+      "step": 3821
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3940453185133683,
+      "learning_rate": 3.0644844841064816e-05,
+      "loss": 0.7059,
+      "step": 3822
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3961650996549398,
+      "learning_rate": 3.06399409147626e-05,
+      "loss": 0.6922,
+      "step": 3823
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.35991224274922234,
+      "learning_rate": 3.0635036096101396e-05,
+      "loss": 0.6948,
+      "step": 3824
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3729894188346732,
+      "learning_rate": 3.0630130385492554e-05,
+      "loss": 0.6716,
+      "step": 3825
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.348774010540366,
+      "learning_rate": 3.062522378334751e-05,
+      "loss": 0.6481,
+      "step": 3826
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3912480219293857,
+      "learning_rate": 3.062031629007779e-05,
+      "loss": 0.663,
+      "step": 3827
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.36560197500722247,
+      "learning_rate": 3.061540790609496e-05,
+      "loss": 0.662,
+      "step": 3828
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4230654073784643,
+      "learning_rate": 3.0610498631810694e-05,
+      "loss": 0.7021,
+      "step": 3829
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9205888356346094,
+      "learning_rate": 3.0605588467636723e-05,
+      "loss": 7.2453,
+      "step": 3830
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4043609235008984,
+      "learning_rate": 3.060067741398485e-05,
+      "loss": 0.6976,
+      "step": 3831
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4164113954176934,
+      "learning_rate": 3.059576547126697e-05,
+      "loss": 0.6958,
+      "step": 3832
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.39310318262233873,
+      "learning_rate": 3.059085263989503e-05,
+      "loss": 0.6757,
+      "step": 3833
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.360501331765083,
+      "learning_rate": 3.0585938920281075e-05,
+      "loss": 0.6463,
+      "step": 3834
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.34449464935614,
+      "learning_rate": 3.05810243128372e-05,
+      "loss": 0.6521,
+      "step": 3835
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4214863746168573,
+      "learning_rate": 3.0576108817975606e-05,
+      "loss": 0.6824,
+      "step": 3836
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4329096997161028,
+      "learning_rate": 3.0571192436108534e-05,
+      "loss": 0.7035,
+      "step": 3837
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3884508714880295,
+      "learning_rate": 3.0566275167648316e-05,
+      "loss": 0.6947,
+      "step": 3838
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.36896728375085824,
+      "learning_rate": 3.056135701300736e-05,
+      "loss": 0.6632,
+      "step": 3839
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3860153403685722,
+      "learning_rate": 3.055643797259815e-05,
+      "loss": 0.6506,
+      "step": 3840
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.39147297548639326,
+      "learning_rate": 3.0551518046833234e-05,
+      "loss": 0.6948,
+      "step": 3841
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4433102008960158,
+      "learning_rate": 3.054659723612525e-05,
+      "loss": 0.7027,
+      "step": 3842
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3931063696703823,
+      "learning_rate": 3.054167554088688e-05,
+      "loss": 0.6725,
+      "step": 3843
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.40238925828065547,
+      "learning_rate": 3.053675296153092e-05,
+      "loss": 0.7055,
+      "step": 3844
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.38491965620365587,
+      "learning_rate": 3.053182949847022e-05,
+      "loss": 0.6761,
+      "step": 3845
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4178368103165302,
+      "learning_rate": 3.0526905152117686e-05,
+      "loss": 0.6713,
+      "step": 3846
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3837636198193575,
+      "learning_rate": 3.052197992288634e-05,
+      "loss": 0.639,
+      "step": 3847
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.842345768431351,
+      "learning_rate": 3.0517053811189245e-05,
+      "loss": 7.2935,
+      "step": 3848
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.37968982384494404,
+      "learning_rate": 3.0512126817439544e-05,
+      "loss": 0.6808,
+      "step": 3849
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.45290115372266787,
+      "learning_rate": 3.0507198942050463e-05,
+      "loss": 0.68,
+      "step": 3850
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.38531793009400633,
+      "learning_rate": 3.0502270185435298e-05,
+      "loss": 0.6702,
+      "step": 3851
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3981150366528139,
+      "learning_rate": 3.0497340548007416e-05,
+      "loss": 0.7184,
+      "step": 3852
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.42206584861340735,
+      "learning_rate": 3.0492410030180255e-05,
+      "loss": 0.6811,
+      "step": 3853
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.379678909684792,
+      "learning_rate": 3.048747863236734e-05,
+      "loss": 0.6883,
+      "step": 3854
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.40022115051724333,
+      "learning_rate": 3.048254635498226e-05,
+      "loss": 0.6923,
+      "step": 3855
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.37711717533004163,
+      "learning_rate": 3.0477613198438674e-05,
+      "loss": 0.6828,
+      "step": 3856
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.42106963828508176,
+      "learning_rate": 3.0472679163150322e-05,
+      "loss": 0.7381,
+      "step": 3857
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3882296849574938,
+      "learning_rate": 3.0467744249531024e-05,
+      "loss": 0.6873,
+      "step": 3858
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.375964877709505,
+      "learning_rate": 3.046280845799465e-05,
+      "loss": 0.7337,
+      "step": 3859
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4135906656843911,
+      "learning_rate": 3.0457871788955163e-05,
+      "loss": 0.653,
+      "step": 3860
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.41293107782872435,
+      "learning_rate": 3.045293424282661e-05,
+      "loss": 0.679,
+      "step": 3861
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.38199587683996644,
+      "learning_rate": 3.0447995820023083e-05,
+      "loss": 0.7083,
+      "step": 3862
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.355860185383416,
+      "learning_rate": 3.0443056520958754e-05,
+      "loss": 0.6056,
+      "step": 3863
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.37250688689193545,
+      "learning_rate": 3.0438116346047897e-05,
+      "loss": 0.688,
+      "step": 3864
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3972496137175383,
+      "learning_rate": 3.043317529570483e-05,
+      "loss": 0.672,
+      "step": 3865
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.39693803919554854,
+      "learning_rate": 3.042823337034394e-05,
+      "loss": 0.7128,
+      "step": 3866
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.31201558751539565,
+      "learning_rate": 3.0423290570379715e-05,
+      "loss": 0.6943,
+      "step": 3867
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4489957100124244,
+      "learning_rate": 3.041834689622671e-05,
+      "loss": 0.7285,
+      "step": 3868
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.44077444095481105,
+      "learning_rate": 3.0413402348299522e-05,
+      "loss": 0.6812,
+      "step": 3869
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.36928627103874856,
+      "learning_rate": 3.040845692701285e-05,
+      "loss": 0.6818,
+      "step": 3870
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.39750955295267953,
+      "learning_rate": 3.0403510632781478e-05,
+      "loss": 0.7377,
+      "step": 3871
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5313161273131113,
+      "learning_rate": 3.039856346602023e-05,
+      "loss": 0.7152,
+      "step": 3872
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7439847471851521,
+      "learning_rate": 3.0393615427144022e-05,
+      "loss": 7.2553,
+      "step": 3873
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.38423678433015,
+      "learning_rate": 3.0388666516567837e-05,
+      "loss": 0.7351,
+      "step": 3874
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3814356090140891,
+      "learning_rate": 3.0383716734706738e-05,
+      "loss": 0.6647,
+      "step": 3875
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3575045902866496,
+      "learning_rate": 3.037876608197586e-05,
+      "loss": 0.6527,
+      "step": 3876
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4218033779364644,
+      "learning_rate": 3.0373814558790406e-05,
+      "loss": 0.6936,
+      "step": 3877
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.383659431541281,
+      "learning_rate": 3.036886216556565e-05,
+      "loss": 0.7001,
+      "step": 3878
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3839807776766466,
+      "learning_rate": 3.036390890271695e-05,
+      "loss": 0.7117,
+      "step": 3879
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5065800810049669,
+      "learning_rate": 3.035895477065972e-05,
+      "loss": 0.6547,
+      "step": 3880
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4345780121343155,
+      "learning_rate": 3.035399976980947e-05,
+      "loss": 0.6893,
+      "step": 3881
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.38471071754819697,
+      "learning_rate": 3.034904390058176e-05,
+      "loss": 0.6852,
+      "step": 3882
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.43762124683406756,
+      "learning_rate": 3.0344087163392236e-05,
+      "loss": 0.6816,
+      "step": 3883
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3799886054269724,
+      "learning_rate": 3.0339129558656618e-05,
+      "loss": 0.6596,
+      "step": 3884
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.43780037858035986,
+      "learning_rate": 3.0334171086790695e-05,
+      "loss": 0.6602,
+      "step": 3885
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.36194933826991976,
+      "learning_rate": 3.0329211748210316e-05,
+      "loss": 0.6534,
+      "step": 3886
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3675018993722933,
+      "learning_rate": 3.032425154333142e-05,
+      "loss": 0.6707,
+      "step": 3887
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3790891688157346,
+      "learning_rate": 3.031929047257002e-05,
+      "loss": 0.645,
+      "step": 3888
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.37427960057063236,
+      "learning_rate": 3.0314328536342193e-05,
+      "loss": 0.6746,
+      "step": 3889
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4443542151680328,
+      "learning_rate": 3.0309365735064085e-05,
+      "loss": 0.7519,
+      "step": 3890
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3719376541257786,
+      "learning_rate": 3.0304402069151925e-05,
+      "loss": 0.676,
+      "step": 3891
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.34527128633803905,
+      "learning_rate": 3.0299437539022007e-05,
+      "loss": 0.6752,
+      "step": 3892
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3597018609474588,
+      "learning_rate": 3.0294472145090696e-05,
+      "loss": 0.6657,
+      "step": 3893
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3601486392561925,
+      "learning_rate": 3.0289505887774445e-05,
+      "loss": 0.6689,
+      "step": 3894
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.40073115441283097,
+      "learning_rate": 3.0284538767489762e-05,
+      "loss": 0.6639,
+      "step": 3895
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3615663587302887,
+      "learning_rate": 3.027957078465323e-05,
+      "loss": 0.6926,
+      "step": 3896
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.42946897644319376,
+      "learning_rate": 3.0274601939681515e-05,
+      "loss": 0.7016,
+      "step": 3897
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3985124240519354,
+      "learning_rate": 3.026963223299134e-05,
+      "loss": 0.6824,
+      "step": 3898
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.42248753597528205,
+      "learning_rate": 3.0264661664999514e-05,
+      "loss": 0.7066,
+      "step": 3899
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3808702741656933,
+      "learning_rate": 3.025969023612291e-05,
+      "loss": 0.6984,
+      "step": 3900
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.41511733093114833,
+      "learning_rate": 3.0254717946778473e-05,
+      "loss": 0.7072,
+      "step": 3901
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.41129813332271364,
+      "learning_rate": 3.0249744797383233e-05,
+      "loss": 0.6918,
+      "step": 3902
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4567952758013475,
+      "learning_rate": 3.0244770788354272e-05,
+      "loss": 0.7648,
+      "step": 3903
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.48411262138180916,
+      "learning_rate": 3.0239795920108758e-05,
+      "loss": 0.6823,
+      "step": 3904
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3891678886741784,
+      "learning_rate": 3.0234820193063927e-05,
+      "loss": 0.6798,
+      "step": 3905
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3783096262984369,
+      "learning_rate": 3.022984360763709e-05,
+      "loss": 0.6251,
+      "step": 3906
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.451225786104712,
+      "learning_rate": 3.0224866164245617e-05,
+      "loss": 0.6919,
+      "step": 3907
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.34980900140386956,
+      "learning_rate": 3.021988786330697e-05,
+      "loss": 0.642,
+      "step": 3908
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4112882611292104,
+      "learning_rate": 3.0214908705238676e-05,
+      "loss": 0.7022,
+      "step": 3909
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.435421375903405,
+      "learning_rate": 3.0209928690458318e-05,
+      "loss": 0.6797,
+      "step": 3910
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4097375493979916,
+      "learning_rate": 3.020494781938357e-05,
+      "loss": 0.6706,
+      "step": 3911
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3744700254663998,
+      "learning_rate": 3.0199966092432178e-05,
+      "loss": 0.6671,
+      "step": 3912
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3749379566615629,
+      "learning_rate": 3.019498351002195e-05,
+      "loss": 0.6644,
+      "step": 3913
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.43549109232398475,
+      "learning_rate": 3.019000007257076e-05,
+      "loss": 0.6909,
+      "step": 3914
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.38231795560928417,
+      "learning_rate": 3.018501578049658e-05,
+      "loss": 0.6566,
+      "step": 3915
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4522411415430469,
+      "learning_rate": 3.018003063421742e-05,
+      "loss": 0.6982,
+      "step": 3916
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4016316564318009,
+      "learning_rate": 3.0175044634151388e-05,
+      "loss": 0.6865,
+      "step": 3917
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4044114698772273,
+      "learning_rate": 3.0170057780716647e-05,
+      "loss": 0.6575,
+      "step": 3918
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3917567330163461,
+      "learning_rate": 3.016507007433145e-05,
+      "loss": 0.6341,
+      "step": 3919
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4283217586108054,
+      "learning_rate": 3.016008151541409e-05,
+      "loss": 0.6915,
+      "step": 3920
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3629417218923544,
+      "learning_rate": 3.015509210438298e-05,
+      "loss": 0.6701,
+      "step": 3921
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3691431561200845,
+      "learning_rate": 3.015010184165655e-05,
+      "loss": 0.663,
+      "step": 3922
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4386372463235991,
+      "learning_rate": 3.0145110727653334e-05,
+      "loss": 0.7342,
+      "step": 3923
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3699554055211854,
+      "learning_rate": 3.0140118762791935e-05,
+      "loss": 0.6666,
+      "step": 3924
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3638772561356654,
+      "learning_rate": 3.013512594749102e-05,
+      "loss": 0.6803,
+      "step": 3925
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3686621038487634,
+      "learning_rate": 3.0130132282169337e-05,
+      "loss": 0.6739,
+      "step": 3926
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3855327994484907,
+      "learning_rate": 3.012513776724569e-05,
+      "loss": 0.6897,
+      "step": 3927
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.44701073651537326,
+      "learning_rate": 3.0120142403138973e-05,
+      "loss": 0.7091,
+      "step": 3928
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4295267346956802,
+      "learning_rate": 3.0115146190268128e-05,
+      "loss": 0.7448,
+      "step": 3929
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.40901805185371015,
+      "learning_rate": 3.011014912905219e-05,
+      "loss": 0.7069,
+      "step": 3930
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.39109492523015027,
+      "learning_rate": 3.0105151219910255e-05,
+      "loss": 0.7207,
+      "step": 3931
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4588727458250715,
+      "learning_rate": 3.0100152463261488e-05,
+      "loss": 0.7486,
+      "step": 3932
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1444549301022526,
+      "learning_rate": 3.0095152859525133e-05,
+      "loss": 7.4483,
+      "step": 3933
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3924552484459659,
+      "learning_rate": 3.0090152409120502e-05,
+      "loss": 0.654,
+      "step": 3934
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3641317271479691,
+      "learning_rate": 3.0085151112466975e-05,
+      "loss": 0.6672,
+      "step": 3935
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4174402662507581,
+      "learning_rate": 3.0080148969984e-05,
+      "loss": 0.7118,
+      "step": 3936
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.39136380669393944,
+      "learning_rate": 3.0075145982091106e-05,
+      "loss": 0.7066,
+      "step": 3937
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.41568298739556475,
+      "learning_rate": 3.007014214920789e-05,
+      "loss": 0.6816,
+      "step": 3938
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.43998836965386,
+      "learning_rate": 3.006513747175401e-05,
+      "loss": 0.7501,
+      "step": 3939
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.35488856271491,
+      "learning_rate": 3.0060131950149202e-05,
+      "loss": 0.6801,
+      "step": 3940
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.37808136427026234,
+      "learning_rate": 3.0055125584813286e-05,
+      "loss": 0.6521,
+      "step": 3941
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.336364393648812,
+      "learning_rate": 3.0050118376166124e-05,
+      "loss": 0.6432,
+      "step": 3942
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3839962157486957,
+      "learning_rate": 3.004511032462767e-05,
+      "loss": 0.6922,
+      "step": 3943
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.38694542554424666,
+      "learning_rate": 3.0040101430617947e-05,
+      "loss": 0.6647,
+      "step": 3944
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3827051092588666,
+      "learning_rate": 3.0035091694557047e-05,
+      "loss": 0.6648,
+      "step": 3945
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.41731955287103834,
+      "learning_rate": 3.0030081116865123e-05,
+      "loss": 0.7378,
+      "step": 3946
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.41665453420973403,
+      "learning_rate": 3.0025069697962403e-05,
+      "loss": 0.724,
+      "step": 3947
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.41184394174647837,
+      "learning_rate": 3.00200574382692e-05,
+      "loss": 0.7469,
+      "step": 3948
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9700666384595246,
+      "learning_rate": 3.0015044338205884e-05,
+      "loss": 7.3415,
+      "step": 3949
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4081551540496339,
+      "learning_rate": 3.001003039819289e-05,
+      "loss": 0.6865,
+      "step": 3950
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.37491522692578555,
+      "learning_rate": 3.0005015618650733e-05,
+      "loss": 0.6532,
+      "step": 3951
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.36433505625781115,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.655,
+      "step": 3952
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.42285890613973387,
+      "learning_rate": 2.9994983542661348e-05,
+      "loss": 0.7174,
+      "step": 3953
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4125425889571214,
+      "learning_rate": 2.99899662470555e-05,
+      "loss": 0.7315,
+      "step": 3954
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.40858440272067637,
+      "learning_rate": 2.9984948113603242e-05,
+      "loss": 0.6926,
+      "step": 3955
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4217081233367582,
+      "learning_rate": 2.9979929142725445e-05,
+      "loss": 0.7384,
+      "step": 3956
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.36087088531117495,
+      "learning_rate": 2.9974909334843042e-05,
+      "loss": 0.6514,
+      "step": 3957
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.40858189007077694,
+      "learning_rate": 2.9969888690377043e-05,
+      "loss": 0.6636,
+      "step": 3958
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4006206088051224,
+      "learning_rate": 2.996486720974852e-05,
+      "loss": 0.6918,
+      "step": 3959
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3966270801403823,
+      "learning_rate": 2.9959844893378617e-05,
+      "loss": 0.709,
+      "step": 3960
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.38397403888040194,
+      "learning_rate": 2.9954821741688548e-05,
+      "loss": 0.6853,
+      "step": 3961
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3924081343657384,
+      "learning_rate": 2.994979775509961e-05,
+      "loss": 0.6097,
+      "step": 3962
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6575840119307761,
+      "learning_rate": 2.9944772934033143e-05,
+      "loss": 7.3568,
+      "step": 3963
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.38404088263426356,
+      "learning_rate": 2.9939747278910585e-05,
+      "loss": 0.6886,
+      "step": 3964
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.43386393158049447,
+      "learning_rate": 2.9934720790153426e-05,
+      "loss": 0.7192,
+      "step": 3965
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.37893441703287944,
+      "learning_rate": 2.992969346818323e-05,
+      "loss": 0.6639,
+      "step": 3966
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.41025669215920557,
+      "learning_rate": 2.9924665313421638e-05,
+      "loss": 0.6536,
+      "step": 3967
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.39076082585833194,
+      "learning_rate": 2.9919636326290348e-05,
+      "loss": 0.6335,
+      "step": 3968
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3479338850916584,
+      "learning_rate": 2.9914606507211145e-05,
+      "loss": 0.6546,
+      "step": 3969
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3679628250046246,
+      "learning_rate": 2.9909575856605867e-05,
+      "loss": 0.6827,
+      "step": 3970
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4040277787641461,
+      "learning_rate": 2.990454437489643e-05,
+      "loss": 0.7023,
+      "step": 3971
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4112322511032165,
+      "learning_rate": 2.9899512062504814e-05,
+      "loss": 0.7246,
+      "step": 3972
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.41742150193047084,
+      "learning_rate": 2.9894478919853077e-05,
+      "loss": 0.721,
+      "step": 3973
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4356060240447128,
+      "learning_rate": 2.9889444947363345e-05,
+      "loss": 0.6908,
+      "step": 3974
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.42686587384751346,
+      "learning_rate": 2.9884410145457806e-05,
+      "loss": 0.7069,
+      "step": 3975
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4469779246281456,
+      "learning_rate": 2.987937451455873e-05,
+      "loss": 0.7368,
+      "step": 3976
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3903298609148703,
+      "learning_rate": 2.987433805508844e-05,
+      "loss": 0.6728,
+      "step": 3977
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.374651942094237,
+      "learning_rate": 2.9869300767469346e-05,
+      "loss": 0.6451,
+      "step": 3978
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.38900755638731566,
+      "learning_rate": 2.986426265212391e-05,
+      "loss": 0.6757,
+      "step": 3979
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4272284706844832,
+      "learning_rate": 2.9859223709474688e-05,
+      "loss": 0.6837,
+      "step": 3980
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3717845207452758,
+      "learning_rate": 2.985418393994427e-05,
+      "loss": 0.6574,
+      "step": 3981
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3741595929895558,
+      "learning_rate": 2.9849143343955353e-05,
+      "loss": 0.6808,
+      "step": 3982
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3563187377544669,
+      "learning_rate": 2.984410192193068e-05,
+      "loss": 0.6609,
+      "step": 3983
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.33757631933358967,
+      "learning_rate": 2.9839059674293058e-05,
+      "loss": 0.694,
+      "step": 3984
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.44575220491056616,
+      "learning_rate": 2.9834016601465393e-05,
+      "loss": 0.6957,
+      "step": 3985
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.850007245611128,
+      "learning_rate": 2.9828972703870625e-05,
+      "loss": 7.2138,
+      "step": 3986
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3347809506857184,
+      "learning_rate": 2.98239279819318e-05,
+      "loss": 0.6418,
+      "step": 3987
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4058553215906326,
+      "learning_rate": 2.9818882436071983e-05,
+      "loss": 0.6856,
+      "step": 3988
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4332805031482382,
+      "learning_rate": 2.9813836066714362e-05,
+      "loss": 0.6853,
+      "step": 3989
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.36647000522003226,
+      "learning_rate": 2.9808788874282162e-05,
+      "loss": 0.6841,
+      "step": 3990
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.34238647691363217,
+      "learning_rate": 2.980374085919868e-05,
+      "loss": 0.6447,
+      "step": 3991
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.44549728688493595,
+      "learning_rate": 2.97986920218873e-05,
+      "loss": 0.7621,
+      "step": 3992
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.405417398781649,
+      "learning_rate": 2.9793642362771452e-05,
+      "loss": 0.663,
+      "step": 3993
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.44774351104249926,
+      "learning_rate": 2.978859188227464e-05,
+      "loss": 0.7233,
+      "step": 3994
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.39830558941547617,
+      "learning_rate": 2.9783540580820456e-05,
+      "loss": 0.6461,
+      "step": 3995
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.42452846168363056,
+      "learning_rate": 2.9778488458832537e-05,
+      "loss": 0.7134,
+      "step": 3996
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4434033706380337,
+      "learning_rate": 2.9773435516734602e-05,
+      "loss": 0.6946,
+      "step": 3997
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4050756296366069,
+      "learning_rate": 2.9768381754950426e-05,
+      "loss": 0.721,
+      "step": 3998
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.39568445678648667,
+      "learning_rate": 2.9763327173903874e-05,
+      "loss": 0.7189,
+      "step": 3999
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3967051097276695,
+      "learning_rate": 2.975827177401887e-05,
+      "loss": 0.6664,
+      "step": 4000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4095549819496539,
+      "learning_rate": 2.975321555571939e-05,
+      "loss": 0.6824,
+      "step": 4001
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4068954961002341,
+      "learning_rate": 2.9748158519429497e-05,
+      "loss": 0.7128,
+      "step": 4002
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3626090360367203,
+      "learning_rate": 2.9743100665573333e-05,
+      "loss": 0.6312,
+      "step": 4003
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3835283362199273,
+      "learning_rate": 2.9738041994575076e-05,
+      "loss": 0.672,
+      "step": 4004
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.363114741321125,
+      "learning_rate": 2.9732982506858992e-05,
+      "loss": 0.6697,
+      "step": 4005
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.34640202100120737,
+      "learning_rate": 2.9727922202849426e-05,
+      "loss": 0.6748,
+      "step": 4006
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.40972819751356726,
+      "learning_rate": 2.9722861082970777e-05,
+      "loss": 0.6941,
+      "step": 4007
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.39870172152049543,
+      "learning_rate": 2.9717799147647505e-05,
+      "loss": 0.6844,
+      "step": 4008
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.38579999187577957,
+      "learning_rate": 2.9712736397304158e-05,
+      "loss": 0.6861,
+      "step": 4009
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3846216793086521,
+      "learning_rate": 2.9707672832365335e-05,
+      "loss": 0.6517,
+      "step": 4010
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4071567372258464,
+      "learning_rate": 2.970260845325572e-05,
+      "loss": 0.6653,
+      "step": 4011
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.386469847132039,
+      "learning_rate": 2.9697543260400045e-05,
+      "loss": 0.6606,
+      "step": 4012
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.482892464881127,
+      "learning_rate": 2.9692477254223137e-05,
+      "loss": 0.6882,
+      "step": 4013
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.44484334190335817,
+      "learning_rate": 2.9687410435149865e-05,
+      "loss": 0.7237,
+      "step": 4014
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.37521857876345194,
+      "learning_rate": 2.9682342803605175e-05,
+      "loss": 0.6587,
+      "step": 4015
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.39855878068888906,
+      "learning_rate": 2.9677274360014095e-05,
+      "loss": 0.6777,
+      "step": 4016
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4529017049495381,
+      "learning_rate": 2.9672205104801697e-05,
+      "loss": 0.7214,
+      "step": 4017
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.392965222682088,
+      "learning_rate": 2.9667135038393135e-05,
+      "loss": 0.7042,
+      "step": 4018
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.39040800446004775,
+      "learning_rate": 2.9662064161213636e-05,
+      "loss": 0.6603,
+      "step": 4019
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3611980545403996,
+      "learning_rate": 2.965699247368848e-05,
+      "loss": 0.6805,
+      "step": 4020
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4246307256153418,
+      "learning_rate": 2.9651919976243034e-05,
+      "loss": 0.7042,
+      "step": 4021
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4115790744008534,
+      "learning_rate": 2.9646846669302712e-05,
+      "loss": 0.6845,
+      "step": 4022
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3696214472455507,
+      "learning_rate": 2.9641772553293012e-05,
+      "loss": 0.685,
+      "step": 4023
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3914492284175548,
+      "learning_rate": 2.9636697628639494e-05,
+      "loss": 0.6743,
+      "step": 4024
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.43223389640477367,
+      "learning_rate": 2.963162189576778e-05,
+      "loss": 0.7329,
+      "step": 4025
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4158133837042,
+      "learning_rate": 2.9626545355103572e-05,
+      "loss": 0.6675,
+      "step": 4026
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4241287780605396,
+      "learning_rate": 2.9621468007072637e-05,
+      "loss": 0.7143,
+      "step": 4027
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.42628031423018536,
+      "learning_rate": 2.961638985210079e-05,
+      "loss": 0.6816,
+      "step": 4028
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.35420379550100983,
+      "learning_rate": 2.961131089061394e-05,
+      "loss": 0.6329,
+      "step": 4029
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4161622332199104,
+      "learning_rate": 2.960623112303806e-05,
+      "loss": 7.4561,
+      "step": 4030
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4111737716112296,
+      "learning_rate": 2.960115054979918e-05,
+      "loss": 0.6782,
+      "step": 4031
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.42246037304305295,
+      "learning_rate": 2.9596069171323392e-05,
+      "loss": 0.6569,
+      "step": 4032
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4173956337017854,
+      "learning_rate": 2.959098698803687e-05,
+      "loss": 0.7099,
+      "step": 4033
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4424569510650833,
+      "learning_rate": 2.9585904000365856e-05,
+      "loss": 0.7096,
+      "step": 4034
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3571140867213609,
+      "learning_rate": 2.9580820208736648e-05,
+      "loss": 0.6957,
+      "step": 4035
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.40228882192157533,
+      "learning_rate": 2.9575735613575615e-05,
+      "loss": 0.7306,
+      "step": 4036
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.35978467357077787,
+      "learning_rate": 2.9570650215309214e-05,
+      "loss": 0.692,
+      "step": 4037
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3984141648344715,
+      "learning_rate": 2.956556401436393e-05,
+      "loss": 0.6753,
+      "step": 4038
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.421093323022916,
+      "learning_rate": 2.9560477011166344e-05,
+      "loss": 0.7002,
+      "step": 4039
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3819229658142759,
+      "learning_rate": 2.9555389206143098e-05,
+      "loss": 0.7213,
+      "step": 4040
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.38159461192777094,
+      "learning_rate": 2.95503005997209e-05,
+      "loss": 0.6996,
+      "step": 4041
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3742799604325021,
+      "learning_rate": 2.954521119232652e-05,
+      "loss": 0.6614,
+      "step": 4042
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4069068086767543,
+      "learning_rate": 2.954012098438681e-05,
+      "loss": 0.6555,
+      "step": 4043
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4863701341725432,
+      "learning_rate": 2.9535029976328675e-05,
+      "loss": 0.6827,
+      "step": 4044
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.37254583866661445,
+      "learning_rate": 2.952993816857909e-05,
+      "loss": 0.6756,
+      "step": 4045
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4132050585469108,
+      "learning_rate": 2.9524845561565096e-05,
+      "loss": 0.6735,
+      "step": 4046
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3925794823440072,
+      "learning_rate": 2.9519752155713814e-05,
+      "loss": 0.6577,
+      "step": 4047
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4127348479993225,
+      "learning_rate": 2.9514657951452414e-05,
+      "loss": 0.6974,
+      "step": 4048
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4237393487831132,
+      "learning_rate": 2.950956294920814e-05,
+      "loss": 0.6975,
+      "step": 4049
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.42470234146858293,
+      "learning_rate": 2.9504467149408306e-05,
+      "loss": 0.6972,
+      "step": 4050
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4211865621369962,
+      "learning_rate": 2.9499370552480298e-05,
+      "loss": 0.6933,
+      "step": 4051
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4240016627342541,
+      "learning_rate": 2.9494273158851542e-05,
+      "loss": 0.7266,
+      "step": 4052
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.35097485781571286,
+      "learning_rate": 2.948917496894957e-05,
+      "loss": 0.6744,
+      "step": 4053
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.41185559449558107,
+      "learning_rate": 2.9484075983201953e-05,
+      "loss": 0.6794,
+      "step": 4054
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3729788424281045,
+      "learning_rate": 2.947897620203634e-05,
+      "loss": 0.6566,
+      "step": 4055
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4232678094433793,
+      "learning_rate": 2.9473875625880435e-05,
+      "loss": 0.7214,
+      "step": 4056
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.45936889394742325,
+      "learning_rate": 2.946877425516202e-05,
+      "loss": 0.7057,
+      "step": 4057
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.37237578000214033,
+      "learning_rate": 2.9463672090308956e-05,
+      "loss": 0.7011,
+      "step": 4058
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.40133882965558576,
+      "learning_rate": 2.945856913174913e-05,
+      "loss": 0.6851,
+      "step": 4059
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4002187941075614,
+      "learning_rate": 2.945346537991054e-05,
+      "loss": 0.7133,
+      "step": 4060
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3503026247241475,
+      "learning_rate": 2.9448360835221223e-05,
+      "loss": 0.6407,
+      "step": 4061
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.41719675613365326,
+      "learning_rate": 2.9443255498109295e-05,
+      "loss": 0.6731,
+      "step": 4062
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.36777890615476044,
+      "learning_rate": 2.9438149369002937e-05,
+      "loss": 0.7293,
+      "step": 4063
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.38409844262579446,
+      "learning_rate": 2.9433042448330384e-05,
+      "loss": 0.7122,
+      "step": 4064
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.39464329952743166,
+      "learning_rate": 2.9427934736519962e-05,
+      "loss": 0.7107,
+      "step": 4065
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.40415924115274116,
+      "learning_rate": 2.9422826234000024e-05,
+      "loss": 0.6852,
+      "step": 4066
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.38493492948602787,
+      "learning_rate": 2.941771694119904e-05,
+      "loss": 0.6883,
+      "step": 4067
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.39044174262005193,
+      "learning_rate": 2.941260685854551e-05,
+      "loss": 0.6918,
+      "step": 4068
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3520746409659562,
+      "learning_rate": 2.9407495986468007e-05,
+      "loss": 0.6736,
+      "step": 4069
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.31649754409463454,
+      "learning_rate": 2.9402384325395182e-05,
+      "loss": 0.6325,
+      "step": 4070
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3771207276527628,
+      "learning_rate": 2.9397271875755734e-05,
+      "loss": 0.7025,
+      "step": 4071
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.38322321680857696,
+      "learning_rate": 2.939215863797844e-05,
+      "loss": 0.6661,
+      "step": 4072
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.37114560330532814,
+      "learning_rate": 2.9387044612492155e-05,
+      "loss": 0.6984,
+      "step": 4073
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.41390845425098294,
+      "learning_rate": 2.9381929799725764e-05,
+      "loss": 0.711,
+      "step": 4074
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3977425294520562,
+      "learning_rate": 2.937681420010826e-05,
+      "loss": 0.6828,
+      "step": 4075
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.38630229468399063,
+      "learning_rate": 2.9371697814068665e-05,
+      "loss": 0.6786,
+      "step": 4076
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3545685554528547,
+      "learning_rate": 2.9366580642036096e-05,
+      "loss": 0.6882,
+      "step": 4077
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3651910295230493,
+      "learning_rate": 2.936146268443972e-05,
+      "loss": 0.6769,
+      "step": 4078
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.40866775363539326,
+      "learning_rate": 2.9356343941708777e-05,
+      "loss": 0.7082,
+      "step": 4079
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.38467648965239193,
+      "learning_rate": 2.9351224414272566e-05,
+      "loss": 0.6953,
+      "step": 4080
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.33371331696995327,
+      "learning_rate": 2.9346104102560455e-05,
+      "loss": 0.6637,
+      "step": 4081
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.33868839343440255,
+      "learning_rate": 2.9340983007001882e-05,
+      "loss": 0.6857,
+      "step": 4082
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.37496380166522514,
+      "learning_rate": 2.9335861128026345e-05,
+      "loss": 0.6547,
+      "step": 4083
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3616625294396917,
+      "learning_rate": 2.933073846606341e-05,
+      "loss": 0.6594,
+      "step": 4084
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4018305200938547,
+      "learning_rate": 2.932561502154271e-05,
+      "loss": 0.6781,
+      "step": 4085
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4072332255632758,
+      "learning_rate": 2.9320490794893937e-05,
+      "loss": 0.7119,
+      "step": 4086
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.37432196240283033,
+      "learning_rate": 2.931536578654686e-05,
+      "loss": 0.681,
+      "step": 4087
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3949357330973757,
+      "learning_rate": 2.9310239996931303e-05,
+      "loss": 0.6612,
+      "step": 4088
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40346644163718204,
+      "learning_rate": 2.930511342647716e-05,
+      "loss": 0.7,
+      "step": 4089
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4063793485758525,
+      "learning_rate": 2.92999860756144e-05,
+      "loss": 0.707,
+      "step": 4090
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40934747751916206,
+      "learning_rate": 2.929485794477303e-05,
+      "loss": 0.6513,
+      "step": 4091
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.395046919005007,
+      "learning_rate": 2.9289729034383156e-05,
+      "loss": 0.7214,
+      "step": 4092
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4258935871057237,
+      "learning_rate": 2.928459934487492e-05,
+      "loss": 0.7087,
+      "step": 4093
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.176775506461088,
+      "learning_rate": 2.9279468876678554e-05,
+      "loss": 7.3265,
+      "step": 4094
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4367112278441963,
+      "learning_rate": 2.927433763022434e-05,
+      "loss": 0.729,
+      "step": 4095
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4215357407061127,
+      "learning_rate": 2.9269205605942628e-05,
+      "loss": 0.698,
+      "step": 4096
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3500991085856153,
+      "learning_rate": 2.9264072804263838e-05,
+      "loss": 0.6539,
+      "step": 4097
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.37880216199073874,
+      "learning_rate": 2.925893922561845e-05,
+      "loss": 0.6951,
+      "step": 4098
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.36775362639914816,
+      "learning_rate": 2.9253804870437008e-05,
+      "loss": 0.6858,
+      "step": 4099
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40661354767672075,
+      "learning_rate": 2.9248669739150134e-05,
+      "loss": 0.729,
+      "step": 4100
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.42441354004544884,
+      "learning_rate": 2.9243533832188497e-05,
+      "loss": 0.6772,
+      "step": 4101
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.41241655318566783,
+      "learning_rate": 2.9238397149982836e-05,
+      "loss": 0.6725,
+      "step": 4102
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.36573000348386925,
+      "learning_rate": 2.9233259692963964e-05,
+      "loss": 0.6343,
+      "step": 4103
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40928997834044556,
+      "learning_rate": 2.9228121461562754e-05,
+      "loss": 0.6725,
+      "step": 4104
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.36593971782617774,
+      "learning_rate": 2.922298245621014e-05,
+      "loss": 0.6261,
+      "step": 4105
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.39620201457049264,
+      "learning_rate": 2.9217842677337125e-05,
+      "loss": 0.6979,
+      "step": 4106
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4333974709084557,
+      "learning_rate": 2.9212702125374782e-05,
+      "loss": 0.6925,
+      "step": 4107
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.37168187865987967,
+      "learning_rate": 2.9207560800754237e-05,
+      "loss": 0.6812,
+      "step": 4108
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40843586704162377,
+      "learning_rate": 2.920241870390669e-05,
+      "loss": 0.7087,
+      "step": 4109
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3259195700136305,
+      "learning_rate": 2.9197275835263395e-05,
+      "loss": 0.6445,
+      "step": 4110
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.37403999901871177,
+      "learning_rate": 2.9192132195255686e-05,
+      "loss": 0.6931,
+      "step": 4111
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.35620398256244795,
+      "learning_rate": 2.9186987784314946e-05,
+      "loss": 0.6721,
+      "step": 4112
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4087461282587025,
+      "learning_rate": 2.9181842602872638e-05,
+      "loss": 0.7431,
+      "step": 4113
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3395590792200781,
+      "learning_rate": 2.9176696651360283e-05,
+      "loss": 0.6863,
+      "step": 4114
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3572219448976761,
+      "learning_rate": 2.9171549930209463e-05,
+      "loss": 0.6564,
+      "step": 4115
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3753785681075452,
+      "learning_rate": 2.9166402439851817e-05,
+      "loss": 0.6988,
+      "step": 4116
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.403791190602948,
+      "learning_rate": 2.916125418071908e-05,
+      "loss": 0.7153,
+      "step": 4117
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.41037477726185395,
+      "learning_rate": 2.9156105153243016e-05,
+      "loss": 0.7298,
+      "step": 4118
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.36296227644058865,
+      "learning_rate": 2.915095535785547e-05,
+      "loss": 0.7029,
+      "step": 4119
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.41266543566311686,
+      "learning_rate": 2.9145804794988345e-05,
+      "loss": 0.6868,
+      "step": 4120
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4185080106110995,
+      "learning_rate": 2.914065346507362e-05,
+      "loss": 0.7127,
+      "step": 4121
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4111444888363245,
+      "learning_rate": 2.913550136854333e-05,
+      "loss": 0.6644,
+      "step": 4122
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.35308084677626006,
+      "learning_rate": 2.913034850582957e-05,
+      "loss": 0.6627,
+      "step": 4123
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.39803301406584424,
+      "learning_rate": 2.912519487736451e-05,
+      "loss": 0.6544,
+      "step": 4124
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.38952210370650375,
+      "learning_rate": 2.9120040483580378e-05,
+      "loss": 0.6999,
+      "step": 4125
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40172075381401295,
+      "learning_rate": 2.9114885324909462e-05,
+      "loss": 0.6683,
+      "step": 4126
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40946013009921073,
+      "learning_rate": 2.9109729401784118e-05,
+      "loss": 0.6904,
+      "step": 4127
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3757273371598515,
+      "learning_rate": 2.910457271463678e-05,
+      "loss": 0.6518,
+      "step": 4128
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.38080390134558306,
+      "learning_rate": 2.9099415263899917e-05,
+      "loss": 0.6647,
+      "step": 4129
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3757795814830267,
+      "learning_rate": 2.909425705000609e-05,
+      "loss": 0.67,
+      "step": 4130
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.43819099060589894,
+      "learning_rate": 2.9089098073387904e-05,
+      "loss": 0.7287,
+      "step": 4131
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4001119192249678,
+      "learning_rate": 2.9083938334478042e-05,
+      "loss": 0.6756,
+      "step": 4132
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3482749962438557,
+      "learning_rate": 2.9078777833709235e-05,
+      "loss": 0.6305,
+      "step": 4133
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.41047793413186906,
+      "learning_rate": 2.90736165715143e-05,
+      "loss": 0.6708,
+      "step": 4134
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3936276703708496,
+      "learning_rate": 2.9068454548326106e-05,
+      "loss": 0.658,
+      "step": 4135
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.348455457922412,
+      "learning_rate": 2.9063291764577575e-05,
+      "loss": 0.7135,
+      "step": 4136
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.38261842018274933,
+      "learning_rate": 2.9058128220701714e-05,
+      "loss": 0.6743,
+      "step": 4137
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.38331958396218013,
+      "learning_rate": 2.9052963917131574e-05,
+      "loss": 0.7123,
+      "step": 4138
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.35179552345682974,
+      "learning_rate": 2.9047798854300288e-05,
+      "loss": 0.6696,
+      "step": 4139
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3704033734699605,
+      "learning_rate": 2.9042633032641034e-05,
+      "loss": 0.6948,
+      "step": 4140
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.41949849940685974,
+      "learning_rate": 2.903746645258707e-05,
+      "loss": 0.6927,
+      "step": 4141
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.35785755053763196,
+      "learning_rate": 2.9032299114571713e-05,
+      "loss": 0.6739,
+      "step": 4142
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3867206142406612,
+      "learning_rate": 2.9027131019028334e-05,
+      "loss": 0.6626,
+      "step": 4143
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.37825803543842307,
+      "learning_rate": 2.9021962166390377e-05,
+      "loss": 0.6622,
+      "step": 4144
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.7236277010288874,
+      "learning_rate": 2.9016792557091355e-05,
+      "loss": 7.168,
+      "step": 4145
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4426678622885949,
+      "learning_rate": 2.9011622191564824e-05,
+      "loss": 0.7259,
+      "step": 4146
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.44236560878029335,
+      "learning_rate": 2.9006451070244422e-05,
+      "loss": 0.7043,
+      "step": 4147
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.36339038008187097,
+      "learning_rate": 2.9001279193563853e-05,
+      "loss": 0.6458,
+      "step": 4148
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4376900511212397,
+      "learning_rate": 2.8996106561956868e-05,
+      "loss": 0.6937,
+      "step": 4149
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.39723257406994955,
+      "learning_rate": 2.899093317585728e-05,
+      "loss": 0.7017,
+      "step": 4150
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3777288062121335,
+      "learning_rate": 2.8985759035698997e-05,
+      "loss": 0.6739,
+      "step": 4151
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4104536478689999,
+      "learning_rate": 2.8980584141915957e-05,
+      "loss": 0.7003,
+      "step": 4152
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3712005810149983,
+      "learning_rate": 2.897540849494216e-05,
+      "loss": 0.7,
+      "step": 4153
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4052993751660521,
+      "learning_rate": 2.8970232095211694e-05,
+      "loss": 0.6906,
+      "step": 4154
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4014701058404007,
+      "learning_rate": 2.8965054943158707e-05,
+      "loss": 0.6185,
+      "step": 4155
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4265618294549581,
+      "learning_rate": 2.8959877039217375e-05,
+      "loss": 0.6871,
+      "step": 4156
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3760591107378667,
+      "learning_rate": 2.895469838382198e-05,
+      "loss": 0.6789,
+      "step": 4157
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.384777336160198,
+      "learning_rate": 2.894951897740685e-05,
+      "loss": 0.6871,
+      "step": 4158
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3626854102343711,
+      "learning_rate": 2.8944338820406376e-05,
+      "loss": 0.6979,
+      "step": 4159
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.35229518135749033,
+      "learning_rate": 2.8939157913254994e-05,
+      "loss": 0.6562,
+      "step": 4160
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3627840290069858,
+      "learning_rate": 2.893397625638724e-05,
+      "loss": 0.6459,
+      "step": 4161
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3632931658484872,
+      "learning_rate": 2.8928793850237694e-05,
+      "loss": 0.674,
+      "step": 4162
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.33290915111777253,
+      "learning_rate": 2.892361069524098e-05,
+      "loss": 0.6413,
+      "step": 4163
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3917712421279665,
+      "learning_rate": 2.8918426791831815e-05,
+      "loss": 0.697,
+      "step": 4164
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.39579277265042373,
+      "learning_rate": 2.891324214044497e-05,
+      "loss": 0.6935,
+      "step": 4165
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.389535368742357,
+      "learning_rate": 2.8908056741515276e-05,
+      "loss": 0.684,
+      "step": 4166
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3494052171194364,
+      "learning_rate": 2.8902870595477615e-05,
+      "loss": 0.6793,
+      "step": 4167
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3768447630747201,
+      "learning_rate": 2.8897683702766946e-05,
+      "loss": 0.6974,
+      "step": 4168
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.33404625534678817,
+      "learning_rate": 2.88924960638183e-05,
+      "loss": 0.6688,
+      "step": 4169
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.38321584761763744,
+      "learning_rate": 2.8887307679066737e-05,
+      "loss": 0.6794,
+      "step": 4170
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4556565931864041,
+      "learning_rate": 2.8882118548947415e-05,
+      "loss": 0.6996,
+      "step": 4171
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4843233722167355,
+      "learning_rate": 2.8876928673895545e-05,
+      "loss": 0.7461,
+      "step": 4172
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4580834792887662,
+      "learning_rate": 2.887173805434638e-05,
+      "loss": 0.6992,
+      "step": 4173
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4101579436875091,
+      "learning_rate": 2.8866546690735258e-05,
+      "loss": 0.6994,
+      "step": 4174
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3655127591382048,
+      "learning_rate": 2.8861354583497577e-05,
+      "loss": 0.6807,
+      "step": 4175
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3293756595100983,
+      "learning_rate": 2.8856161733068794e-05,
+      "loss": 0.6489,
+      "step": 4176
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4136127801719381,
+      "learning_rate": 2.8850968139884416e-05,
+      "loss": 0.6201,
+      "step": 4177
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.35871411955538646,
+      "learning_rate": 2.8845773804380028e-05,
+      "loss": 0.668,
+      "step": 4178
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3909661076302462,
+      "learning_rate": 2.8840578726991278e-05,
+      "loss": 0.6828,
+      "step": 4179
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4398391373652183,
+      "learning_rate": 2.8835382908153865e-05,
+      "loss": 0.6765,
+      "step": 4180
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3708099672194635,
+      "learning_rate": 2.8830186348303555e-05,
+      "loss": 0.6493,
+      "step": 4181
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40683321584086535,
+      "learning_rate": 2.8824989047876185e-05,
+      "loss": 0.6836,
+      "step": 4182
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.38885057548450386,
+      "learning_rate": 2.8819791007307638e-05,
+      "loss": 0.601,
+      "step": 4183
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.35411523570692627,
+      "learning_rate": 2.881459222703387e-05,
+      "loss": 0.658,
+      "step": 4184
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.4521881946511157,
+      "learning_rate": 2.88093927074909e-05,
+      "loss": 7.4488,
+      "step": 4185
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4075067871447769,
+      "learning_rate": 2.8804192449114805e-05,
+      "loss": 0.6678,
+      "step": 4186
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3751595771233198,
+      "learning_rate": 2.8798991452341715e-05,
+      "loss": 0.663,
+      "step": 4187
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3584682789742843,
+      "learning_rate": 2.879378971760784e-05,
+      "loss": 0.6362,
+      "step": 4188
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3865992289531617,
+      "learning_rate": 2.8788587245349448e-05,
+      "loss": 0.6717,
+      "step": 4189
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.37105480392222356,
+      "learning_rate": 2.8783384036002853e-05,
+      "loss": 0.6895,
+      "step": 4190
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.43367910231193657,
+      "learning_rate": 2.877818009000445e-05,
+      "loss": 0.6627,
+      "step": 4191
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.381541857555003,
+      "learning_rate": 2.877297540779068e-05,
+      "loss": 0.6572,
+      "step": 4192
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3875204366642779,
+      "learning_rate": 2.876776998979806e-05,
+      "loss": 0.6914,
+      "step": 4193
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.37114531045707283,
+      "learning_rate": 2.8762563836463155e-05,
+      "loss": 0.6877,
+      "step": 4194
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3847541149870324,
+      "learning_rate": 2.875735694822261e-05,
+      "loss": 0.6637,
+      "step": 4195
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3971026051562751,
+      "learning_rate": 2.875214932551311e-05,
+      "loss": 0.6493,
+      "step": 4196
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.38108640197933585,
+      "learning_rate": 2.874694096877142e-05,
+      "loss": 0.6975,
+      "step": 4197
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3662349862261279,
+      "learning_rate": 2.874173187843435e-05,
+      "loss": 0.7199,
+      "step": 4198
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.38731711105355454,
+      "learning_rate": 2.873652205493879e-05,
+      "loss": 0.6921,
+      "step": 4199
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3781200016731424,
+      "learning_rate": 2.8731311498721667e-05,
+      "loss": 0.7033,
+      "step": 4200
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.33616901707099095,
+      "learning_rate": 2.8726100210220004e-05,
+      "loss": 0.6639,
+      "step": 4201
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.36419318156016117,
+      "learning_rate": 2.8720888189870846e-05,
+      "loss": 0.6928,
+      "step": 4202
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.35958759752140085,
+      "learning_rate": 2.8715675438111334e-05,
+      "loss": 0.6897,
+      "step": 4203
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3670811316596534,
+      "learning_rate": 2.8710461955378645e-05,
+      "loss": 0.6798,
+      "step": 4204
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.40225777063466606,
+      "learning_rate": 2.8705247742110037e-05,
+      "loss": 0.6746,
+      "step": 4205
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.426753169296377,
+      "learning_rate": 2.870003279874281e-05,
+      "loss": 0.7073,
+      "step": 4206
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.37617336109592425,
+      "learning_rate": 2.8694817125714335e-05,
+      "loss": 0.6549,
+      "step": 4207
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4111284577148159,
+      "learning_rate": 2.8689600723462056e-05,
+      "loss": 0.7234,
+      "step": 4208
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4301765382873988,
+      "learning_rate": 2.8684383592423452e-05,
+      "loss": 0.6975,
+      "step": 4209
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.42352885841042326,
+      "learning_rate": 2.867916573303609e-05,
+      "loss": 0.7193,
+      "step": 4210
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.45439428178415875,
+      "learning_rate": 2.8673947145737577e-05,
+      "loss": 0.7082,
+      "step": 4211
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4456230450539726,
+      "learning_rate": 2.8668727830965592e-05,
+      "loss": 0.7408,
+      "step": 4212
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4333824831148669,
+      "learning_rate": 2.8663507789157873e-05,
+      "loss": 0.6804,
+      "step": 4213
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3861133581222309,
+      "learning_rate": 2.8658287020752217e-05,
+      "loss": 0.6818,
+      "step": 4214
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.36745501915716583,
+      "learning_rate": 2.8653065526186492e-05,
+      "loss": 0.675,
+      "step": 4215
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.39275946669008227,
+      "learning_rate": 2.8647843305898605e-05,
+      "loss": 0.6734,
+      "step": 4216
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.347179161379187,
+      "learning_rate": 2.8642620360326546e-05,
+      "loss": 0.6594,
+      "step": 4217
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.37401857706821895,
+      "learning_rate": 2.8637396689908353e-05,
+      "loss": 0.664,
+      "step": 4218
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3714250169734857,
+      "learning_rate": 2.8632172295082132e-05,
+      "loss": 0.6841,
+      "step": 4219
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.36903282082117017,
+      "learning_rate": 2.862694717628605e-05,
+      "loss": 0.7003,
+      "step": 4220
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3844914004956783,
+      "learning_rate": 2.862172133395832e-05,
+      "loss": 0.6581,
+      "step": 4221
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4098527802168956,
+      "learning_rate": 2.861649476853724e-05,
+      "loss": 0.6454,
+      "step": 4222
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.37379030621520426,
+      "learning_rate": 2.8611267480461148e-05,
+      "loss": 0.6433,
+      "step": 4223
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.43355990495646934,
+      "learning_rate": 2.860603947016845e-05,
+      "loss": 0.6668,
+      "step": 4224
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.38022548287502134,
+      "learning_rate": 2.860081073809762e-05,
+      "loss": 0.6897,
+      "step": 4225
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.33422937403975805,
+      "learning_rate": 2.8595581284687176e-05,
+      "loss": 0.6548,
+      "step": 4226
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.43621232687280637,
+      "learning_rate": 2.859035111037571e-05,
+      "loss": 0.6777,
+      "step": 4227
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.38829239754816225,
+      "learning_rate": 2.8585120215601872e-05,
+      "loss": 0.6944,
+      "step": 4228
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3620121564535617,
+      "learning_rate": 2.857988860080437e-05,
+      "loss": 0.6591,
+      "step": 4229
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4233887586238042,
+      "learning_rate": 2.8574656266421975e-05,
+      "loss": 0.686,
+      "step": 4230
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.39466922450238107,
+      "learning_rate": 2.856942321289351e-05,
+      "loss": 0.6987,
+      "step": 4231
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.39009838924910917,
+      "learning_rate": 2.8564189440657876e-05,
+      "loss": 0.6781,
+      "step": 4232
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3870757455798979,
+      "learning_rate": 2.855895495015401e-05,
+      "loss": 0.7057,
+      "step": 4233
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3445054876758797,
+      "learning_rate": 2.8553719741820928e-05,
+      "loss": 0.6554,
+      "step": 4234
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3193401156057875,
+      "learning_rate": 2.854848381609771e-05,
+      "loss": 0.6631,
+      "step": 4235
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4091894181178523,
+      "learning_rate": 2.8543247173423477e-05,
+      "loss": 0.6508,
+      "step": 4236
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3963880674668443,
+      "learning_rate": 2.8538009814237412e-05,
+      "loss": 0.6816,
+      "step": 4237
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.34753958821823033,
+      "learning_rate": 2.853277173897878e-05,
+      "loss": 0.6849,
+      "step": 4238
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3868409908640387,
+      "learning_rate": 2.852753294808689e-05,
+      "loss": 0.6362,
+      "step": 4239
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4675743592871955,
+      "learning_rate": 2.8522293442001108e-05,
+      "loss": 0.7168,
+      "step": 4240
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.34613456778359114,
+      "learning_rate": 2.8517053221160868e-05,
+      "loss": 0.6424,
+      "step": 4241
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.41433677004056096,
+      "learning_rate": 2.8511812286005662e-05,
+      "loss": 0.7247,
+      "step": 4242
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4425101963381768,
+      "learning_rate": 2.8506570636975038e-05,
+      "loss": 0.6651,
+      "step": 4243
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3658310838448775,
+      "learning_rate": 2.8501328274508604e-05,
+      "loss": 0.6665,
+      "step": 4244
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4027448294226748,
+      "learning_rate": 2.849608519904604e-05,
+      "loss": 0.6864,
+      "step": 4245
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.43248340549699354,
+      "learning_rate": 2.849084141102707e-05,
+      "loss": 0.7414,
+      "step": 4246
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4570792572056199,
+      "learning_rate": 2.8485596910891482e-05,
+      "loss": 0.7443,
+      "step": 4247
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.660301010111874,
+      "learning_rate": 2.8480351699079133e-05,
+      "loss": 7.2191,
+      "step": 4248
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4493658917753823,
+      "learning_rate": 2.8475105776029924e-05,
+      "loss": 0.6848,
+      "step": 4249
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.40005701940779637,
+      "learning_rate": 2.8469859142183836e-05,
+      "loss": 0.6732,
+      "step": 4250
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3742977207308368,
+      "learning_rate": 2.8464611797980885e-05,
+      "loss": 0.712,
+      "step": 4251
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.41345516980248204,
+      "learning_rate": 2.8459363743861166e-05,
+      "loss": 0.6845,
+      "step": 4252
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3879264547067304,
+      "learning_rate": 2.8454114980264827e-05,
+      "loss": 0.631,
+      "step": 4253
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.36598500938598705,
+      "learning_rate": 2.8448865507632075e-05,
+      "loss": 0.6811,
+      "step": 4254
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3995624345514939,
+      "learning_rate": 2.844361532640317e-05,
+      "loss": 0.6816,
+      "step": 4255
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.45750101211774075,
+      "learning_rate": 2.8438364437018457e-05,
+      "loss": 0.7443,
+      "step": 4256
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4215830892352896,
+      "learning_rate": 2.84331128399183e-05,
+      "loss": 0.6806,
+      "step": 4257
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3973322698065129,
+      "learning_rate": 2.8427860535543157e-05,
+      "loss": 0.7029,
+      "step": 4258
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.40560053541499147,
+      "learning_rate": 2.842260752433353e-05,
+      "loss": 0.6591,
+      "step": 4259
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6434355954297257,
+      "learning_rate": 2.8417353806729986e-05,
+      "loss": 7.2431,
+      "step": 4260
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3672888628434924,
+      "learning_rate": 2.841209938317313e-05,
+      "loss": 0.6587,
+      "step": 4261
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.44685535783307595,
+      "learning_rate": 2.840684425410367e-05,
+      "loss": 0.6993,
+      "step": 4262
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3876579466219332,
+      "learning_rate": 2.8401588419962337e-05,
+      "loss": 0.674,
+      "step": 4263
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.37450407113429357,
+      "learning_rate": 2.8396331881189923e-05,
+      "loss": 0.6672,
+      "step": 4264
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.40591505343934126,
+      "learning_rate": 2.8391074638227297e-05,
+      "loss": 0.6577,
+      "step": 4265
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.388003495090219,
+      "learning_rate": 2.8385816691515376e-05,
+      "loss": 0.6629,
+      "step": 4266
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.39501347321828323,
+      "learning_rate": 2.8380558041495137e-05,
+      "loss": 0.6892,
+      "step": 4267
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.33621397542999376,
+      "learning_rate": 2.8375298688607605e-05,
+      "loss": 0.6635,
+      "step": 4268
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.36921920075916986,
+      "learning_rate": 2.8370038633293902e-05,
+      "loss": 0.6764,
+      "step": 4269
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4192217059467799,
+      "learning_rate": 2.836477787599516e-05,
+      "loss": 0.6628,
+      "step": 4270
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3937427253709522,
+      "learning_rate": 2.83595164171526e-05,
+      "loss": 0.7035,
+      "step": 4271
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.40603261834967597,
+      "learning_rate": 2.8354254257207495e-05,
+      "loss": 0.7247,
+      "step": 4272
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3809676924357507,
+      "learning_rate": 2.8348991396601176e-05,
+      "loss": 0.6548,
+      "step": 4273
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.40375628099689653,
+      "learning_rate": 2.8343727835775027e-05,
+      "loss": 0.6921,
+      "step": 4274
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6450456420613486,
+      "learning_rate": 2.83384635751705e-05,
+      "loss": 7.2634,
+      "step": 4275
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.40133691505415703,
+      "learning_rate": 2.8333198615229113e-05,
+      "loss": 0.7412,
+      "step": 4276
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4066401785873176,
+      "learning_rate": 2.832793295639242e-05,
+      "loss": 0.7026,
+      "step": 4277
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.37590854277986,
+      "learning_rate": 2.832266659910204e-05,
+      "loss": 0.6519,
+      "step": 4278
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4344395335844498,
+      "learning_rate": 2.8317399543799667e-05,
+      "loss": 0.6607,
+      "step": 4279
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.1933828539080171,
+      "learning_rate": 2.8312131790927044e-05,
+      "loss": 7.2947,
+      "step": 4280
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4106136879118101,
+      "learning_rate": 2.830686334092596e-05,
+      "loss": 0.6933,
+      "step": 4281
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.38563649857586474,
+      "learning_rate": 2.830159419423828e-05,
+      "loss": 0.6825,
+      "step": 4282
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.38129644547028396,
+      "learning_rate": 2.8296324351305923e-05,
+      "loss": 0.6588,
+      "step": 4283
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4049078542543032,
+      "learning_rate": 2.8291053812570862e-05,
+      "loss": 0.6746,
+      "step": 4284
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4114763879968166,
+      "learning_rate": 2.8285782578475127e-05,
+      "loss": 0.6789,
+      "step": 4285
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4103659024544386,
+      "learning_rate": 2.828051064946082e-05,
+      "loss": 0.6786,
+      "step": 4286
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.37548356249719766,
+      "learning_rate": 2.8275238025970085e-05,
+      "loss": 0.6505,
+      "step": 4287
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.36222704705124104,
+      "learning_rate": 2.8269964708445127e-05,
+      "loss": 0.6237,
+      "step": 4288
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3553957405878825,
+      "learning_rate": 2.8264690697328215e-05,
+      "loss": 0.667,
+      "step": 4289
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.41208094157102176,
+      "learning_rate": 2.8259415993061678e-05,
+      "loss": 0.6993,
+      "step": 4290
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4185719254834075,
+      "learning_rate": 2.8254140596087897e-05,
+      "loss": 0.6986,
+      "step": 4291
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.39774080073764245,
+      "learning_rate": 2.824886450684931e-05,
+      "loss": 0.6978,
+      "step": 4292
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.34219451558230135,
+      "learning_rate": 2.824358772578842e-05,
+      "loss": 0.6508,
+      "step": 4293
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.386972295401281,
+      "learning_rate": 2.8238310253347793e-05,
+      "loss": 0.66,
+      "step": 4294
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3841147361669332,
+      "learning_rate": 2.8233032089970023e-05,
+      "loss": 0.6664,
+      "step": 4295
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4472094529688035,
+      "learning_rate": 2.8227753236097792e-05,
+      "loss": 0.6602,
+      "step": 4296
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.377883421823919,
+      "learning_rate": 2.8222473692173843e-05,
+      "loss": 0.686,
+      "step": 4297
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3971918202921587,
+      "learning_rate": 2.8217193458640947e-05,
+      "loss": 0.7112,
+      "step": 4298
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.35554968886307026,
+      "learning_rate": 2.8211912535941964e-05,
+      "loss": 0.6656,
+      "step": 4299
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.43385348703855553,
+      "learning_rate": 2.8206630924519796e-05,
+      "loss": 0.7121,
+      "step": 4300
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3979328962596579,
+      "learning_rate": 2.8201348624817403e-05,
+      "loss": 0.7006,
+      "step": 4301
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.38440741281617535,
+      "learning_rate": 2.81960656372778e-05,
+      "loss": 0.6459,
+      "step": 4302
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3695845015370357,
+      "learning_rate": 2.8190781962344078e-05,
+      "loss": 0.7068,
+      "step": 4303
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.38036527874336995,
+      "learning_rate": 2.8185497600459368e-05,
+      "loss": 0.6765,
+      "step": 4304
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.43475372705578835,
+      "learning_rate": 2.8180212552066847e-05,
+      "loss": 0.6835,
+      "step": 4305
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3993965726947506,
+      "learning_rate": 2.8174926817609782e-05,
+      "loss": 0.703,
+      "step": 4306
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3670677394877076,
+      "learning_rate": 2.8169640397531486e-05,
+      "loss": 0.6674,
+      "step": 4307
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 24.807594965586514,
+      "learning_rate": 2.8164353292275307e-05,
+      "loss": 7.7528,
+      "step": 4308
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.2091229172626377,
+      "learning_rate": 2.815906550228468e-05,
+      "loss": 7.3316,
+      "step": 4309
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 13.221658621554829,
+      "learning_rate": 2.8153777028003087e-05,
+      "loss": 1.3348,
+      "step": 4310
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.359664009346692,
+      "learning_rate": 2.8148487869874062e-05,
+      "loss": 0.7728,
+      "step": 4311
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 9.242948359081401,
+      "learning_rate": 2.8143198028341197e-05,
+      "loss": 0.8951,
+      "step": 4312
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 8.030506178540993,
+      "learning_rate": 2.813790750384814e-05,
+      "loss": 1.0991,
+      "step": 4313
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 9.180777174677724,
+      "learning_rate": 2.8132616296838623e-05,
+      "loss": 1.0107,
+      "step": 4314
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.2804791193098044,
+      "learning_rate": 2.812732440775639e-05,
+      "loss": 0.95,
+      "step": 4315
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 69.27841375669765,
+      "learning_rate": 2.8122031837045274e-05,
+      "loss": 1.6998,
+      "step": 4316
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.152759165531739,
+      "learning_rate": 2.8116738585149163e-05,
+      "loss": 0.8897,
+      "step": 4317
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 99.00918739840353,
+      "learning_rate": 2.8111444652511984e-05,
+      "loss": 2.0033,
+      "step": 4318
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.358389250599343,
+      "learning_rate": 2.8106150039577742e-05,
+      "loss": 0.8562,
+      "step": 4319
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8801729571522912,
+      "learning_rate": 2.8100854746790482e-05,
+      "loss": 0.9266,
+      "step": 4320
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 13.130456834865026,
+      "learning_rate": 2.8095558774594325e-05,
+      "loss": 0.9702,
+      "step": 4321
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 42.63591882078234,
+      "learning_rate": 2.8090262123433426e-05,
+      "loss": 1.1382,
+      "step": 4322
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 23.919525992146717,
+      "learning_rate": 2.808496479375201e-05,
+      "loss": 1.2965,
+      "step": 4323
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 23.269550888945954,
+      "learning_rate": 2.8079666785994362e-05,
+      "loss": 1.3953,
+      "step": 4324
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.645804972462818,
+      "learning_rate": 2.807436810060482e-05,
+      "loss": 0.9848,
+      "step": 4325
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.13161267576206,
+      "learning_rate": 2.8069068738027777e-05,
+      "loss": 0.9464,
+      "step": 4326
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6430213547539307,
+      "learning_rate": 2.8063768698707683e-05,
+      "loss": 0.8735,
+      "step": 4327
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.7081863741081746,
+      "learning_rate": 2.8058467983089053e-05,
+      "loss": 0.8708,
+      "step": 4328
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.6098276221108156,
+      "learning_rate": 2.8053166591616435e-05,
+      "loss": 0.8826,
+      "step": 4329
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.283720242088362,
+      "learning_rate": 2.8047864524734466e-05,
+      "loss": 0.8443,
+      "step": 4330
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.3430211130681655,
+      "learning_rate": 2.8042561782887818e-05,
+      "loss": 0.873,
+      "step": 4331
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7063553077256145,
+      "learning_rate": 2.8037258366521224e-05,
+      "loss": 0.7556,
+      "step": 4332
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7553121917005199,
+      "learning_rate": 2.8031954276079486e-05,
+      "loss": 0.75,
+      "step": 4333
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.965994068706917,
+      "learning_rate": 2.8026649512007433e-05,
+      "loss": 0.7854,
+      "step": 4334
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9280210043805132,
+      "learning_rate": 2.8021344074749983e-05,
+      "loss": 0.8502,
+      "step": 4335
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6138515833114954,
+      "learning_rate": 2.801603796475209e-05,
+      "loss": 0.7955,
+      "step": 4336
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5999717474928756,
+      "learning_rate": 2.801073118245878e-05,
+      "loss": 0.73,
+      "step": 4337
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7366906748176153,
+      "learning_rate": 2.8005423728315118e-05,
+      "loss": 0.7513,
+      "step": 4338
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.631376003400838,
+      "learning_rate": 2.800011560276623e-05,
+      "loss": 0.7803,
+      "step": 4339
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9495391135087158,
+      "learning_rate": 2.799480680625732e-05,
+      "loss": 0.8122,
+      "step": 4340
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5854184689959674,
+      "learning_rate": 2.7989497339233607e-05,
+      "loss": 0.7482,
+      "step": 4341
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5221411304666562,
+      "learning_rate": 2.7984187202140405e-05,
+      "loss": 0.7652,
+      "step": 4342
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4681364320410623,
+      "learning_rate": 2.7978876395423066e-05,
+      "loss": 0.6907,
+      "step": 4343
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.556434173331339,
+      "learning_rate": 2.7973564919526998e-05,
+      "loss": 0.7577,
+      "step": 4344
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4372700158743075,
+      "learning_rate": 2.7968252774897677e-05,
+      "loss": 0.6989,
+      "step": 4345
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5221042408438025,
+      "learning_rate": 2.7962939961980607e-05,
+      "loss": 0.7348,
+      "step": 4346
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4612643740177062,
+      "learning_rate": 2.7957626481221382e-05,
+      "loss": 0.7257,
+      "step": 4347
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.51190057703336,
+      "learning_rate": 2.7952312333065643e-05,
+      "loss": 0.7119,
+      "step": 4348
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5547218828067833,
+      "learning_rate": 2.7946997517959062e-05,
+      "loss": 0.7894,
+      "step": 4349
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5572863390096852,
+      "learning_rate": 2.7941682036347405e-05,
+      "loss": 0.7371,
+      "step": 4350
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8047444609489183,
+      "learning_rate": 2.793636588867646e-05,
+      "loss": 0.7443,
+      "step": 4351
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.44477964350447396,
+      "learning_rate": 2.7931049075392097e-05,
+      "loss": 0.7326,
+      "step": 4352
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4941280547371568,
+      "learning_rate": 2.7925731596940226e-05,
+      "loss": 0.7379,
+      "step": 4353
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4594373702037416,
+      "learning_rate": 2.7920413453766815e-05,
+      "loss": 0.6966,
+      "step": 4354
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.48670020434780986,
+      "learning_rate": 2.7915094646317896e-05,
+      "loss": 0.7331,
+      "step": 4355
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4890805968495413,
+      "learning_rate": 2.7909775175039548e-05,
+      "loss": 0.695,
+      "step": 4356
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4559052846920509,
+      "learning_rate": 2.7904455040377912e-05,
+      "loss": 0.709,
+      "step": 4357
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5284985090515357,
+      "learning_rate": 2.7899134242779173e-05,
+      "loss": 0.7274,
+      "step": 4358
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.48941161081993934,
+      "learning_rate": 2.7893812782689584e-05,
+      "loss": 0.7064,
+      "step": 4359
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.384167258584514,
+      "learning_rate": 2.788849066055545e-05,
+      "loss": 0.6662,
+      "step": 4360
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5085179656304311,
+      "learning_rate": 2.7883167876823134e-05,
+      "loss": 0.7356,
+      "step": 4361
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5379235115671397,
+      "learning_rate": 2.7877844431939047e-05,
+      "loss": 0.7236,
+      "step": 4362
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.44220683135544003,
+      "learning_rate": 2.7872520326349664e-05,
+      "loss": 0.6864,
+      "step": 4363
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4985880779265213,
+      "learning_rate": 2.7867195560501508e-05,
+      "loss": 0.6931,
+      "step": 4364
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4167476459753828,
+      "learning_rate": 2.7861870134841163e-05,
+      "loss": 0.6975,
+      "step": 4365
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.43238724965606906,
+      "learning_rate": 2.7856544049815258e-05,
+      "loss": 0.7185,
+      "step": 4366
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4866755779759849,
+      "learning_rate": 2.78512173058705e-05,
+      "loss": 0.6809,
+      "step": 4367
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4868801367373978,
+      "learning_rate": 2.7845889903453623e-05,
+      "loss": 0.7509,
+      "step": 4368
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4420014785894388,
+      "learning_rate": 2.784056184301143e-05,
+      "loss": 0.7345,
+      "step": 4369
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.46251834213378207,
+      "learning_rate": 2.7835233124990795e-05,
+      "loss": 0.7009,
+      "step": 4370
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4402713828873047,
+      "learning_rate": 2.7829903749838618e-05,
+      "loss": 0.7017,
+      "step": 4371
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3769471348484148,
+      "learning_rate": 2.7824573718001868e-05,
+      "loss": 0.6626,
+      "step": 4372
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5257812418455999,
+      "learning_rate": 2.781924302992757e-05,
+      "loss": 0.7098,
+      "step": 4373
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 11.474668221895012,
+      "learning_rate": 2.7813911686062804e-05,
+      "loss": 7.667,
+      "step": 4374
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.48946216131479314,
+      "learning_rate": 2.7808579686854696e-05,
+      "loss": 0.7315,
+      "step": 4375
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5131546016618294,
+      "learning_rate": 2.780324703275044e-05,
+      "loss": 0.6999,
+      "step": 4376
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.6094342858196176,
+      "learning_rate": 2.779791372419729e-05,
+      "loss": 0.6818,
+      "step": 4377
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6200518758762118,
+      "learning_rate": 2.7792579761642527e-05,
+      "loss": 0.6812,
+      "step": 4378
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6057332872363954,
+      "learning_rate": 2.7787245145533505e-05,
+      "loss": 0.6923,
+      "step": 4379
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 66.46246790173578,
+      "learning_rate": 2.7781909876317645e-05,
+      "loss": 7.7684,
+      "step": 4380
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.4248654456596705,
+      "learning_rate": 2.7776573954442397e-05,
+      "loss": 0.766,
+      "step": 4381
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 36.9735527261453,
+      "learning_rate": 2.7771237380355284e-05,
+      "loss": 7.6775,
+      "step": 4382
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2134948135062498,
+      "learning_rate": 2.776590015450388e-05,
+      "loss": 0.7493,
+      "step": 4383
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.89057093783802,
+      "learning_rate": 2.776056227733581e-05,
+      "loss": 0.7679,
+      "step": 4384
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.321847657510832,
+      "learning_rate": 2.775522374929875e-05,
+      "loss": 0.7557,
+      "step": 4385
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7321201652975721,
+      "learning_rate": 2.7749884570840443e-05,
+      "loss": 0.7326,
+      "step": 4386
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.507981991991146,
+      "learning_rate": 2.7744544742408676e-05,
+      "loss": 0.7267,
+      "step": 4387
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 7.0644272416888425,
+      "learning_rate": 2.7739204264451297e-05,
+      "loss": 0.7522,
+      "step": 4388
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.1953310459172906,
+      "learning_rate": 2.7733863137416196e-05,
+      "loss": 0.7003,
+      "step": 4389
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.315237944848358,
+      "learning_rate": 2.7728521361751334e-05,
+      "loss": 0.7062,
+      "step": 4390
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.1503678602457588,
+      "learning_rate": 2.7723178937904728e-05,
+      "loss": 0.7259,
+      "step": 4391
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.786432027924719,
+      "learning_rate": 2.7717835866324424e-05,
+      "loss": 0.7395,
+      "step": 4392
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.3710143107654649,
+      "learning_rate": 2.7712492147458544e-05,
+      "loss": 0.7434,
+      "step": 4393
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 3.0555865370849675,
+      "learning_rate": 2.770714778175527e-05,
+      "loss": 0.6901,
+      "step": 4394
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5109331441489348,
+      "learning_rate": 2.770180276966282e-05,
+      "loss": 0.7466,
+      "step": 4395
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5570362997621436,
+      "learning_rate": 2.7696457111629463e-05,
+      "loss": 0.7102,
+      "step": 4396
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4565399517956296,
+      "learning_rate": 2.769111080810355e-05,
+      "loss": 0.6755,
+      "step": 4397
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.43236451226993167,
+      "learning_rate": 2.7685763859533463e-05,
+      "loss": 0.7027,
+      "step": 4398
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.44377126984020643,
+      "learning_rate": 2.768041626636764e-05,
+      "loss": 0.6804,
+      "step": 4399
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.45357590958083743,
+      "learning_rate": 2.7675068029054583e-05,
+      "loss": 0.6871,
+      "step": 4400
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4857739870891743,
+      "learning_rate": 2.766971914804284e-05,
+      "loss": 0.7145,
+      "step": 4401
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5328277250608056,
+      "learning_rate": 2.7664369623781016e-05,
+      "loss": 0.7354,
+      "step": 4402
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.643654659149366,
+      "learning_rate": 2.7659019456717762e-05,
+      "loss": 0.7239,
+      "step": 4403
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4022840646941321,
+      "learning_rate": 2.7653668647301797e-05,
+      "loss": 0.6658,
+      "step": 4404
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.45545010609571324,
+      "learning_rate": 2.7648317195981893e-05,
+      "loss": 0.7138,
+      "step": 4405
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.41372844280809096,
+      "learning_rate": 2.764296510320685e-05,
+      "loss": 0.6193,
+      "step": 4406
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.45998881668192787,
+      "learning_rate": 2.763761236942556e-05,
+      "loss": 0.773,
+      "step": 4407
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.39099554867139785,
+      "learning_rate": 2.7632258995086952e-05,
+      "loss": 0.6685,
+      "step": 4408
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4770312723222404,
+      "learning_rate": 2.7626904980639993e-05,
+      "loss": 0.7419,
+      "step": 4409
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4502095469434096,
+      "learning_rate": 2.7621550326533722e-05,
+      "loss": 0.6851,
+      "step": 4410
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3936576887851466,
+      "learning_rate": 2.7616195033217235e-05,
+      "loss": 0.6631,
+      "step": 4411
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4212031422730891,
+      "learning_rate": 2.7610839101139668e-05,
+      "loss": 0.7307,
+      "step": 4412
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.38675534933040334,
+      "learning_rate": 2.760548253075021e-05,
+      "loss": 0.6747,
+      "step": 4413
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.47485856788917385,
+      "learning_rate": 2.7600125322498127e-05,
+      "loss": 0.7675,
+      "step": 4414
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4763515881041056,
+      "learning_rate": 2.759476747683271e-05,
+      "loss": 0.7089,
+      "step": 4415
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3715983352421246,
+      "learning_rate": 2.758940899420331e-05,
+      "loss": 0.6867,
+      "step": 4416
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4673868170338077,
+      "learning_rate": 2.7584049875059346e-05,
+      "loss": 0.6358,
+      "step": 4417
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.48470674694753646,
+      "learning_rate": 2.7578690119850283e-05,
+      "loss": 0.7348,
+      "step": 4418
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5259237857635927,
+      "learning_rate": 2.7573329729025627e-05,
+      "loss": 0.7282,
+      "step": 4419
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6173922356322158,
+      "learning_rate": 2.7567968703034954e-05,
+      "loss": 0.7571,
+      "step": 4420
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.48413844997669825,
+      "learning_rate": 2.7562607042327888e-05,
+      "loss": 0.7178,
+      "step": 4421
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4697055785875091,
+      "learning_rate": 2.75572447473541e-05,
+      "loss": 0.709,
+      "step": 4422
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4573003679568581,
+      "learning_rate": 2.7551881818563324e-05,
+      "loss": 0.714,
+      "step": 4423
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.46388548809775537,
+      "learning_rate": 2.754651825640534e-05,
+      "loss": 0.711,
+      "step": 4424
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3920602128002772,
+      "learning_rate": 2.7541154061329984e-05,
+      "loss": 0.6905,
+      "step": 4425
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.473354516535292,
+      "learning_rate": 2.7535789233787143e-05,
+      "loss": 0.6914,
+      "step": 4426
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.43085548548503616,
+      "learning_rate": 2.7530423774226758e-05,
+      "loss": 0.7128,
+      "step": 4427
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4344381703929394,
+      "learning_rate": 2.752505768309883e-05,
+      "loss": 0.6807,
+      "step": 4428
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.37770563660364315,
+      "learning_rate": 2.7519690960853402e-05,
+      "loss": 0.6144,
+      "step": 4429
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4946805410182031,
+      "learning_rate": 2.7514323607940566e-05,
+      "loss": 0.6903,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4211699800305512,
+      "learning_rate": 2.7508955624810493e-05,
+      "loss": 0.6804,
+      "step": 4431
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.48408879224130663,
+      "learning_rate": 2.750358701191338e-05,
+      "loss": 0.6861,
+      "step": 4432
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.48606989678281837,
+      "learning_rate": 2.749821776969948e-05,
+      "loss": 0.7026,
+      "step": 4433
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6054255360256147,
+      "learning_rate": 2.7492847898619115e-05,
+      "loss": 0.6797,
+      "step": 4434
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5236633157013705,
+      "learning_rate": 2.7487477399122648e-05,
+      "loss": 0.7498,
+      "step": 4435
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.46847049449801054,
+      "learning_rate": 2.7482106271660494e-05,
+      "loss": 0.7123,
+      "step": 4436
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.41139274860579533,
+      "learning_rate": 2.7476734516683115e-05,
+      "loss": 0.6842,
+      "step": 4437
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5181613795939438,
+      "learning_rate": 2.7471362134641056e-05,
+      "loss": 0.6671,
+      "step": 4438
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.431321600571868,
+      "learning_rate": 2.746598912598487e-05,
+      "loss": 0.6866,
+      "step": 4439
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3643261331613426,
+      "learning_rate": 2.7460615491165187e-05,
+      "loss": 0.6275,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4356281505627021,
+      "learning_rate": 2.7455241230632702e-05,
+      "loss": 0.683,
+      "step": 4441
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4573227533775497,
+      "learning_rate": 2.7449866344838138e-05,
+      "loss": 0.7193,
+      "step": 4442
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5192940675158921,
+      "learning_rate": 2.744449083423227e-05,
+      "loss": 0.6853,
+      "step": 4443
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.42809157948262133,
+      "learning_rate": 2.7439114699265954e-05,
+      "loss": 0.7145,
+      "step": 4444
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.436151667631182,
+      "learning_rate": 2.743373794039008e-05,
+      "loss": 0.6949,
+      "step": 4445
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4320469675816767,
+      "learning_rate": 2.7428360558055572e-05,
+      "loss": 0.6756,
+      "step": 4446
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.491805118271199,
+      "learning_rate": 2.742298255271344e-05,
+      "loss": 0.7563,
+      "step": 4447
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.41580027137212516,
+      "learning_rate": 2.741760392481472e-05,
+      "loss": 0.6624,
+      "step": 4448
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.38217733144012944,
+      "learning_rate": 2.741222467481053e-05,
+      "loss": 0.6775,
+      "step": 4449
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5046291375784815,
+      "learning_rate": 2.7406844803151992e-05,
+      "loss": 0.6882,
+      "step": 4450
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4446672989507128,
+      "learning_rate": 2.7401464310290334e-05,
+      "loss": 0.7102,
+      "step": 4451
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.408866093306861,
+      "learning_rate": 2.7396083196676805e-05,
+      "loss": 0.7327,
+      "step": 4452
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.37184642533231566,
+      "learning_rate": 2.7390701462762703e-05,
+      "loss": 0.682,
+      "step": 4453
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.43793983303867057,
+      "learning_rate": 2.7385319108999407e-05,
+      "loss": 0.6967,
+      "step": 4454
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3822831656457298,
+      "learning_rate": 2.7379936135838305e-05,
+      "loss": 0.6758,
+      "step": 4455
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.63359712259788,
+      "learning_rate": 2.737455254373088e-05,
+      "loss": 7.6504,
+      "step": 4456
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.395250206287121,
+      "learning_rate": 2.7369168333128634e-05,
+      "loss": 0.6922,
+      "step": 4457
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.39507239817361095,
+      "learning_rate": 2.7363783504483137e-05,
+      "loss": 0.6981,
+      "step": 4458
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5592955386592522,
+      "learning_rate": 2.7358398058246017e-05,
+      "loss": 0.721,
+      "step": 4459
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.42260574946735235,
+      "learning_rate": 2.7353011994868935e-05,
+      "loss": 0.7149,
+      "step": 4460
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.41404785052498544,
+      "learning_rate": 2.734762531480362e-05,
+      "loss": 0.6787,
+      "step": 4461
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4244274485268712,
+      "learning_rate": 2.7342238018501843e-05,
+      "loss": 0.6653,
+      "step": 4462
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5530335558663972,
+      "learning_rate": 2.7336850106415428e-05,
+      "loss": 0.6933,
+      "step": 4463
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.45909277438156587,
+      "learning_rate": 2.733146157899626e-05,
+      "loss": 0.7112,
+      "step": 4464
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4224251516030201,
+      "learning_rate": 2.7326072436696258e-05,
+      "loss": 0.6657,
+      "step": 4465
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3619896937068763,
+      "learning_rate": 2.7320682679967413e-05,
+      "loss": 0.6759,
+      "step": 4466
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.37610185944048263,
+      "learning_rate": 2.731529230926175e-05,
+      "loss": 0.6687,
+      "step": 4467
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.460408715611993,
+      "learning_rate": 2.7309901325031357e-05,
+      "loss": 0.6502,
+      "step": 4468
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3990352734109636,
+      "learning_rate": 2.730450972772837e-05,
+      "loss": 0.6701,
+      "step": 4469
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4540952476220776,
+      "learning_rate": 2.7299117517804967e-05,
+      "loss": 0.693,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4362551017834005,
+      "learning_rate": 2.72937246957134e-05,
+      "loss": 0.7044,
+      "step": 4471
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4333415297694066,
+      "learning_rate": 2.7288331261905944e-05,
+      "loss": 0.6522,
+      "step": 4472
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4742953609704774,
+      "learning_rate": 2.7282937216834956e-05,
+      "loss": 0.7049,
+      "step": 4473
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 4.041549020483168,
+      "learning_rate": 2.7277542560952813e-05,
+      "loss": 7.6298,
+      "step": 4474
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5697589026827382,
+      "learning_rate": 2.7272147294711966e-05,
+      "loss": 0.647,
+      "step": 4475
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4281095666867797,
+      "learning_rate": 2.726675141856491e-05,
+      "loss": 0.6995,
+      "step": 4476
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.47855739240040684,
+      "learning_rate": 2.7261354932964183e-05,
+      "loss": 0.6996,
+      "step": 4477
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.509665339463006,
+      "learning_rate": 2.7255957838362393e-05,
+      "loss": 0.7237,
+      "step": 4478
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.36556591535673466,
+      "learning_rate": 2.725056013521218e-05,
+      "loss": 0.6815,
+      "step": 4479
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5064661174528388,
+      "learning_rate": 2.7245161823966242e-05,
+      "loss": 0.7125,
+      "step": 4480
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4255341827519882,
+      "learning_rate": 2.723976290507734e-05,
+      "loss": 0.7156,
+      "step": 4481
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4130486067292804,
+      "learning_rate": 2.7234363378998255e-05,
+      "loss": 0.6682,
+      "step": 4482
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.45961542114514947,
+      "learning_rate": 2.7228963246181864e-05,
+      "loss": 0.7178,
+      "step": 4483
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.37258904828089573,
+      "learning_rate": 2.7223562507081042e-05,
+      "loss": 0.6299,
+      "step": 4484
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.43690374671556786,
+      "learning_rate": 2.7218161162148767e-05,
+      "loss": 0.6953,
+      "step": 4485
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4827264259252175,
+      "learning_rate": 2.7212759211838027e-05,
+      "loss": 0.6669,
+      "step": 4486
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4071243646847685,
+      "learning_rate": 2.720735665660188e-05,
+      "loss": 0.6832,
+      "step": 4487
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5650599910852216,
+      "learning_rate": 2.7201953496893443e-05,
+      "loss": 0.659,
+      "step": 4488
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4396115417033345,
+      "learning_rate": 2.719654973316586e-05,
+      "loss": 0.6852,
+      "step": 4489
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.44734892976826024,
+      "learning_rate": 2.7191145365872342e-05,
+      "loss": 0.7195,
+      "step": 4490
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.394648877721056,
+      "learning_rate": 2.7185740395466143e-05,
+      "loss": 0.6896,
+      "step": 4491
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.240463911688562,
+      "learning_rate": 2.7180334822400586e-05,
+      "loss": 7.3071,
+      "step": 4492
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5447307897899557,
+      "learning_rate": 2.7174928647129007e-05,
+      "loss": 0.6718,
+      "step": 4493
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.44190129014021357,
+      "learning_rate": 2.7169521870104833e-05,
+      "loss": 0.6889,
+      "step": 4494
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3966286802193353,
+      "learning_rate": 2.7164114491781516e-05,
+      "loss": 0.6762,
+      "step": 4495
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5615097737144494,
+      "learning_rate": 2.7158706512612574e-05,
+      "loss": 0.6919,
+      "step": 4496
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.505707149472963,
+      "learning_rate": 2.7153297933051557e-05,
+      "loss": 0.6827,
+      "step": 4497
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3899228757823971,
+      "learning_rate": 2.7147888753552086e-05,
+      "loss": 0.6658,
+      "step": 4498
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4117369330203868,
+      "learning_rate": 2.7142478974567814e-05,
+      "loss": 0.6793,
+      "step": 4499
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4291408499090631,
+      "learning_rate": 2.7137068596552458e-05,
+      "loss": 0.6647,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4439613266785471,
+      "learning_rate": 2.7131657619959774e-05,
+      "loss": 0.6665,
+      "step": 4501
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.42511153635652593,
+      "learning_rate": 2.7126246045243584e-05,
+      "loss": 0.7023,
+      "step": 4502
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.40649958768372824,
+      "learning_rate": 2.712083387285774e-05,
+      "loss": 0.6956,
+      "step": 4503
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.40850008594014325,
+      "learning_rate": 2.711542110325616e-05,
+      "loss": 0.6983,
+      "step": 4504
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4368798389708804,
+      "learning_rate": 2.7110007736892808e-05,
+      "loss": 0.6493,
+      "step": 4505
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3775442659525542,
+      "learning_rate": 2.7104593774221696e-05,
+      "loss": 0.6484,
+      "step": 4506
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5145356181587647,
+      "learning_rate": 2.7099179215696878e-05,
+      "loss": 0.7133,
+      "step": 4507
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4381894291472329,
+      "learning_rate": 2.7093764061772473e-05,
+      "loss": 0.6617,
+      "step": 4508
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.500093276569577,
+      "learning_rate": 2.708834831290265e-05,
+      "loss": 0.742,
+      "step": 4509
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.43581220902451134,
+      "learning_rate": 2.7082931969541605e-05,
+      "loss": 0.7424,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.40449936845519685,
+      "learning_rate": 2.7077515032143617e-05,
+      "loss": 0.6835,
+      "step": 4511
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4123383044632193,
+      "learning_rate": 2.7072097501162987e-05,
+      "loss": 0.7018,
+      "step": 4512
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.49271900593808526,
+      "learning_rate": 2.7066679377054083e-05,
+      "loss": 0.7025,
+      "step": 4513
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4186418923930494,
+      "learning_rate": 2.706126066027131e-05,
+      "loss": 0.6471,
+      "step": 4514
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.36834706369337344,
+      "learning_rate": 2.705584135126914e-05,
+      "loss": 0.6676,
+      "step": 4515
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4037664571587978,
+      "learning_rate": 2.7050421450502074e-05,
+      "loss": 0.7137,
+      "step": 4516
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.40776612518280464,
+      "learning_rate": 2.7045000958424674e-05,
+      "loss": 0.6684,
+      "step": 4517
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.37965822073541444,
+      "learning_rate": 2.703957987549155e-05,
+      "loss": 0.6251,
+      "step": 4518
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4591724490874376,
+      "learning_rate": 2.703415820215737e-05,
+      "loss": 0.7125,
+      "step": 4519
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3886771584062771,
+      "learning_rate": 2.702873593887683e-05,
+      "loss": 0.6668,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.49461571816255906,
+      "learning_rate": 2.70233130861047e-05,
+      "loss": 0.7181,
+      "step": 4521
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.421686567045417,
+      "learning_rate": 2.701788964429578e-05,
+      "loss": 0.6995,
+      "step": 4522
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3686444173733964,
+      "learning_rate": 2.7012465613904937e-05,
+      "loss": 0.6742,
+      "step": 4523
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3900001859548111,
+      "learning_rate": 2.7007040995387065e-05,
+      "loss": 0.7113,
+      "step": 4524
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.42209266349342145,
+      "learning_rate": 2.7001615789197128e-05,
+      "loss": 0.7046,
+      "step": 4525
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3475416368289195,
+      "learning_rate": 2.6996189995790138e-05,
+      "loss": 0.6537,
+      "step": 4526
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3779940512311007,
+      "learning_rate": 2.6990763615621135e-05,
+      "loss": 0.7176,
+      "step": 4527
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4925148237763519,
+      "learning_rate": 2.698533664914523e-05,
+      "loss": 0.7225,
+      "step": 4528
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.36626854128143693,
+      "learning_rate": 2.697990909681758e-05,
+      "loss": 0.6622,
+      "step": 4529
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4774517893214914,
+      "learning_rate": 2.6974480959093386e-05,
+      "loss": 0.7275,
+      "step": 4530
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4141333451085937,
+      "learning_rate": 2.6969052236427886e-05,
+      "loss": 0.669,
+      "step": 4531
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.46443620343406905,
+      "learning_rate": 2.69636229292764e-05,
+      "loss": 0.6899,
+      "step": 4532
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.37960771013661493,
+      "learning_rate": 2.6958193038094273e-05,
+      "loss": 0.6862,
+      "step": 4533
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4153965570751294,
+      "learning_rate": 2.6952762563336893e-05,
+      "loss": 0.7079,
+      "step": 4534
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4378016558937292,
+      "learning_rate": 2.694733150545971e-05,
+      "loss": 0.6824,
+      "step": 4535
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.467591211955216,
+      "learning_rate": 2.6941899864918232e-05,
+      "loss": 0.6936,
+      "step": 4536
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3955380626505945,
+      "learning_rate": 2.6936467642167992e-05,
+      "loss": 0.719,
+      "step": 4537
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4463114075412188,
+      "learning_rate": 2.6931034837664583e-05,
+      "loss": 0.7044,
+      "step": 4538
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.38540578530542846,
+      "learning_rate": 2.6925601451863662e-05,
+      "loss": 0.7173,
+      "step": 4539
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3806998221020038,
+      "learning_rate": 2.692016748522091e-05,
+      "loss": 0.6493,
+      "step": 4540
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4550257543325201,
+      "learning_rate": 2.691473293819206e-05,
+      "loss": 0.7169,
+      "step": 4541
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4546301186403844,
+      "learning_rate": 2.690929781123292e-05,
+      "loss": 0.7324,
+      "step": 4542
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.393580478650384,
+      "learning_rate": 2.6903862104799312e-05,
+      "loss": 0.6836,
+      "step": 4543
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3718002574956781,
+      "learning_rate": 2.6898425819347128e-05,
+      "loss": 0.6459,
+      "step": 4544
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.37340212947614193,
+      "learning_rate": 2.6892988955332295e-05,
+      "loss": 0.6706,
+      "step": 4545
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.37597813588458334,
+      "learning_rate": 2.688755151321081e-05,
+      "loss": 0.6482,
+      "step": 4546
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3862761094879225,
+      "learning_rate": 2.6882113493438698e-05,
+      "loss": 0.6435,
+      "step": 4547
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.37793178423333423,
+      "learning_rate": 2.6876674896472033e-05,
+      "loss": 0.7003,
+      "step": 4548
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3850104006582654,
+      "learning_rate": 2.6871235722766956e-05,
+      "loss": 0.6521,
+      "step": 4549
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.38177379986066895,
+      "learning_rate": 2.686579597277964e-05,
+      "loss": 0.6854,
+      "step": 4550
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.43066376284383157,
+      "learning_rate": 2.6860355646966297e-05,
+      "loss": 0.7004,
+      "step": 4551
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4104463551063939,
+      "learning_rate": 2.6854914745783217e-05,
+      "loss": 0.6685,
+      "step": 4552
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.39401965358169483,
+      "learning_rate": 2.6849473269686716e-05,
+      "loss": 0.6951,
+      "step": 4553
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3620931114951165,
+      "learning_rate": 2.6844031219133164e-05,
+      "loss": 0.6802,
+      "step": 4554
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4654350983575318,
+      "learning_rate": 2.6838588594578973e-05,
+      "loss": 0.6962,
+      "step": 4555
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4146766682250426,
+      "learning_rate": 2.6833145396480627e-05,
+      "loss": 0.7029,
+      "step": 4556
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4499013184334551,
+      "learning_rate": 2.6827701625294623e-05,
+      "loss": 0.6595,
+      "step": 4557
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7887843982382167,
+      "learning_rate": 2.6822257281477523e-05,
+      "loss": 0.6879,
+      "step": 4558
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4352587624336297,
+      "learning_rate": 2.6816812365485955e-05,
+      "loss": 0.7036,
+      "step": 4559
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3896294672740719,
+      "learning_rate": 2.6811366877776563e-05,
+      "loss": 0.7042,
+      "step": 4560
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.40293842454236845,
+      "learning_rate": 2.6805920818806054e-05,
+      "loss": 0.6812,
+      "step": 4561
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.343246524986533,
+      "learning_rate": 2.680047418903119e-05,
+      "loss": 0.6362,
+      "step": 4562
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.447465172835536,
+      "learning_rate": 2.6795026988908765e-05,
+      "loss": 0.6994,
+      "step": 4563
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.46113794270393277,
+      "learning_rate": 2.6789579218895636e-05,
+      "loss": 0.7114,
+      "step": 4564
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.41772225705494453,
+      "learning_rate": 2.678413087944869e-05,
+      "loss": 0.7039,
+      "step": 4565
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.37728967862047286,
+      "learning_rate": 2.6778681971024886e-05,
+      "loss": 0.7353,
+      "step": 4566
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.37338003943755893,
+      "learning_rate": 2.6773232494081214e-05,
+      "loss": 0.6735,
+      "step": 4567
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4010962451798855,
+      "learning_rate": 2.6767782449074706e-05,
+      "loss": 0.6504,
+      "step": 4568
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.46498614125346743,
+      "learning_rate": 2.6762331836462458e-05,
+      "loss": 0.7284,
+      "step": 4569
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.39862331550080055,
+      "learning_rate": 2.6756880656701615e-05,
+      "loss": 0.6746,
+      "step": 4570
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.41739479045998285,
+      "learning_rate": 2.6751428910249343e-05,
+      "loss": 0.6415,
+      "step": 4571
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.39387118525946585,
+      "learning_rate": 2.6745976597562885e-05,
+      "loss": 0.6646,
+      "step": 4572
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4340744954425488,
+      "learning_rate": 2.6740523719099516e-05,
+      "loss": 0.7282,
+      "step": 4573
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.48651666618418415,
+      "learning_rate": 2.6735070275316566e-05,
+      "loss": 0.6699,
+      "step": 4574
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4175611021518981,
+      "learning_rate": 2.6729616266671406e-05,
+      "loss": 0.6967,
+      "step": 4575
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4079318776918292,
+      "learning_rate": 2.672416169362145e-05,
+      "loss": 0.6775,
+      "step": 4576
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.40743123610675647,
+      "learning_rate": 2.671870655662418e-05,
+      "loss": 0.7046,
+      "step": 4577
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.46870293097907456,
+      "learning_rate": 2.6713250856137102e-05,
+      "loss": 0.7023,
+      "step": 4578
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.38606943837432683,
+      "learning_rate": 2.670779459261778e-05,
+      "loss": 0.6962,
+      "step": 4579
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.40516457321815313,
+      "learning_rate": 2.6702337766523832e-05,
+      "loss": 0.6939,
+      "step": 4580
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4058267770310165,
+      "learning_rate": 2.669688037831291e-05,
+      "loss": 0.641,
+      "step": 4581
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.42638840792655985,
+      "learning_rate": 2.669142242844272e-05,
+      "loss": 0.7086,
+      "step": 4582
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3940863748298331,
+      "learning_rate": 2.6685963917371004e-05,
+      "loss": 0.6436,
+      "step": 4583
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5139156140559834,
+      "learning_rate": 2.6680504845555577e-05,
+      "loss": 0.7131,
+      "step": 4584
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4483449815082185,
+      "learning_rate": 2.6675045213454265e-05,
+      "loss": 0.7217,
+      "step": 4585
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.42111255412659476,
+      "learning_rate": 2.6669585021524983e-05,
+      "loss": 0.7005,
+      "step": 4586
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3803686838424539,
+      "learning_rate": 2.6664124270225662e-05,
+      "loss": 0.6812,
+      "step": 4587
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4172363051168431,
+      "learning_rate": 2.665866296001428e-05,
+      "loss": 0.6502,
+      "step": 4588
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4290709722347414,
+      "learning_rate": 2.665320109134888e-05,
+      "loss": 0.693,
+      "step": 4589
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3730027137707393,
+      "learning_rate": 2.6647738664687536e-05,
+      "loss": 0.6575,
+      "step": 4590
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3987692365539463,
+      "learning_rate": 2.6642275680488383e-05,
+      "loss": 0.6568,
+      "step": 4591
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.33501372392622836,
+      "learning_rate": 2.663681213920959e-05,
+      "loss": 0.6439,
+      "step": 4592
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3690562050536763,
+      "learning_rate": 2.6631348041309375e-05,
+      "loss": 0.6664,
+      "step": 4593
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4022168689677141,
+      "learning_rate": 2.662588338724601e-05,
+      "loss": 0.708,
+      "step": 4594
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4870806847107952,
+      "learning_rate": 2.6620418177477807e-05,
+      "loss": 0.71,
+      "step": 4595
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.514048968614532,
+      "learning_rate": 2.6614952412463124e-05,
+      "loss": 0.6955,
+      "step": 4596
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4404350708606141,
+      "learning_rate": 2.660948609266038e-05,
+      "loss": 0.6711,
+      "step": 4597
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.43613694848483026,
+      "learning_rate": 2.6604019218528012e-05,
+      "loss": 0.7146,
+      "step": 4598
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.38791255031877203,
+      "learning_rate": 2.6598551790524532e-05,
+      "loss": 0.7213,
+      "step": 4599
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4402291885317376,
+      "learning_rate": 2.6593083809108478e-05,
+      "loss": 0.7051,
+      "step": 4600
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.39624407306956844,
+      "learning_rate": 2.6587615274738454e-05,
+      "loss": 0.7091,
+      "step": 4601
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4592600861040283,
+      "learning_rate": 2.6582146187873087e-05,
+      "loss": 0.7366,
+      "step": 4602
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4117493539729781,
+      "learning_rate": 2.6576676548971066e-05,
+      "loss": 0.725,
+      "step": 4603
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9894342600687922,
+      "learning_rate": 2.657120635849113e-05,
+      "loss": 0.6804,
+      "step": 4604
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.43765789970668856,
+      "learning_rate": 2.6565735616892054e-05,
+      "loss": 0.6924,
+      "step": 4605
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4526716547298382,
+      "learning_rate": 2.656026432463266e-05,
+      "loss": 0.694,
+      "step": 4606
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.39612939846518214,
+      "learning_rate": 2.6554792482171818e-05,
+      "loss": 0.6591,
+      "step": 4607
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5156185545911858,
+      "learning_rate": 2.654932008996845e-05,
+      "loss": 0.6627,
+      "step": 4608
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.41207872663806,
+      "learning_rate": 2.654384714848151e-05,
+      "loss": 0.6463,
+      "step": 4609
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.38617331086157836,
+      "learning_rate": 2.6538373658170017e-05,
+      "loss": 0.681,
+      "step": 4610
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4106801297566234,
+      "learning_rate": 2.653289961949302e-05,
+      "loss": 0.7025,
+      "step": 4611
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5088248029049676,
+      "learning_rate": 2.6527425032909623e-05,
+      "loss": 0.6959,
+      "step": 4612
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.43779175510766327,
+      "learning_rate": 2.6521949898878975e-05,
+      "loss": 0.6791,
+      "step": 4613
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.39792069145201064,
+      "learning_rate": 2.651647421786026e-05,
+      "loss": 0.6803,
+      "step": 4614
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4835512512988913,
+      "learning_rate": 2.651099799031272e-05,
+      "loss": 0.6884,
+      "step": 4615
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.41466070622831613,
+      "learning_rate": 2.6505521216695653e-05,
+      "loss": 0.6683,
+      "step": 4616
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.37650051667819256,
+      "learning_rate": 2.6500043897468367e-05,
+      "loss": 0.694,
+      "step": 4617
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4131434620917964,
+      "learning_rate": 2.6494566033090256e-05,
+      "loss": 0.6974,
+      "step": 4618
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4518197399515573,
+      "learning_rate": 2.6489087624020727e-05,
+      "loss": 0.7124,
+      "step": 4619
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.39403554872235774,
+      "learning_rate": 2.648360867071927e-05,
+      "loss": 0.6659,
+      "step": 4620
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5564809761887783,
+      "learning_rate": 2.6478129173645372e-05,
+      "loss": 7.3731,
+      "step": 4621
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3492073706037215,
+      "learning_rate": 2.6472649133258603e-05,
+      "loss": 0.6101,
+      "step": 4622
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3908357383730995,
+      "learning_rate": 2.6467168550018576e-05,
+      "loss": 0.6949,
+      "step": 4623
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3865218535445385,
+      "learning_rate": 2.6461687424384925e-05,
+      "loss": 0.6584,
+      "step": 4624
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4371466676165781,
+      "learning_rate": 2.6456205756817365e-05,
+      "loss": 0.7481,
+      "step": 4625
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.36755546240497156,
+      "learning_rate": 2.645072354777561e-05,
+      "loss": 0.6502,
+      "step": 4626
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3755753521545814,
+      "learning_rate": 2.6445240797719477e-05,
+      "loss": 0.6777,
+      "step": 4627
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3601255351406925,
+      "learning_rate": 2.643975750710877e-05,
+      "loss": 0.6554,
+      "step": 4628
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5136106881124031,
+      "learning_rate": 2.6434273676403383e-05,
+      "loss": 0.6758,
+      "step": 4629
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3991948225957597,
+      "learning_rate": 2.6428789306063233e-05,
+      "loss": 0.6604,
+      "step": 4630
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3921027147427486,
+      "learning_rate": 2.642330439654829e-05,
+      "loss": 0.6807,
+      "step": 4631
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4098595619314601,
+      "learning_rate": 2.6417818948318557e-05,
+      "loss": 0.6901,
+      "step": 4632
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3654008141244112,
+      "learning_rate": 2.641233296183411e-05,
+      "loss": 0.6887,
+      "step": 4633
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.40053746677182445,
+      "learning_rate": 2.640684643755504e-05,
+      "loss": 0.672,
+      "step": 4634
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3652721054051023,
+      "learning_rate": 2.6401359375941495e-05,
+      "loss": 0.687,
+      "step": 4635
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.437277422182518,
+      "learning_rate": 2.6395871777453666e-05,
+      "loss": 0.7058,
+      "step": 4636
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.373739839422572,
+      "learning_rate": 2.63903836425518e-05,
+      "loss": 0.6576,
+      "step": 4637
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.38262672322756264,
+      "learning_rate": 2.6384894971696176e-05,
+      "loss": 0.67,
+      "step": 4638
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4147294480758352,
+      "learning_rate": 2.6379405765347113e-05,
+      "loss": 0.678,
+      "step": 4639
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4612569239339981,
+      "learning_rate": 2.6373916023965004e-05,
+      "loss": 0.6966,
+      "step": 4640
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.801534340859204,
+      "learning_rate": 2.636842574801025e-05,
+      "loss": 7.2879,
+      "step": 4641
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39425501827546994,
+      "learning_rate": 2.636293493794332e-05,
+      "loss": 0.7186,
+      "step": 4642
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3959489501529634,
+      "learning_rate": 2.6357443594224723e-05,
+      "loss": 0.6972,
+      "step": 4643
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3141153890862177,
+      "learning_rate": 2.6351951717315012e-05,
+      "loss": 0.6134,
+      "step": 4644
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4215592848252636,
+      "learning_rate": 2.6346459307674774e-05,
+      "loss": 0.6805,
+      "step": 4645
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4225490091758521,
+      "learning_rate": 2.6340966365764664e-05,
+      "loss": 0.6859,
+      "step": 4646
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39731581069532096,
+      "learning_rate": 2.633547289204536e-05,
+      "loss": 0.6977,
+      "step": 4647
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.34319882812370245,
+      "learning_rate": 2.6329978886977595e-05,
+      "loss": 0.6359,
+      "step": 4648
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3526073437546563,
+      "learning_rate": 2.632448435102215e-05,
+      "loss": 0.6509,
+      "step": 4649
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3986546849329614,
+      "learning_rate": 2.631898928463984e-05,
+      "loss": 0.7186,
+      "step": 4650
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4187522679600893,
+      "learning_rate": 2.6313493688291535e-05,
+      "loss": 0.7189,
+      "step": 4651
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.36156843672296807,
+      "learning_rate": 2.630799756243813e-05,
+      "loss": 0.6845,
+      "step": 4652
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39174299808139823,
+      "learning_rate": 2.6302500907540584e-05,
+      "loss": 0.661,
+      "step": 4653
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.46047688867647657,
+      "learning_rate": 2.629700372405991e-05,
+      "loss": 0.6654,
+      "step": 4654
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.41597823140811085,
+      "learning_rate": 2.629150601245713e-05,
+      "loss": 0.726,
+      "step": 4655
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4634930255258862,
+      "learning_rate": 2.628600777319334e-05,
+      "loss": 0.7335,
+      "step": 4656
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3649432363784877,
+      "learning_rate": 2.6280509006729675e-05,
+      "loss": 0.6944,
+      "step": 4657
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.37207728959492803,
+      "learning_rate": 2.6275009713527303e-05,
+      "loss": 0.6915,
+      "step": 4658
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.38029017203229765,
+      "learning_rate": 2.626950989404744e-05,
+      "loss": 0.6621,
+      "step": 4659
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.41939730403145015,
+      "learning_rate": 2.6264009548751358e-05,
+      "loss": 0.6911,
+      "step": 4660
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.46402319762367095,
+      "learning_rate": 2.6258508678100362e-05,
+      "loss": 0.673,
+      "step": 4661
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6334119197889433,
+      "learning_rate": 2.6253007282555795e-05,
+      "loss": 0.6675,
+      "step": 4662
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.35906344270853463,
+      "learning_rate": 2.624750536257906e-05,
+      "loss": 0.6963,
+      "step": 4663
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.37761091111998407,
+      "learning_rate": 2.62420029186316e-05,
+      "loss": 0.6498,
+      "step": 4664
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3602834981460067,
+      "learning_rate": 2.6236499951174895e-05,
+      "loss": 0.6574,
+      "step": 4665
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.42632598007296374,
+      "learning_rate": 2.623099646067046e-05,
+      "loss": 0.6967,
+      "step": 4666
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4213207710990043,
+      "learning_rate": 2.6225492447579886e-05,
+      "loss": 0.681,
+      "step": 4667
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3766401140934084,
+      "learning_rate": 2.621998791236478e-05,
+      "loss": 0.686,
+      "step": 4668
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4190096012760872,
+      "learning_rate": 2.6214482855486795e-05,
+      "loss": 0.7009,
+      "step": 4669
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4571345741716837,
+      "learning_rate": 2.620897727740764e-05,
+      "loss": 0.7577,
+      "step": 4670
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3489377407251144,
+      "learning_rate": 2.620347117858906e-05,
+      "loss": 0.662,
+      "step": 4671
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.43664015293141956,
+      "learning_rate": 2.6197964559492845e-05,
+      "loss": 0.6891,
+      "step": 4672
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39025285931346954,
+      "learning_rate": 2.6192457420580823e-05,
+      "loss": 0.6696,
+      "step": 4673
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4035533231895854,
+      "learning_rate": 2.618694976231488e-05,
+      "loss": 0.6687,
+      "step": 4674
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3600427990278606,
+      "learning_rate": 2.618144158515693e-05,
+      "loss": 0.6916,
+      "step": 4675
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.35972672259542937,
+      "learning_rate": 2.6175932889568935e-05,
+      "loss": 0.6367,
+      "step": 4676
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.40633215883080925,
+      "learning_rate": 2.6170423676012915e-05,
+      "loss": 0.6469,
+      "step": 4677
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3988830175935536,
+      "learning_rate": 2.6164913944950918e-05,
+      "loss": 0.6834,
+      "step": 4678
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3798761680542349,
+      "learning_rate": 2.615940369684502e-05,
+      "loss": 0.6769,
+      "step": 4679
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3670577840970225,
+      "learning_rate": 2.6153892932157382e-05,
+      "loss": 0.6971,
+      "step": 4680
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39275224715841783,
+      "learning_rate": 2.6148381651350176e-05,
+      "loss": 0.6495,
+      "step": 4681
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.40614775923941193,
+      "learning_rate": 2.6142869854885626e-05,
+      "loss": 0.7072,
+      "step": 4682
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3926007376644689,
+      "learning_rate": 2.613735754322599e-05,
+      "loss": 0.6772,
+      "step": 4683
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.40349589730871427,
+      "learning_rate": 2.6131844716833595e-05,
+      "loss": 0.7458,
+      "step": 4684
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.36563790895280557,
+      "learning_rate": 2.6126331376170794e-05,
+      "loss": 0.6843,
+      "step": 4685
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.38374518433508586,
+      "learning_rate": 2.6120817521699975e-05,
+      "loss": 0.6906,
+      "step": 4686
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3590930286683697,
+      "learning_rate": 2.6115303153883585e-05,
+      "loss": 0.6804,
+      "step": 4687
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.42574753256404346,
+      "learning_rate": 2.6109788273184103e-05,
+      "loss": 0.6941,
+      "step": 4688
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3768176425802136,
+      "learning_rate": 2.6104272880064058e-05,
+      "loss": 0.6402,
+      "step": 4689
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.407194309711068,
+      "learning_rate": 2.6098756974986013e-05,
+      "loss": 0.7319,
+      "step": 4690
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39576225747403476,
+      "learning_rate": 2.6093240558412595e-05,
+      "loss": 0.6814,
+      "step": 4691
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4119665151855067,
+      "learning_rate": 2.608772363080645e-05,
+      "loss": 0.667,
+      "step": 4692
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4884682660908241,
+      "learning_rate": 2.6082206192630263e-05,
+      "loss": 0.7017,
+      "step": 4693
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4181292857431316,
+      "learning_rate": 2.60766882443468e-05,
+      "loss": 0.7125,
+      "step": 4694
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6740475767092704,
+      "learning_rate": 2.6071169786418833e-05,
+      "loss": 7.4563,
+      "step": 4695
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3740747268512163,
+      "learning_rate": 2.6065650819309186e-05,
+      "loss": 0.6635,
+      "step": 4696
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3955612240908064,
+      "learning_rate": 2.6060131343480723e-05,
+      "loss": 0.6908,
+      "step": 4697
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.38671369274827494,
+      "learning_rate": 2.6054611359396377e-05,
+      "loss": 0.6809,
+      "step": 4698
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4514884369453111,
+      "learning_rate": 2.6049090867519073e-05,
+      "loss": 0.7416,
+      "step": 4699
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4574885192715481,
+      "learning_rate": 2.6043569868311832e-05,
+      "loss": 0.7277,
+      "step": 4700
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.398823469271929,
+      "learning_rate": 2.603804836223769e-05,
+      "loss": 0.6935,
+      "step": 4701
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3896157465565577,
+      "learning_rate": 2.603252634975972e-05,
+      "loss": 0.7042,
+      "step": 4702
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4103135123585624,
+      "learning_rate": 2.6027003831341047e-05,
+      "loss": 0.6894,
+      "step": 4703
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.32685766707104363,
+      "learning_rate": 2.602148080744484e-05,
+      "loss": 0.66,
+      "step": 4704
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.38776881557915943,
+      "learning_rate": 2.6015957278534315e-05,
+      "loss": 0.6771,
+      "step": 4705
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.33729792465320446,
+      "learning_rate": 2.601043324507271e-05,
+      "loss": 0.623,
+      "step": 4706
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.37600457957287864,
+      "learning_rate": 2.6004908707523332e-05,
+      "loss": 0.7242,
+      "step": 4707
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3934805517997591,
+      "learning_rate": 2.5999383666349517e-05,
+      "loss": 0.676,
+      "step": 4708
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3680101413320166,
+      "learning_rate": 2.5993858122014642e-05,
+      "loss": 0.6709,
+      "step": 4709
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4171226761256253,
+      "learning_rate": 2.5988332074982115e-05,
+      "loss": 0.6977,
+      "step": 4710
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.6231448614207686,
+      "learning_rate": 2.5982805525715415e-05,
+      "loss": 7.4397,
+      "step": 4711
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4103722535097416,
+      "learning_rate": 2.5977278474678046e-05,
+      "loss": 0.6878,
+      "step": 4712
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.46260934225663775,
+      "learning_rate": 2.597175092233354e-05,
+      "loss": 0.6696,
+      "step": 4713
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4247515248311153,
+      "learning_rate": 2.5966222869145497e-05,
+      "loss": 0.6302,
+      "step": 4714
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.681915720062776,
+      "learning_rate": 2.5960694315577555e-05,
+      "loss": 7.2795,
+      "step": 4715
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4471678122350831,
+      "learning_rate": 2.5955165262093372e-05,
+      "loss": 0.6831,
+      "step": 4716
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39618320885518443,
+      "learning_rate": 2.5949635709156677e-05,
+      "loss": 0.6806,
+      "step": 4717
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.37205063450619064,
+      "learning_rate": 2.5944105657231212e-05,
+      "loss": 0.6294,
+      "step": 4718
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39280738600286697,
+      "learning_rate": 2.5938575106780796e-05,
+      "loss": 0.7055,
+      "step": 4719
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3755044701557937,
+      "learning_rate": 2.5933044058269244e-05,
+      "loss": 0.6761,
+      "step": 4720
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4274244600747471,
+      "learning_rate": 2.5927512512160462e-05,
+      "loss": 0.7225,
+      "step": 4721
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3365113804371376,
+      "learning_rate": 2.5921980468918363e-05,
+      "loss": 0.6654,
+      "step": 4722
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.44780069183711546,
+      "learning_rate": 2.591644792900691e-05,
+      "loss": 0.7136,
+      "step": 4723
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3804744332345944,
+      "learning_rate": 2.5910914892890113e-05,
+      "loss": 0.6419,
+      "step": 4724
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.35437323886946415,
+      "learning_rate": 2.590538136103203e-05,
+      "loss": 0.6653,
+      "step": 4725
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.442389447686566,
+      "learning_rate": 2.589984733389674e-05,
+      "loss": 0.7012,
+      "step": 4726
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4136777238760349,
+      "learning_rate": 2.5894312811948378e-05,
+      "loss": 0.6665,
+      "step": 4727
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5961331684820288,
+      "learning_rate": 2.5888777795651118e-05,
+      "loss": 0.6866,
+      "step": 4728
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3623117044888467,
+      "learning_rate": 2.5883242285469187e-05,
+      "loss": 0.6507,
+      "step": 4729
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39568497718841517,
+      "learning_rate": 2.5877706281866816e-05,
+      "loss": 0.6653,
+      "step": 4730
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.35811883207067363,
+      "learning_rate": 2.5872169785308323e-05,
+      "loss": 0.6743,
+      "step": 4731
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3383474749286928,
+      "learning_rate": 2.5866632796258047e-05,
+      "loss": 0.6314,
+      "step": 4732
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4193439933093994,
+      "learning_rate": 2.5861095315180355e-05,
+      "loss": 0.7036,
+      "step": 4733
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.38570825676741655,
+      "learning_rate": 2.5855557342539683e-05,
+      "loss": 0.6628,
+      "step": 4734
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5232497540324832,
+      "learning_rate": 2.585001887880049e-05,
+      "loss": 7.3749,
+      "step": 4735
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4288771246364012,
+      "learning_rate": 2.5844479924427278e-05,
+      "loss": 0.7066,
+      "step": 4736
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3770350735387208,
+      "learning_rate": 2.5838940479884585e-05,
+      "loss": 0.6539,
+      "step": 4737
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.37500135660786454,
+      "learning_rate": 2.583340054563701e-05,
+      "loss": 0.657,
+      "step": 4738
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.34248549146525736,
+      "learning_rate": 2.5827860122149182e-05,
+      "loss": 0.6457,
+      "step": 4739
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.35621162612633034,
+      "learning_rate": 2.5822319209885756e-05,
+      "loss": 0.664,
+      "step": 4740
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4512926522390738,
+      "learning_rate": 2.5816777809311457e-05,
+      "loss": 0.7486,
+      "step": 4741
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.46355340275865226,
+      "learning_rate": 2.5811235920891022e-05,
+      "loss": 0.7085,
+      "step": 4742
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4223122320924337,
+      "learning_rate": 2.580569354508925e-05,
+      "loss": 0.6896,
+      "step": 4743
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.37399947946471723,
+      "learning_rate": 2.5800150682370976e-05,
+      "loss": 0.6687,
+      "step": 4744
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4358533019352591,
+      "learning_rate": 2.579460733320106e-05,
+      "loss": 0.6878,
+      "step": 4745
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3611581913511251,
+      "learning_rate": 2.578906349804443e-05,
+      "loss": 0.6605,
+      "step": 4746
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.35025163232541173,
+      "learning_rate": 2.578351917736604e-05,
+      "loss": 0.6291,
+      "step": 4747
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3738598828439574,
+      "learning_rate": 2.5777974371630876e-05,
+      "loss": 0.6454,
+      "step": 4748
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3982064205203893,
+      "learning_rate": 2.5772429081303982e-05,
+      "loss": 0.6604,
+      "step": 4749
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4545041341417207,
+      "learning_rate": 2.576688330685043e-05,
+      "loss": 0.6872,
+      "step": 4750
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.36509258660840177,
+      "learning_rate": 2.5761337048735343e-05,
+      "loss": 0.6605,
+      "step": 4751
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4332013894576544,
+      "learning_rate": 2.575579030742387e-05,
+      "loss": 0.7291,
+      "step": 4752
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3860207136668821,
+      "learning_rate": 2.5750243083381222e-05,
+      "loss": 0.6889,
+      "step": 4753
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4821493541291304,
+      "learning_rate": 2.5744695377072624e-05,
+      "loss": 0.7157,
+      "step": 4754
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.464346096825784,
+      "learning_rate": 2.573914718896337e-05,
+      "loss": 7.3949,
+      "step": 4755
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4009265365332148,
+      "learning_rate": 2.5733598519518773e-05,
+      "loss": 0.616,
+      "step": 4756
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3650532852550856,
+      "learning_rate": 2.5728049369204185e-05,
+      "loss": 0.6586,
+      "step": 4757
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4163981855998301,
+      "learning_rate": 2.5722499738485018e-05,
+      "loss": 0.6625,
+      "step": 4758
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3818446854760132,
+      "learning_rate": 2.5716949627826707e-05,
+      "loss": 0.705,
+      "step": 4759
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.39434742066207495,
+      "learning_rate": 2.5711399037694735e-05,
+      "loss": 0.7143,
+      "step": 4760
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.37846948881113956,
+      "learning_rate": 2.570584796855463e-05,
+      "loss": 0.6511,
+      "step": 4761
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.40061981073704467,
+      "learning_rate": 2.5700296420871943e-05,
+      "loss": 0.7209,
+      "step": 4762
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4000253275335605,
+      "learning_rate": 2.5694744395112278e-05,
+      "loss": 0.6788,
+      "step": 4763
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.37241155464627357,
+      "learning_rate": 2.5689191891741274e-05,
+      "loss": 0.6285,
+      "step": 4764
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3955159757606591,
+      "learning_rate": 2.5683638911224624e-05,
+      "loss": 0.6846,
+      "step": 4765
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42669433200260865,
+      "learning_rate": 2.567808545402804e-05,
+      "loss": 0.6827,
+      "step": 4766
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.39710653922322536,
+      "learning_rate": 2.5672531520617285e-05,
+      "loss": 0.6637,
+      "step": 4767
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.38754831260287803,
+      "learning_rate": 2.5666977111458162e-05,
+      "loss": 0.6782,
+      "step": 4768
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.366007924572159,
+      "learning_rate": 2.5661422227016514e-05,
+      "loss": 0.6545,
+      "step": 4769
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.41194705558184064,
+      "learning_rate": 2.5655866867758224e-05,
+      "loss": 0.6416,
+      "step": 4770
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.6937129819652414,
+      "learning_rate": 2.56503110341492e-05,
+      "loss": 7.2054,
+      "step": 4771
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.41471188030022543,
+      "learning_rate": 2.5644754726655426e-05,
+      "loss": 0.702,
+      "step": 4772
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4576357307194786,
+      "learning_rate": 2.5639197945742886e-05,
+      "loss": 0.6547,
+      "step": 4773
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3895466328152654,
+      "learning_rate": 2.563364069187762e-05,
+      "loss": 0.6935,
+      "step": 4774
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.41252026675838765,
+      "learning_rate": 2.562808296552572e-05,
+      "loss": 0.6657,
+      "step": 4775
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4313685544647395,
+      "learning_rate": 2.562252476715329e-05,
+      "loss": 0.6644,
+      "step": 4776
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.41582950562773746,
+      "learning_rate": 2.5616966097226503e-05,
+      "loss": 0.6554,
+      "step": 4777
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4700509528316792,
+      "learning_rate": 2.5611406956211554e-05,
+      "loss": 0.7337,
+      "step": 4778
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.49194404645528605,
+      "learning_rate": 2.5605847344574683e-05,
+      "loss": 0.6752,
+      "step": 4779
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4055256288889944,
+      "learning_rate": 2.5600287262782164e-05,
+      "loss": 0.6995,
+      "step": 4780
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4275630554140332,
+      "learning_rate": 2.5594726711300307e-05,
+      "loss": 0.6325,
+      "step": 4781
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42806868797308484,
+      "learning_rate": 2.558916569059549e-05,
+      "loss": 0.6908,
+      "step": 4782
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42180987109301066,
+      "learning_rate": 2.5583604201134087e-05,
+      "loss": 0.6793,
+      "step": 4783
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4042870682714223,
+      "learning_rate": 2.5578042243382547e-05,
+      "loss": 0.6871,
+      "step": 4784
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.39745287888062397,
+      "learning_rate": 2.557247981780734e-05,
+      "loss": 0.6915,
+      "step": 4785
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.37373187200890917,
+      "learning_rate": 2.5566916924874985e-05,
+      "loss": 0.6773,
+      "step": 4786
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.35120681674561144,
+      "learning_rate": 2.5561353565052027e-05,
+      "loss": 0.656,
+      "step": 4787
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.35410710427295766,
+      "learning_rate": 2.555578973880506e-05,
+      "loss": 0.647,
+      "step": 4788
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3693617555952988,
+      "learning_rate": 2.555022544660072e-05,
+      "loss": 0.6771,
+      "step": 4789
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3250082982793328,
+      "learning_rate": 2.554466068890567e-05,
+      "loss": 7.2267,
+      "step": 4790
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4094028953585751,
+      "learning_rate": 2.5539095466186626e-05,
+      "loss": 0.6833,
+      "step": 4791
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3982976639676009,
+      "learning_rate": 2.553352977891034e-05,
+      "loss": 0.644,
+      "step": 4792
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.40084437889097063,
+      "learning_rate": 2.552796362754359e-05,
+      "loss": 0.6561,
+      "step": 4793
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.33689988069153426,
+      "learning_rate": 2.5522397012553204e-05,
+      "loss": 0.62,
+      "step": 4794
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42966052154578127,
+      "learning_rate": 2.5516829934406055e-05,
+      "loss": 0.7257,
+      "step": 4795
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.37711206297203775,
+      "learning_rate": 2.551126239356904e-05,
+      "loss": 0.6344,
+      "step": 4796
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.39720826081357397,
+      "learning_rate": 2.550569439050911e-05,
+      "loss": 0.6662,
+      "step": 4797
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3837951180383426,
+      "learning_rate": 2.550012592569323e-05,
+      "loss": 0.646,
+      "step": 4798
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4319203899058684,
+      "learning_rate": 2.549455699958844e-05,
+      "loss": 0.6963,
+      "step": 4799
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.34060596189532527,
+      "learning_rate": 2.548898761266178e-05,
+      "loss": 0.6495,
+      "step": 4800
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4082620146678765,
+      "learning_rate": 2.5483417765380364e-05,
+      "loss": 0.689,
+      "step": 4801
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.39305275238258097,
+      "learning_rate": 2.5477847458211326e-05,
+      "loss": 0.6928,
+      "step": 4802
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.38144355575192906,
+      "learning_rate": 2.5472276691621837e-05,
+      "loss": 0.7025,
+      "step": 4803
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.35058967844147476,
+      "learning_rate": 2.54667054660791e-05,
+      "loss": 0.6596,
+      "step": 4804
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.34611203452913725,
+      "learning_rate": 2.546113378205039e-05,
+      "loss": 0.666,
+      "step": 4805
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3981237143851215,
+      "learning_rate": 2.5455561640002985e-05,
+      "loss": 0.6905,
+      "step": 4806
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.33587560036458164,
+      "learning_rate": 2.5449989040404202e-05,
+      "loss": 0.652,
+      "step": 4807
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4253801576271215,
+      "learning_rate": 2.5444415983721426e-05,
+      "loss": 0.6978,
+      "step": 4808
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.37293801636653123,
+      "learning_rate": 2.5438842470422064e-05,
+      "loss": 0.6838,
+      "step": 4809
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.390795203402973,
+      "learning_rate": 2.5433268500973546e-05,
+      "loss": 0.6803,
+      "step": 4810
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9349913532191586,
+      "learning_rate": 2.5427694075843358e-05,
+      "loss": 7.3215,
+      "step": 4811
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4459075167340244,
+      "learning_rate": 2.5422119195499026e-05,
+      "loss": 0.7321,
+      "step": 4812
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.43615432833134066,
+      "learning_rate": 2.5416543860408106e-05,
+      "loss": 0.6881,
+      "step": 4813
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4220837424272445,
+      "learning_rate": 2.541096807103819e-05,
+      "loss": 0.704,
+      "step": 4814
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.37249834882259725,
+      "learning_rate": 2.540539182785692e-05,
+      "loss": 0.6849,
+      "step": 4815
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3841916425952141,
+      "learning_rate": 2.5399815131331965e-05,
+      "loss": 0.6659,
+      "step": 4816
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.28026195563124456,
+      "learning_rate": 2.539423798193104e-05,
+      "loss": 0.6168,
+      "step": 4817
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.35281792770599335,
+      "learning_rate": 2.5388660380121885e-05,
+      "loss": 0.6845,
+      "step": 4818
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3376426200749011,
+      "learning_rate": 2.5383082326372296e-05,
+      "loss": 0.6385,
+      "step": 4819
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3420045280348983,
+      "learning_rate": 2.537750382115009e-05,
+      "loss": 0.6579,
+      "step": 4820
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.37464023042108935,
+      "learning_rate": 2.5371924864923126e-05,
+      "loss": 0.6824,
+      "step": 4821
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.41763994688423717,
+      "learning_rate": 2.5366345458159326e-05,
+      "loss": 0.6659,
+      "step": 4822
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4140418301570985,
+      "learning_rate": 2.5360765601326606e-05,
+      "loss": 0.7127,
+      "step": 4823
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4053483397702792,
+      "learning_rate": 2.5355185294892945e-05,
+      "loss": 0.706,
+      "step": 4824
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4021552748526867,
+      "learning_rate": 2.5349604539326365e-05,
+      "loss": 0.6414,
+      "step": 4825
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42639799105465503,
+      "learning_rate": 2.5344023335094914e-05,
+      "loss": 0.681,
+      "step": 4826
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42792969218551985,
+      "learning_rate": 2.5338441682666685e-05,
+      "loss": 0.6997,
+      "step": 4827
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4443905368517963,
+      "learning_rate": 2.5332859582509786e-05,
+      "loss": 0.6992,
+      "step": 4828
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.39307864213626725,
+      "learning_rate": 2.5327277035092403e-05,
+      "loss": 0.6524,
+      "step": 4829
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4210792692206135,
+      "learning_rate": 2.532169404088273e-05,
+      "loss": 0.6518,
+      "step": 4830
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4323969934070017,
+      "learning_rate": 2.5316110600348998e-05,
+      "loss": 0.6728,
+      "step": 4831
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42052449237052575,
+      "learning_rate": 2.5310526713959496e-05,
+      "loss": 0.7068,
+      "step": 4832
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4027174730945617,
+      "learning_rate": 2.5304942382182533e-05,
+      "loss": 0.7081,
+      "step": 4833
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3814283224787604,
+      "learning_rate": 2.529935760548645e-05,
+      "loss": 0.6669,
+      "step": 4834
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3908318234008154,
+      "learning_rate": 2.5293772384339653e-05,
+      "loss": 0.6822,
+      "step": 4835
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42038254304871403,
+      "learning_rate": 2.5288186719210557e-05,
+      "loss": 0.6915,
+      "step": 4836
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4011542350434515,
+      "learning_rate": 2.5282600610567628e-05,
+      "loss": 0.654,
+      "step": 4837
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4359386130065434,
+      "learning_rate": 2.5277014058879368e-05,
+      "loss": 0.6856,
+      "step": 4838
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4761621870936991,
+      "learning_rate": 2.5271427064614303e-05,
+      "loss": 0.6193,
+      "step": 4839
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.426206699723178,
+      "learning_rate": 2.5265839628241028e-05,
+      "loss": 0.7014,
+      "step": 4840
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.39156125881116466,
+      "learning_rate": 2.5260251750228136e-05,
+      "loss": 0.6861,
+      "step": 4841
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42151563263590835,
+      "learning_rate": 2.5254663431044285e-05,
+      "loss": 0.6826,
+      "step": 4842
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.37287940129756153,
+      "learning_rate": 2.5249074671158157e-05,
+      "loss": 0.6623,
+      "step": 4843
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42635664863392614,
+      "learning_rate": 2.5243485471038477e-05,
+      "loss": 0.6825,
+      "step": 4844
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.40570494805647705,
+      "learning_rate": 2.5237895831154e-05,
+      "loss": 0.6905,
+      "step": 4845
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.39598662703993354,
+      "learning_rate": 2.5232305751973525e-05,
+      "loss": 0.6841,
+      "step": 4846
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.46888433194120765,
+      "learning_rate": 2.5226715233965894e-05,
+      "loss": 0.6953,
+      "step": 4847
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.45438750418065454,
+      "learning_rate": 2.5221124277599955e-05,
+      "loss": 0.7184,
+      "step": 4848
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.34949387507595653,
+      "learning_rate": 2.5215532883344637e-05,
+      "loss": 0.664,
+      "step": 4849
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.39835735876710465,
+      "learning_rate": 2.5209941051668874e-05,
+      "loss": 0.6949,
+      "step": 4850
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.36428112099054727,
+      "learning_rate": 2.5204348783041644e-05,
+      "loss": 0.7017,
+      "step": 4851
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3584725525635114,
+      "learning_rate": 2.519875607793197e-05,
+      "loss": 0.6666,
+      "step": 4852
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.38694644829364727,
+      "learning_rate": 2.5193162936808903e-05,
+      "loss": 0.7074,
+      "step": 4853
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3829987163183963,
+      "learning_rate": 2.5187569360141534e-05,
+      "loss": 7.4153,
+      "step": 4854
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.44616189879651863,
+      "learning_rate": 2.5181975348398982e-05,
+      "loss": 0.7029,
+      "step": 4855
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.37453241319819236,
+      "learning_rate": 2.5176380902050418e-05,
+      "loss": 0.6958,
+      "step": 4856
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4171195852338107,
+      "learning_rate": 2.517078602156504e-05,
+      "loss": 0.6637,
+      "step": 4857
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.40796849742781305,
+      "learning_rate": 2.5165190707412085e-05,
+      "loss": 0.6971,
+      "step": 4858
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4062952034237757,
+      "learning_rate": 2.5159594960060813e-05,
+      "loss": 0.6402,
+      "step": 4859
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4481966852832561,
+      "learning_rate": 2.5153998779980557e-05,
+      "loss": 0.7075,
+      "step": 4860
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42410734743191497,
+      "learning_rate": 2.514840216764064e-05,
+      "loss": 0.7001,
+      "step": 4861
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.358105385021053,
+      "learning_rate": 2.5142805123510458e-05,
+      "loss": 0.6372,
+      "step": 4862
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3820690872363391,
+      "learning_rate": 2.5137207648059412e-05,
+      "loss": 0.6713,
+      "step": 4863
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3226802101479371,
+      "learning_rate": 2.5131609741756967e-05,
+      "loss": 0.6856,
+      "step": 4864
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4410001910959925,
+      "learning_rate": 2.5126011405072612e-05,
+      "loss": 0.7138,
+      "step": 4865
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3823078056455019,
+      "learning_rate": 2.5120412638475872e-05,
+      "loss": 0.6944,
+      "step": 4866
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.41227823520564794,
+      "learning_rate": 2.511481344243631e-05,
+      "loss": 0.7198,
+      "step": 4867
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4137702944903973,
+      "learning_rate": 2.5109213817423514e-05,
+      "loss": 0.7408,
+      "step": 4868
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3873902899544941,
+      "learning_rate": 2.510361376390713e-05,
+      "loss": 0.6742,
+      "step": 4869
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.37925312905924774,
+      "learning_rate": 2.5098013282356825e-05,
+      "loss": 0.6736,
+      "step": 4870
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4452801328619183,
+      "learning_rate": 2.50924123732423e-05,
+      "loss": 0.7317,
+      "step": 4871
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.36098339463637935,
+      "learning_rate": 2.5086811037033302e-05,
+      "loss": 0.6808,
+      "step": 4872
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.43419358350847204,
+      "learning_rate": 2.50812092741996e-05,
+      "loss": 0.6804,
+      "step": 4873
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4191569623336952,
+      "learning_rate": 2.5075607085211023e-05,
+      "loss": 0.6946,
+      "step": 4874
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.43179545594054536,
+      "learning_rate": 2.50700044705374e-05,
+      "loss": 0.7137,
+      "step": 4875
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.42887245673101154,
+      "learning_rate": 2.5064401430648633e-05,
+      "loss": 0.6851,
+      "step": 4876
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.41548683561575633,
+      "learning_rate": 2.5058797966014625e-05,
+      "loss": 0.629,
+      "step": 4877
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3878137786192555,
+      "learning_rate": 2.5053194077105344e-05,
+      "loss": 0.6784,
+      "step": 4878
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4138508281584996,
+      "learning_rate": 2.5047589764390786e-05,
+      "loss": 0.7071,
+      "step": 4879
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.42578753935340197,
+      "learning_rate": 2.504198502834096e-05,
+      "loss": 0.6764,
+      "step": 4880
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3686024710422419,
+      "learning_rate": 2.503637986942595e-05,
+      "loss": 0.6524,
+      "step": 4881
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3478090712584098,
+      "learning_rate": 2.503077428811583e-05,
+      "loss": 0.6684,
+      "step": 4882
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.39192872106131976,
+      "learning_rate": 2.5025168284880756e-05,
+      "loss": 0.6614,
+      "step": 4883
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.38244345024819204,
+      "learning_rate": 2.5019561860190883e-05,
+      "loss": 0.6703,
+      "step": 4884
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3644846388250423,
+      "learning_rate": 2.501395501451642e-05,
+      "loss": 0.6589,
+      "step": 4885
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.37015308883138337,
+      "learning_rate": 2.5008347748327603e-05,
+      "loss": 0.6728,
+      "step": 4886
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3437782587375705,
+      "learning_rate": 2.500274006209471e-05,
+      "loss": 0.7088,
+      "step": 4887
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40984836130180397,
+      "learning_rate": 2.499713195628805e-05,
+      "loss": 0.6656,
+      "step": 4888
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.34454713613257254,
+      "learning_rate": 2.4991523431377965e-05,
+      "loss": 0.6349,
+      "step": 4889
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4136764316759829,
+      "learning_rate": 2.4985914487834845e-05,
+      "loss": 0.6691,
+      "step": 4890
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.39223282816974664,
+      "learning_rate": 2.4980305126129093e-05,
+      "loss": 0.6882,
+      "step": 4891
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4373846474079105,
+      "learning_rate": 2.4974695346731165e-05,
+      "loss": 0.6827,
+      "step": 4892
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.39716573098658103,
+      "learning_rate": 2.4969085150111552e-05,
+      "loss": 0.6803,
+      "step": 4893
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3238696224603552,
+      "learning_rate": 2.4963474536740758e-05,
+      "loss": 0.6591,
+      "step": 4894
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3566128697818521,
+      "learning_rate": 2.495786350708935e-05,
+      "loss": 0.6733,
+      "step": 4895
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4116248286892564,
+      "learning_rate": 2.495225206162793e-05,
+      "loss": 0.673,
+      "step": 4896
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3927695975599469,
+      "learning_rate": 2.4946640200827103e-05,
+      "loss": 0.681,
+      "step": 4897
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40826147836120624,
+      "learning_rate": 2.4941027925157535e-05,
+      "loss": 0.7014,
+      "step": 4898
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.390128615666643,
+      "learning_rate": 2.4935415235089918e-05,
+      "loss": 0.6541,
+      "step": 4899
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.37259205777319326,
+      "learning_rate": 2.4929802131094996e-05,
+      "loss": 0.6803,
+      "step": 4900
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4000085931417083,
+      "learning_rate": 2.4924188613643513e-05,
+      "loss": 0.6512,
+      "step": 4901
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.41768706432938346,
+      "learning_rate": 2.4918574683206283e-05,
+      "loss": 0.7381,
+      "step": 4902
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40053280633621735,
+      "learning_rate": 2.4912960340254135e-05,
+      "loss": 0.6869,
+      "step": 4903
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.42229578947648166,
+      "learning_rate": 2.4907345585257938e-05,
+      "loss": 0.6801,
+      "step": 4904
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40347320678305976,
+      "learning_rate": 2.4901730418688594e-05,
+      "loss": 0.657,
+      "step": 4905
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4253283094001151,
+      "learning_rate": 2.4896114841017033e-05,
+      "loss": 0.703,
+      "step": 4906
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.35271897751996856,
+      "learning_rate": 2.489049885271424e-05,
+      "loss": 0.672,
+      "step": 4907
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.39977412815080743,
+      "learning_rate": 2.4884882454251207e-05,
+      "loss": 0.6583,
+      "step": 4908
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.44827815232497753,
+      "learning_rate": 2.4879265646098983e-05,
+      "loss": 0.7154,
+      "step": 4909
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3424582783649697,
+      "learning_rate": 2.487364842872865e-05,
+      "loss": 0.6571,
+      "step": 4910
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.44530243958985155,
+      "learning_rate": 2.4868030802611306e-05,
+      "loss": 0.7111,
+      "step": 4911
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40541723158677817,
+      "learning_rate": 2.486241276821809e-05,
+      "loss": 0.6765,
+      "step": 4912
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4758153754089267,
+      "learning_rate": 2.4856794326020196e-05,
+      "loss": 0.7276,
+      "step": 4913
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.42076543864144683,
+      "learning_rate": 2.485117547648883e-05,
+      "loss": 0.6814,
+      "step": 4914
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3781902749836343,
+      "learning_rate": 2.4845556220095227e-05,
+      "loss": 0.6827,
+      "step": 4915
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40383768991290114,
+      "learning_rate": 2.483993655731068e-05,
+      "loss": 0.6842,
+      "step": 4916
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3774162368186878,
+      "learning_rate": 2.4834316488606498e-05,
+      "loss": 0.6719,
+      "step": 4917
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.39038070491981075,
+      "learning_rate": 2.482869601445403e-05,
+      "loss": 0.6661,
+      "step": 4918
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4104950954192529,
+      "learning_rate": 2.4823075135324658e-05,
+      "loss": 0.6824,
+      "step": 4919
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3668584146468052,
+      "learning_rate": 2.48174538516898e-05,
+      "loss": 0.6664,
+      "step": 4920
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.35017315105113234,
+      "learning_rate": 2.4811832164020913e-05,
+      "loss": 0.6456,
+      "step": 4921
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3887150242189556,
+      "learning_rate": 2.4806210072789458e-05,
+      "loss": 0.7016,
+      "step": 4922
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.3991214495415867,
+      "learning_rate": 2.480058757846698e-05,
+      "loss": 7.1104,
+      "step": 4923
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4529079950594759,
+      "learning_rate": 2.479496468152502e-05,
+      "loss": 0.7076,
+      "step": 4924
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40745136863507503,
+      "learning_rate": 2.4789341382435158e-05,
+      "loss": 0.6772,
+      "step": 4925
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3896201320698493,
+      "learning_rate": 2.478371768166902e-05,
+      "loss": 0.6676,
+      "step": 4926
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40229420566304147,
+      "learning_rate": 2.4778093579698256e-05,
+      "loss": 0.6962,
+      "step": 4927
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3640794058304054,
+      "learning_rate": 2.4772469076994562e-05,
+      "loss": 0.6414,
+      "step": 4928
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3890467284287986,
+      "learning_rate": 2.4766844174029636e-05,
+      "loss": 0.6606,
+      "step": 4929
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3956275128686304,
+      "learning_rate": 2.4761218871275256e-05,
+      "loss": 0.6355,
+      "step": 4930
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.39541493156709273,
+      "learning_rate": 2.47555931692032e-05,
+      "loss": 0.691,
+      "step": 4931
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3919315646044405,
+      "learning_rate": 2.4749967068285284e-05,
+      "loss": 0.7046,
+      "step": 4932
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.46133161999653877,
+      "learning_rate": 2.4744340568993365e-05,
+      "loss": 0.7358,
+      "step": 4933
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3934867254868428,
+      "learning_rate": 2.4738713671799337e-05,
+      "loss": 0.6818,
+      "step": 4934
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.41066849922620174,
+      "learning_rate": 2.4733086377175112e-05,
+      "loss": 0.6835,
+      "step": 4935
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.45044880223611583,
+      "learning_rate": 2.472745868559265e-05,
+      "loss": 0.7234,
+      "step": 4936
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3541185513851363,
+      "learning_rate": 2.4721830597523943e-05,
+      "loss": 0.6618,
+      "step": 4937
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4143974403939413,
+      "learning_rate": 2.4716202113441003e-05,
+      "loss": 0.6991,
+      "step": 4938
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3381279723630861,
+      "learning_rate": 2.471057323381588e-05,
+      "loss": 0.6536,
+      "step": 4939
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.37517373056610887,
+      "learning_rate": 2.4704943959120677e-05,
+      "loss": 0.6951,
+      "step": 4940
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4043000649472544,
+      "learning_rate": 2.4699314289827506e-05,
+      "loss": 0.6801,
+      "step": 4941
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3874325569719771,
+      "learning_rate": 2.4693684226408515e-05,
+      "loss": 0.7159,
+      "step": 4942
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.36939546746413027,
+      "learning_rate": 2.46880537693359e-05,
+      "loss": 0.6825,
+      "step": 4943
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3978159506224017,
+      "learning_rate": 2.468242291908188e-05,
+      "loss": 0.6839,
+      "step": 4944
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3627701678961596,
+      "learning_rate": 2.4676791676118706e-05,
+      "loss": 0.6587,
+      "step": 4945
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.38283434177405645,
+      "learning_rate": 2.467116004091865e-05,
+      "loss": 0.6606,
+      "step": 4946
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.39196262198043597,
+      "learning_rate": 2.4665528013954054e-05,
+      "loss": 0.6862,
+      "step": 4947
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.45664740497020473,
+      "learning_rate": 2.465989559569726e-05,
+      "loss": 0.6827,
+      "step": 4948
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40375844185477683,
+      "learning_rate": 2.465426278662064e-05,
+      "loss": 0.6797,
+      "step": 4949
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4085208237514952,
+      "learning_rate": 2.464862958719663e-05,
+      "loss": 0.6813,
+      "step": 4950
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4035800966611084,
+      "learning_rate": 2.464299599789767e-05,
+      "loss": 0.6867,
+      "step": 4951
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.41059405616131445,
+      "learning_rate": 2.4637362019196235e-05,
+      "loss": 0.6589,
+      "step": 4952
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.31976988278964635,
+      "learning_rate": 2.4631727651564852e-05,
+      "loss": 0.6498,
+      "step": 4953
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3805794517145168,
+      "learning_rate": 2.462609289547607e-05,
+      "loss": 0.7034,
+      "step": 4954
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3913318486815797,
+      "learning_rate": 2.462045775140246e-05,
+      "loss": 0.6814,
+      "step": 4955
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3913412781639698,
+      "learning_rate": 2.461482221981663e-05,
+      "loss": 0.6888,
+      "step": 4956
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.35043728376915667,
+      "learning_rate": 2.4609186301191245e-05,
+      "loss": 0.6781,
+      "step": 4957
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3610760537853345,
+      "learning_rate": 2.460354999599897e-05,
+      "loss": 0.6685,
+      "step": 4958
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3844308708749437,
+      "learning_rate": 2.4597913304712516e-05,
+      "loss": 0.6903,
+      "step": 4959
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4034808008010445,
+      "learning_rate": 2.4592276227804618e-05,
+      "loss": 0.6992,
+      "step": 4960
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40759031546930924,
+      "learning_rate": 2.458663876574807e-05,
+      "loss": 0.6779,
+      "step": 4961
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4232182504056296,
+      "learning_rate": 2.4581000919015664e-05,
+      "loss": 0.6761,
+      "step": 4962
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4004579087247222,
+      "learning_rate": 2.4575362688080236e-05,
+      "loss": 0.6996,
+      "step": 4963
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.38315209012383467,
+      "learning_rate": 2.456972407341468e-05,
+      "loss": 0.6712,
+      "step": 4964
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.38787643916432685,
+      "learning_rate": 2.4564085075491876e-05,
+      "loss": 0.6519,
+      "step": 4965
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.41853437673054883,
+      "learning_rate": 2.455844569478477e-05,
+      "loss": 0.6723,
+      "step": 4966
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.44486105605908066,
+      "learning_rate": 2.4552805931766322e-05,
+      "loss": 0.6683,
+      "step": 4967
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40949982442919375,
+      "learning_rate": 2.4547165786909548e-05,
+      "loss": 0.6889,
+      "step": 4968
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.46774275555782996,
+      "learning_rate": 2.4541525260687468e-05,
+      "loss": 0.7165,
+      "step": 4969
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4358708269394856,
+      "learning_rate": 2.4535884353573148e-05,
+      "loss": 0.7186,
+      "step": 4970
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3620623267328505,
+      "learning_rate": 2.4530243066039688e-05,
+      "loss": 0.6476,
+      "step": 4971
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.34855593872707447,
+      "learning_rate": 2.452460139856021e-05,
+      "loss": 0.6445,
+      "step": 4972
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.41825902089929906,
+      "learning_rate": 2.451895935160788e-05,
+      "loss": 0.7065,
+      "step": 4973
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3989659315050734,
+      "learning_rate": 2.4513316925655882e-05,
+      "loss": 0.6728,
+      "step": 4974
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3841054105750787,
+      "learning_rate": 2.450767412117745e-05,
+      "loss": 0.7059,
+      "step": 4975
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3774590952181273,
+      "learning_rate": 2.450203093864583e-05,
+      "loss": 0.6882,
+      "step": 4976
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.40879004289353493,
+      "learning_rate": 2.449638737853431e-05,
+      "loss": 0.6935,
+      "step": 4977
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.41877004658202605,
+      "learning_rate": 2.4490743441316217e-05,
+      "loss": 0.6822,
+      "step": 4978
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.40380700394944463,
+      "learning_rate": 2.448509912746489e-05,
+      "loss": 0.6552,
+      "step": 4979
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.3798350348380426,
+      "learning_rate": 2.4479454437453716e-05,
+      "loss": 7.4585,
+      "step": 4980
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.45962005709446613,
+      "learning_rate": 2.447380937175611e-05,
+      "loss": 0.6832,
+      "step": 4981
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4194135891272633,
+      "learning_rate": 2.446816393084552e-05,
+      "loss": 0.644,
+      "step": 4982
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.40969728039960723,
+      "learning_rate": 2.4462518115195408e-05,
+      "loss": 0.6981,
+      "step": 4983
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4238575786683269,
+      "learning_rate": 2.4456871925279293e-05,
+      "loss": 0.7189,
+      "step": 4984
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.43213028997521147,
+      "learning_rate": 2.4451225361570717e-05,
+      "loss": 0.6862,
+      "step": 4985
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3626428646855617,
+      "learning_rate": 2.4445578424543245e-05,
+      "loss": 0.681,
+      "step": 4986
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.41894224734014474,
+      "learning_rate": 2.443993111467047e-05,
+      "loss": 0.6794,
+      "step": 4987
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4352487258117622,
+      "learning_rate": 2.4434283432426048e-05,
+      "loss": 0.6952,
+      "step": 4988
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.5033476661124756,
+      "learning_rate": 2.4428635378283624e-05,
+      "loss": 7.3574,
+      "step": 4989
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4156458143720424,
+      "learning_rate": 2.4422986952716897e-05,
+      "loss": 0.708,
+      "step": 4990
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4020346649467947,
+      "learning_rate": 2.4417338156199596e-05,
+      "loss": 0.717,
+      "step": 4991
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3957086541324942,
+      "learning_rate": 2.4411688989205482e-05,
+      "loss": 0.6717,
+      "step": 4992
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4115654757746123,
+      "learning_rate": 2.4406039452208336e-05,
+      "loss": 0.6874,
+      "step": 4993
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3339135006429137,
+      "learning_rate": 2.4400389545681983e-05,
+      "loss": 0.649,
+      "step": 4994
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.42751425108206476,
+      "learning_rate": 2.439473927010028e-05,
+      "loss": 0.6992,
+      "step": 4995
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4034190500987647,
+      "learning_rate": 2.4389088625937098e-05,
+      "loss": 0.6745,
+      "step": 4996
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4120218599761797,
+      "learning_rate": 2.4383437613666353e-05,
+      "loss": 0.6683,
+      "step": 4997
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4385452667364385,
+      "learning_rate": 2.4377786233761985e-05,
+      "loss": 0.6381,
+      "step": 4998
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4283509382376838,
+      "learning_rate": 2.4372134486697982e-05,
+      "loss": 0.7304,
+      "step": 4999
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.389310579733277,
+      "learning_rate": 2.436648237294833e-05,
+      "loss": 0.6807,
+      "step": 5000
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4633034714420927,
+      "learning_rate": 2.4360829892987078e-05,
+      "loss": 0.7075,
+      "step": 5001
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.361124900369346,
+      "learning_rate": 2.435517704728829e-05,
+      "loss": 0.6711,
+      "step": 5002
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3347707474881972,
+      "learning_rate": 2.434952383632607e-05,
+      "loss": 0.6544,
+      "step": 5003
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.40064657941194903,
+      "learning_rate": 2.4343870260574528e-05,
+      "loss": 0.6604,
+      "step": 5004
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.41336828560485184,
+      "learning_rate": 2.433821632050784e-05,
+      "loss": 0.6585,
+      "step": 5005
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4089536279510622,
+      "learning_rate": 2.4332562016600186e-05,
+      "loss": 0.7089,
+      "step": 5006
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3461458028403347,
+      "learning_rate": 2.4326907349325784e-05,
+      "loss": 0.6772,
+      "step": 5007
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.37737623591738545,
+      "learning_rate": 2.4321252319158893e-05,
+      "loss": 0.6536,
+      "step": 5008
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4324697809719155,
+      "learning_rate": 2.4315596926573787e-05,
+      "loss": 0.6751,
+      "step": 5009
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3602335150245621,
+      "learning_rate": 2.430994117204478e-05,
+      "loss": 0.684,
+      "step": 5010
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.37465116633820883,
+      "learning_rate": 2.4304285056046215e-05,
+      "loss": 0.656,
+      "step": 5011
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3776019584555366,
+      "learning_rate": 2.4298628579052456e-05,
+      "loss": 0.6715,
+      "step": 5012
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.38632473409236734,
+      "learning_rate": 2.4292971741537907e-05,
+      "loss": 0.6424,
+      "step": 5013
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.44665128899430073,
+      "learning_rate": 2.428731454397701e-05,
+      "loss": 0.7038,
+      "step": 5014
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.37861740403574334,
+      "learning_rate": 2.4281656986844217e-05,
+      "loss": 0.663,
+      "step": 5015
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3649966637086012,
+      "learning_rate": 2.427599907061402e-05,
+      "loss": 0.6576,
+      "step": 5016
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4216699590884934,
+      "learning_rate": 2.427034079576095e-05,
+      "loss": 0.6946,
+      "step": 5017
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.4608932004596604,
+      "learning_rate": 2.4264682162759555e-05,
+      "loss": 7.1439,
+      "step": 5018
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4312875695980344,
+      "learning_rate": 2.4259023172084412e-05,
+      "loss": 0.7077,
+      "step": 5019
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4290297081410972,
+      "learning_rate": 2.425336382421014e-05,
+      "loss": 7.2732,
+      "step": 5020
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4158287258311452,
+      "learning_rate": 2.4247704119611385e-05,
+      "loss": 0.6795,
+      "step": 5021
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.36346387769809824,
+      "learning_rate": 2.4242044058762812e-05,
+      "loss": 0.6528,
+      "step": 5022
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3802329105712019,
+      "learning_rate": 2.423638364213913e-05,
+      "loss": 0.6477,
+      "step": 5023
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3328498922798293,
+      "learning_rate": 2.4230722870215062e-05,
+      "loss": 0.6581,
+      "step": 5024
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3816685442593181,
+      "learning_rate": 2.422506174346538e-05,
+      "loss": 0.6714,
+      "step": 5025
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.43053758409256343,
+      "learning_rate": 2.4219400262364875e-05,
+      "loss": 0.6982,
+      "step": 5026
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3915660172711186,
+      "learning_rate": 2.4213738427388362e-05,
+      "loss": 0.6777,
+      "step": 5027
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.40565875739252777,
+      "learning_rate": 2.42080762390107e-05,
+      "loss": 0.6479,
+      "step": 5028
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4084625211731897,
+      "learning_rate": 2.4202413697706764e-05,
+      "loss": 0.6704,
+      "step": 5029
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3819476598541026,
+      "learning_rate": 2.419675080395146e-05,
+      "loss": 0.6756,
+      "step": 5030
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.44005773005128684,
+      "learning_rate": 2.4191087558219744e-05,
+      "loss": 0.6722,
+      "step": 5031
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3701718385336954,
+      "learning_rate": 2.418542396098657e-05,
+      "loss": 0.6506,
+      "step": 5032
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.370620499846445,
+      "learning_rate": 2.417976001272695e-05,
+      "loss": 0.6437,
+      "step": 5033
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.41333887983730566,
+      "learning_rate": 2.4174095713915898e-05,
+      "loss": 0.7116,
+      "step": 5034
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5014616071150109,
+      "learning_rate": 2.4168431065028487e-05,
+      "loss": 0.7253,
+      "step": 5035
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.43919526874003667,
+      "learning_rate": 2.416276606653979e-05,
+      "loss": 0.6869,
+      "step": 5036
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3426886501846818,
+      "learning_rate": 2.415710071892493e-05,
+      "loss": 0.6448,
+      "step": 5037
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4220938856035365,
+      "learning_rate": 2.415143502265906e-05,
+      "loss": 0.7001,
+      "step": 5038
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.40704480545201965,
+      "learning_rate": 2.414576897821734e-05,
+      "loss": 0.6565,
+      "step": 5039
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3920735375366674,
+      "learning_rate": 2.4140102586074986e-05,
+      "loss": 0.6542,
+      "step": 5040
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3907819121998663,
+      "learning_rate": 2.4134435846707232e-05,
+      "loss": 0.6976,
+      "step": 5041
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.35979326826121233,
+      "learning_rate": 2.4128768760589333e-05,
+      "loss": 0.6928,
+      "step": 5042
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.43145508077070976,
+      "learning_rate": 2.412310132819659e-05,
+      "loss": 0.6585,
+      "step": 5043
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.37946092976003937,
+      "learning_rate": 2.4117433550004304e-05,
+      "loss": 0.6577,
+      "step": 5044
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3704465963471798,
+      "learning_rate": 2.4111765426487854e-05,
+      "loss": 0.6536,
+      "step": 5045
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3514812527005296,
+      "learning_rate": 2.4106096958122594e-05,
+      "loss": 0.66,
+      "step": 5046
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.39843245344046685,
+      "learning_rate": 2.410042814538394e-05,
+      "loss": 0.6759,
+      "step": 5047
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.36310677121346546,
+      "learning_rate": 2.409475898874734e-05,
+      "loss": 0.7312,
+      "step": 5048
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.375461382085811,
+      "learning_rate": 2.4089089488688245e-05,
+      "loss": 0.6282,
+      "step": 5049
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.33957263712195934,
+      "learning_rate": 2.408341964568215e-05,
+      "loss": 0.6519,
+      "step": 5050
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.373497336245297,
+      "learning_rate": 2.4077749460204576e-05,
+      "loss": 0.6482,
+      "step": 5051
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4400949129867158,
+      "learning_rate": 2.4072078932731088e-05,
+      "loss": 0.7199,
+      "step": 5052
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3684955689086381,
+      "learning_rate": 2.4066408063737253e-05,
+      "loss": 0.6943,
+      "step": 5053
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3546426395786624,
+      "learning_rate": 2.4060736853698683e-05,
+      "loss": 0.6864,
+      "step": 5054
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3382847755657906,
+      "learning_rate": 2.4055065303091025e-05,
+      "loss": 0.6388,
+      "step": 5055
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3932610778198451,
+      "learning_rate": 2.4049393412389935e-05,
+      "loss": 0.6675,
+      "step": 5056
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.37745762469726396,
+      "learning_rate": 2.40437211820711e-05,
+      "loss": 0.6639,
+      "step": 5057
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4292128419548519,
+      "learning_rate": 2.4038048612610264e-05,
+      "loss": 0.7003,
+      "step": 5058
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3648003631476323,
+      "learning_rate": 2.4032375704483166e-05,
+      "loss": 0.6269,
+      "step": 5059
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.46534503727875615,
+      "learning_rate": 2.4026702458165587e-05,
+      "loss": 0.7259,
+      "step": 5060
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.41853719546782653,
+      "learning_rate": 2.402102887413333e-05,
+      "loss": 0.7299,
+      "step": 5061
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3742378366681166,
+      "learning_rate": 2.4015354952862248e-05,
+      "loss": 0.6744,
+      "step": 5062
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3663211570931236,
+      "learning_rate": 2.4009680694828184e-05,
+      "loss": 0.6863,
+      "step": 5063
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.37405034694665124,
+      "learning_rate": 2.4004006100507048e-05,
+      "loss": 0.662,
+      "step": 5064
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3712591794955222,
+      "learning_rate": 2.399833117037476e-05,
+      "loss": 0.6852,
+      "step": 5065
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4007745206130673,
+      "learning_rate": 2.399265590490726e-05,
+      "loss": 0.6648,
+      "step": 5066
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.33784712539191764,
+      "learning_rate": 2.3986980304580533e-05,
+      "loss": 0.6642,
+      "step": 5067
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4026494321168532,
+      "learning_rate": 2.3981304369870578e-05,
+      "loss": 0.662,
+      "step": 5068
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.41609777716272994,
+      "learning_rate": 2.3975628101253446e-05,
+      "loss": 0.7158,
+      "step": 5069
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.40309621844421606,
+      "learning_rate": 2.3969951499205175e-05,
+      "loss": 0.6573,
+      "step": 5070
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.40533025836297487,
+      "learning_rate": 2.396427456420187e-05,
+      "loss": 0.6643,
+      "step": 5071
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.386775677473754,
+      "learning_rate": 2.3958597296719646e-05,
+      "loss": 0.6769,
+      "step": 5072
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4446157905086338,
+      "learning_rate": 2.3952919697234646e-05,
+      "loss": 0.7342,
+      "step": 5073
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.40964787638027017,
+      "learning_rate": 2.394724176622304e-05,
+      "loss": 0.6652,
+      "step": 5074
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4131957744510276,
+      "learning_rate": 2.3941563504161044e-05,
+      "loss": 0.6531,
+      "step": 5075
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.36450169879681144,
+      "learning_rate": 2.3935884911524876e-05,
+      "loss": 0.6621,
+      "step": 5076
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.44700248895979217,
+      "learning_rate": 2.393020598879079e-05,
+      "loss": 0.7081,
+      "step": 5077
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4002266860658442,
+      "learning_rate": 2.3924526736435068e-05,
+      "loss": 0.6519,
+      "step": 5078
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4471303955698012,
+      "learning_rate": 2.391884715493404e-05,
+      "loss": 0.6709,
+      "step": 5079
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.42184022165594687,
+      "learning_rate": 2.3913167244764034e-05,
+      "loss": 0.712,
+      "step": 5080
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.37208429498811874,
+      "learning_rate": 2.3907487006401403e-05,
+      "loss": 0.6492,
+      "step": 5081
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3784202482473267,
+      "learning_rate": 2.390180644032257e-05,
+      "loss": 0.6759,
+      "step": 5082
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3755930514137703,
+      "learning_rate": 2.3896125547003938e-05,
+      "loss": 0.6385,
+      "step": 5083
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.38854609142431107,
+      "learning_rate": 2.389044432692196e-05,
+      "loss": 0.6813,
+      "step": 5084
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.42104546325078823,
+      "learning_rate": 2.3884762780553114e-05,
+      "loss": 0.6669,
+      "step": 5085
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.37156098192272735,
+      "learning_rate": 2.387908090837391e-05,
+      "loss": 0.6687,
+      "step": 5086
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3785510845066396,
+      "learning_rate": 2.3873398710860873e-05,
+      "loss": 0.6453,
+      "step": 5087
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.38409500601477203,
+      "learning_rate": 2.3867716188490563e-05,
+      "loss": 0.6963,
+      "step": 5088
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.44711469062124887,
+      "learning_rate": 2.386203334173957e-05,
+      "loss": 0.7075,
+      "step": 5089
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5350685781515022,
+      "learning_rate": 2.38563501710845e-05,
+      "loss": 0.6751,
+      "step": 5090
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.39704717008142554,
+      "learning_rate": 2.3850666677002e-05,
+      "loss": 0.6493,
+      "step": 5091
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.373884089636565,
+      "learning_rate": 2.3844982859968736e-05,
+      "loss": 0.6421,
+      "step": 5092
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.37101417069844406,
+      "learning_rate": 2.3839298720461405e-05,
+      "loss": 0.63,
+      "step": 5093
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.39854637509378127,
+      "learning_rate": 2.3833614258956725e-05,
+      "loss": 0.673,
+      "step": 5094
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3353984398471668,
+      "learning_rate": 2.3827929475931447e-05,
+      "loss": 0.6471,
+      "step": 5095
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.43189726996119293,
+      "learning_rate": 2.3822244371862352e-05,
+      "loss": 0.6869,
+      "step": 5096
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.38192161740572717,
+      "learning_rate": 2.3816558947226232e-05,
+      "loss": 0.6864,
+      "step": 5097
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35617020267898375,
+      "learning_rate": 2.381087320249992e-05,
+      "loss": 0.6563,
+      "step": 5098
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.38625483523854537,
+      "learning_rate": 2.380518713816028e-05,
+      "loss": 0.6747,
+      "step": 5099
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.38352278960431285,
+      "learning_rate": 2.3799500754684196e-05,
+      "loss": 0.6528,
+      "step": 5100
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4203744677430944,
+      "learning_rate": 2.379381405254856e-05,
+      "loss": 0.7149,
+      "step": 5101
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.37684163208197163,
+      "learning_rate": 2.378812703223033e-05,
+      "loss": 0.7098,
+      "step": 5102
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.36677605936695884,
+      "learning_rate": 2.3782439694206456e-05,
+      "loss": 0.6458,
+      "step": 5103
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.37420140456051904,
+      "learning_rate": 2.3776752038953932e-05,
+      "loss": 0.6439,
+      "step": 5104
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4196284622189283,
+      "learning_rate": 2.3771064066949777e-05,
+      "loss": 0.7155,
+      "step": 5105
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.38494524149896575,
+      "learning_rate": 2.3765375778671035e-05,
+      "loss": 0.6977,
+      "step": 5106
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3745764158031804,
+      "learning_rate": 2.3759687174594778e-05,
+      "loss": 0.6637,
+      "step": 5107
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3418052179451073,
+      "learning_rate": 2.3753998255198085e-05,
+      "loss": 0.6432,
+      "step": 5108
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.36875198349409477,
+      "learning_rate": 2.3748309020958105e-05,
+      "loss": 0.6157,
+      "step": 5109
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3363792753007904,
+      "learning_rate": 2.374261947235197e-05,
+      "loss": 0.5992,
+      "step": 5110
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.40357508033091044,
+      "learning_rate": 2.373692960985686e-05,
+      "loss": 0.6653,
+      "step": 5111
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.32969860359008973,
+      "learning_rate": 2.3731239433949972e-05,
+      "loss": 0.6871,
+      "step": 5112
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3618915819074089,
+      "learning_rate": 2.3725548945108544e-05,
+      "loss": 0.6841,
+      "step": 5113
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.362120737044373,
+      "learning_rate": 2.3719858143809824e-05,
+      "loss": 0.6817,
+      "step": 5114
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3868529375429973,
+      "learning_rate": 2.3714167030531093e-05,
+      "loss": 0.6908,
+      "step": 5115
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.39218299211578134,
+      "learning_rate": 2.3708475605749663e-05,
+      "loss": 0.6928,
+      "step": 5116
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.38658453718176067,
+      "learning_rate": 2.370278386994286e-05,
+      "loss": 0.7271,
+      "step": 5117
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3584265021131719,
+      "learning_rate": 2.369709182358804e-05,
+      "loss": 0.664,
+      "step": 5118
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35095593992335194,
+      "learning_rate": 2.36913994671626e-05,
+      "loss": 0.6563,
+      "step": 5119
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3670674371344031,
+      "learning_rate": 2.3685706801143944e-05,
+      "loss": 0.6874,
+      "step": 5120
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3756885161978683,
+      "learning_rate": 2.3680013826009507e-05,
+      "loss": 0.6689,
+      "step": 5121
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.34764053820330126,
+      "learning_rate": 2.3674320542236757e-05,
+      "loss": 0.6689,
+      "step": 5122
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.40636907385967874,
+      "learning_rate": 2.366862695030318e-05,
+      "loss": 0.6539,
+      "step": 5123
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.37418295074452224,
+      "learning_rate": 2.3662933050686293e-05,
+      "loss": 0.6649,
+      "step": 5124
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3441844632244847,
+      "learning_rate": 2.3657238843863636e-05,
+      "loss": 0.6827,
+      "step": 5125
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3868291018656716,
+      "learning_rate": 2.365154433031277e-05,
+      "loss": 0.656,
+      "step": 5126
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35219709216046513,
+      "learning_rate": 2.3645849510511295e-05,
+      "loss": 0.6495,
+      "step": 5127
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3850313910913583,
+      "learning_rate": 2.364015438493682e-05,
+      "loss": 0.6809,
+      "step": 5128
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.36351393834375684,
+      "learning_rate": 2.3634458954066992e-05,
+      "loss": 0.6388,
+      "step": 5129
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3638587424987737,
+      "learning_rate": 2.3628763218379485e-05,
+      "loss": 0.6408,
+      "step": 5130
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 3.724458496261448,
+      "learning_rate": 2.362306717835199e-05,
+      "loss": 7.3055,
+      "step": 5131
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4380445756495802,
+      "learning_rate": 2.3617370834462223e-05,
+      "loss": 0.6902,
+      "step": 5132
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4259823159489984,
+      "learning_rate": 2.361167418718793e-05,
+      "loss": 0.7047,
+      "step": 5133
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.41461039546220674,
+      "learning_rate": 2.3605977237006898e-05,
+      "loss": 0.6737,
+      "step": 5134
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.40388097033215936,
+      "learning_rate": 2.3600279984396894e-05,
+      "loss": 0.6532,
+      "step": 5135
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.38520195473929175,
+      "learning_rate": 2.3594582429835763e-05,
+      "loss": 0.6506,
+      "step": 5136
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9870388537849404,
+      "learning_rate": 2.358888457380135e-05,
+      "loss": 7.2605,
+      "step": 5137
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.43162002212896966,
+      "learning_rate": 2.3583186416771513e-05,
+      "loss": 0.6516,
+      "step": 5138
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4051043517895467,
+      "learning_rate": 2.3577487959224162e-05,
+      "loss": 0.686,
+      "step": 5139
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.366319018474789,
+      "learning_rate": 2.3571789201637216e-05,
+      "loss": 0.6622,
+      "step": 5140
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.396031123389406,
+      "learning_rate": 2.356609014448862e-05,
+      "loss": 0.6214,
+      "step": 5141
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3529409928983969,
+      "learning_rate": 2.356039078825636e-05,
+      "loss": 0.6386,
+      "step": 5142
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35933109453728573,
+      "learning_rate": 2.355469113341841e-05,
+      "loss": 0.647,
+      "step": 5143
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3436496194920967,
+      "learning_rate": 2.3548991180452824e-05,
+      "loss": 0.6676,
+      "step": 5144
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3980163024074107,
+      "learning_rate": 2.3543290929837626e-05,
+      "loss": 0.6676,
+      "step": 5145
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3529795308014352,
+      "learning_rate": 2.3537590382050893e-05,
+      "loss": 0.6804,
+      "step": 5146
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3863953982853381,
+      "learning_rate": 2.3531889537570735e-05,
+      "loss": 0.6724,
+      "step": 5147
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.40638171686252494,
+      "learning_rate": 2.352618839687526e-05,
+      "loss": 0.6529,
+      "step": 5148
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4395326901345472,
+      "learning_rate": 2.3520486960442627e-05,
+      "loss": 0.6999,
+      "step": 5149
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.39811608254229486,
+      "learning_rate": 2.3514785228751e-05,
+      "loss": 0.6863,
+      "step": 5150
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3781305258718326,
+      "learning_rate": 2.3509083202278588e-05,
+      "loss": 0.6521,
+      "step": 5151
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4052127003881731,
+      "learning_rate": 2.3503380881503598e-05,
+      "loss": 0.6697,
+      "step": 5152
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3500819482716826,
+      "learning_rate": 2.3497678266904287e-05,
+      "loss": 0.6747,
+      "step": 5153
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3588286814015664,
+      "learning_rate": 2.3491975358958925e-05,
+      "loss": 0.6498,
+      "step": 5154
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35432211662926905,
+      "learning_rate": 2.3486272158145807e-05,
+      "loss": 0.6607,
+      "step": 5155
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4147001590121596,
+      "learning_rate": 2.3480568664943255e-05,
+      "loss": 0.6994,
+      "step": 5156
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3879985196009324,
+      "learning_rate": 2.3474864879829608e-05,
+      "loss": 0.6989,
+      "step": 5157
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.38790791476570585,
+      "learning_rate": 2.346916080328324e-05,
+      "loss": 0.6975,
+      "step": 5158
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4480687713232306,
+      "learning_rate": 2.3463456435782554e-05,
+      "loss": 0.6543,
+      "step": 5159
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4389816575296455,
+      "learning_rate": 2.345775177780595e-05,
+      "loss": 0.7124,
+      "step": 5160
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.41518760405009897,
+      "learning_rate": 2.3452046829831884e-05,
+      "loss": 0.6944,
+      "step": 5161
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4371464695019962,
+      "learning_rate": 2.344634159233882e-05,
+      "loss": 0.6328,
+      "step": 5162
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3549976135284465,
+      "learning_rate": 2.344063606580525e-05,
+      "loss": 0.6785,
+      "step": 5163
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 4.364043647962927,
+      "learning_rate": 2.343493025070968e-05,
+      "loss": 7.3428,
+      "step": 5164
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.37062542579811125,
+      "learning_rate": 2.3429224147530664e-05,
+      "loss": 0.6365,
+      "step": 5165
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3608877457316091,
+      "learning_rate": 2.3423517756746764e-05,
+      "loss": 0.6918,
+      "step": 5166
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3700866093956325,
+      "learning_rate": 2.341781107883656e-05,
+      "loss": 0.6778,
+      "step": 5167
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.47320867680729034,
+      "learning_rate": 2.341210411427867e-05,
+      "loss": 0.691,
+      "step": 5168
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3977766966515981,
+      "learning_rate": 2.340639686355173e-05,
+      "loss": 0.674,
+      "step": 5169
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3947087518476362,
+      "learning_rate": 2.3400689327134395e-05,
+      "loss": 0.6969,
+      "step": 5170
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.40108472335357975,
+      "learning_rate": 2.3394981505505357e-05,
+      "loss": 0.7231,
+      "step": 5171
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.36851788370121535,
+      "learning_rate": 2.3389273399143317e-05,
+      "loss": 0.6944,
+      "step": 5172
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.38812566197151505,
+      "learning_rate": 2.3383565008527008e-05,
+      "loss": 0.6965,
+      "step": 5173
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3408880424362588,
+      "learning_rate": 2.3377856334135194e-05,
+      "loss": 0.6382,
+      "step": 5174
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.37148328206591535,
+      "learning_rate": 2.337214737644664e-05,
+      "loss": 0.6934,
+      "step": 5175
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35714486294606246,
+      "learning_rate": 2.336643813594017e-05,
+      "loss": 0.6688,
+      "step": 5176
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4086715611251787,
+      "learning_rate": 2.3360728613094595e-05,
+      "loss": 0.66,
+      "step": 5177
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35279638180246514,
+      "learning_rate": 2.3355018808388765e-05,
+      "loss": 0.6606,
+      "step": 5178
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.46996469980817046,
+      "learning_rate": 2.334930872230156e-05,
+      "loss": 0.7262,
+      "step": 5179
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4147832608431751,
+      "learning_rate": 2.3343598355311886e-05,
+      "loss": 0.6873,
+      "step": 5180
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.40826422020242675,
+      "learning_rate": 2.3337887707898648e-05,
+      "loss": 0.6718,
+      "step": 5181
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3844989321147029,
+      "learning_rate": 2.33321767805408e-05,
+      "loss": 0.6844,
+      "step": 5182
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.36161795575738676,
+      "learning_rate": 2.332646557371731e-05,
+      "loss": 0.6308,
+      "step": 5183
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4055440524525312,
+      "learning_rate": 2.3320754087907176e-05,
+      "loss": 0.6821,
+      "step": 5184
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.42693704510803143,
+      "learning_rate": 2.33150423235894e-05,
+      "loss": 0.724,
+      "step": 5185
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3965580441405362,
+      "learning_rate": 2.330933028124303e-05,
+      "loss": 0.6652,
+      "step": 5186
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.37468772840940906,
+      "learning_rate": 2.3303617961347126e-05,
+      "loss": 0.6839,
+      "step": 5187
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4290418701434604,
+      "learning_rate": 2.3297905364380773e-05,
+      "loss": 0.7086,
+      "step": 5188
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7080814523020627,
+      "learning_rate": 2.3292192490823075e-05,
+      "loss": 0.6735,
+      "step": 5189
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.38440818644500285,
+      "learning_rate": 2.328647934115318e-05,
+      "loss": 0.656,
+      "step": 5190
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3491432880794704,
+      "learning_rate": 2.3280765915850224e-05,
+      "loss": 0.6813,
+      "step": 5191
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.40049266784183,
+      "learning_rate": 2.327505221539339e-05,
+      "loss": 0.6538,
+      "step": 5192
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4183181140312501,
+      "learning_rate": 2.3269338240261886e-05,
+      "loss": 0.6786,
+      "step": 5193
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3884451882313566,
+      "learning_rate": 2.3263623990934936e-05,
+      "loss": 0.6759,
+      "step": 5194
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.48589708970812406,
+      "learning_rate": 2.325790946789178e-05,
+      "loss": 0.6975,
+      "step": 5195
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3807492973877077,
+      "learning_rate": 2.3252194671611686e-05,
+      "loss": 0.6254,
+      "step": 5196
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3559323987069765,
+      "learning_rate": 2.324647960257396e-05,
+      "loss": 0.6261,
+      "step": 5197
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3746593054756361,
+      "learning_rate": 2.3240764261257908e-05,
+      "loss": 0.6895,
+      "step": 5198
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3741938019943328,
+      "learning_rate": 2.3235048648142867e-05,
+      "loss": 0.6681,
+      "step": 5199
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35649543782691007,
+      "learning_rate": 2.3229332763708214e-05,
+      "loss": 0.647,
+      "step": 5200
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3171762307974289,
+      "learning_rate": 2.3223616608433317e-05,
+      "loss": 0.65,
+      "step": 5201
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.31492645452934187,
+      "learning_rate": 2.321790018279758e-05,
+      "loss": 0.64,
+      "step": 5202
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3337240714560984,
+      "learning_rate": 2.3212183487280448e-05,
+      "loss": 0.6369,
+      "step": 5203
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.415526990291042,
+      "learning_rate": 2.320646652236137e-05,
+      "loss": 0.7099,
+      "step": 5204
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39705273354831677,
+      "learning_rate": 2.3200749288519812e-05,
+      "loss": 0.649,
+      "step": 5205
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4169279384227062,
+      "learning_rate": 2.3195031786235272e-05,
+      "loss": 0.6594,
+      "step": 5206
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4375477611954182,
+      "learning_rate": 2.3189314015987286e-05,
+      "loss": 0.6523,
+      "step": 5207
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.43981101490437413,
+      "learning_rate": 2.3183595978255382e-05,
+      "loss": 0.6754,
+      "step": 5208
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.44569692664559646,
+      "learning_rate": 2.317787767351912e-05,
+      "loss": 0.7106,
+      "step": 5209
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4579612470622031,
+      "learning_rate": 2.3172159102258106e-05,
+      "loss": 0.6852,
+      "step": 5210
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39699571660885946,
+      "learning_rate": 2.3166440264951935e-05,
+      "loss": 0.6665,
+      "step": 5211
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3791041737283034,
+      "learning_rate": 2.316072116208024e-05,
+      "loss": 0.6502,
+      "step": 5212
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3239416127947213,
+      "learning_rate": 2.3155001794122676e-05,
+      "loss": 0.6435,
+      "step": 5213
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.37386854020395455,
+      "learning_rate": 2.314928216155893e-05,
+      "loss": 0.6339,
+      "step": 5214
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.42325918773760873,
+      "learning_rate": 2.3143562264868686e-05,
+      "loss": 0.6616,
+      "step": 5215
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.41763831426357345,
+      "learning_rate": 2.313784210453167e-05,
+      "loss": 0.6798,
+      "step": 5216
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4080355030860911,
+      "learning_rate": 2.3132121681027635e-05,
+      "loss": 0.6657,
+      "step": 5217
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3634132517285384,
+      "learning_rate": 2.3126400994836337e-05,
+      "loss": 0.6546,
+      "step": 5218
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4499770615266012,
+      "learning_rate": 2.3120680046437554e-05,
+      "loss": 0.6921,
+      "step": 5219
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.37349613202646786,
+      "learning_rate": 2.311495883631111e-05,
+      "loss": 0.6872,
+      "step": 5220
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39940203802410473,
+      "learning_rate": 2.3109237364936834e-05,
+      "loss": 0.6437,
+      "step": 5221
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.41338964790433913,
+      "learning_rate": 2.310351563279457e-05,
+      "loss": 0.6499,
+      "step": 5222
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3657570339882347,
+      "learning_rate": 2.309779364036419e-05,
+      "loss": 0.6326,
+      "step": 5223
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.42574349808594797,
+      "learning_rate": 2.3092071388125614e-05,
+      "loss": 0.6687,
+      "step": 5224
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3651339716868074,
+      "learning_rate": 2.308634887655874e-05,
+      "loss": 0.6752,
+      "step": 5225
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.49207364424403166,
+      "learning_rate": 2.3080626106143503e-05,
+      "loss": 0.7114,
+      "step": 5226
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.40585904136741485,
+      "learning_rate": 2.3074903077359888e-05,
+      "loss": 0.7051,
+      "step": 5227
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3645966489417215,
+      "learning_rate": 2.3069179790687867e-05,
+      "loss": 0.6692,
+      "step": 5228
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3793743310211244,
+      "learning_rate": 2.3063456246607432e-05,
+      "loss": 0.6523,
+      "step": 5229
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3782591284827242,
+      "learning_rate": 2.3057732445598625e-05,
+      "loss": 0.6778,
+      "step": 5230
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4174290276273555,
+      "learning_rate": 2.3052008388141495e-05,
+      "loss": 0.7193,
+      "step": 5231
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.38327474379589155,
+      "learning_rate": 2.30462840747161e-05,
+      "loss": 0.7023,
+      "step": 5232
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4194026218291612,
+      "learning_rate": 2.3040559505802545e-05,
+      "loss": 0.7018,
+      "step": 5233
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.40201708124301994,
+      "learning_rate": 2.3034834681880938e-05,
+      "loss": 0.6682,
+      "step": 5234
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3373168566826583,
+      "learning_rate": 2.302910960343141e-05,
+      "loss": 0.6511,
+      "step": 5235
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4315530534191385,
+      "learning_rate": 2.3023384270934107e-05,
+      "loss": 0.6495,
+      "step": 5236
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3933508192350129,
+      "learning_rate": 2.301765868486923e-05,
+      "loss": 0.695,
+      "step": 5237
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3241900785319391,
+      "learning_rate": 2.3011932845716963e-05,
+      "loss": 0.6589,
+      "step": 5238
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3681877005229578,
+      "learning_rate": 2.3006206753957518e-05,
+      "loss": 0.6546,
+      "step": 5239
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.38965615742471477,
+      "learning_rate": 2.3000480410071147e-05,
+      "loss": 0.6839,
+      "step": 5240
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4373668555960367,
+      "learning_rate": 2.2994753814538107e-05,
+      "loss": 0.6953,
+      "step": 5241
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.382661507336245,
+      "learning_rate": 2.2989026967838684e-05,
+      "loss": 0.6745,
+      "step": 5242
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3444166141519952,
+      "learning_rate": 2.2983299870453172e-05,
+      "loss": 0.6396,
+      "step": 5243
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3515442288904463,
+      "learning_rate": 2.297757252286191e-05,
+      "loss": 0.6721,
+      "step": 5244
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3900923946127971,
+      "learning_rate": 2.2971844925545234e-05,
+      "loss": 0.7073,
+      "step": 5245
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3999675010860228,
+      "learning_rate": 2.296611707898351e-05,
+      "loss": 0.7032,
+      "step": 5246
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4337067642371014,
+      "learning_rate": 2.2960388983657134e-05,
+      "loss": 0.6998,
+      "step": 5247
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3808036497252479,
+      "learning_rate": 2.2954660640046507e-05,
+      "loss": 0.668,
+      "step": 5248
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.42128735122181427,
+      "learning_rate": 2.294893204863206e-05,
+      "loss": 0.7067,
+      "step": 5249
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4286824545796449,
+      "learning_rate": 2.294320320989424e-05,
+      "loss": 0.7046,
+      "step": 5250
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4024357373882909,
+      "learning_rate": 2.293747412431353e-05,
+      "loss": 0.6292,
+      "step": 5251
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4489611016829186,
+      "learning_rate": 2.293174479237041e-05,
+      "loss": 0.7006,
+      "step": 5252
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.42481036302206443,
+      "learning_rate": 2.2926015214545393e-05,
+      "loss": 0.6947,
+      "step": 5253
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4076698440314306,
+      "learning_rate": 2.2920285391319007e-05,
+      "loss": 0.6901,
+      "step": 5254
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.43046389452666,
+      "learning_rate": 2.2914555323171824e-05,
+      "loss": 0.7505,
+      "step": 5255
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.36093083247101715,
+      "learning_rate": 2.29088250105844e-05,
+      "loss": 0.6503,
+      "step": 5256
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.34471759786759026,
+      "learning_rate": 2.2903094454037332e-05,
+      "loss": 0.6575,
+      "step": 5257
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4092030087064172,
+      "learning_rate": 2.2897363654011246e-05,
+      "loss": 0.6982,
+      "step": 5258
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.35635607897601274,
+      "learning_rate": 2.2891632610986766e-05,
+      "loss": 0.6564,
+      "step": 5259
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4145698044227466,
+      "learning_rate": 2.288590132544455e-05,
+      "loss": 0.6974,
+      "step": 5260
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.680935584947838,
+      "learning_rate": 2.2880169797865273e-05,
+      "loss": 7.2882,
+      "step": 5261
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4306289890010488,
+      "learning_rate": 2.287443802872964e-05,
+      "loss": 0.6915,
+      "step": 5262
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4113988710485602,
+      "learning_rate": 2.2868706018518356e-05,
+      "loss": 0.652,
+      "step": 5263
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.44007981766363125,
+      "learning_rate": 2.2862973767712165e-05,
+      "loss": 0.6934,
+      "step": 5264
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.41831835732742206,
+      "learning_rate": 2.285724127679182e-05,
+      "loss": 0.6606,
+      "step": 5265
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3999536254352327,
+      "learning_rate": 2.28515085462381e-05,
+      "loss": 0.7115,
+      "step": 5266
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3876950934163225,
+      "learning_rate": 2.2845775576531802e-05,
+      "loss": 0.6448,
+      "step": 5267
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3628974064417175,
+      "learning_rate": 2.2840042368153746e-05,
+      "loss": 0.6463,
+      "step": 5268
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.37029296718584015,
+      "learning_rate": 2.2834308921584763e-05,
+      "loss": 0.669,
+      "step": 5269
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.38767264960259584,
+      "learning_rate": 2.2828575237305715e-05,
+      "loss": 0.6543,
+      "step": 5270
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.36634687617249057,
+      "learning_rate": 2.2822841315797474e-05,
+      "loss": 0.665,
+      "step": 5271
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4048395644154523,
+      "learning_rate": 2.2817107157540947e-05,
+      "loss": 0.7125,
+      "step": 5272
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3968458108198756,
+      "learning_rate": 2.281137276301704e-05,
+      "loss": 0.6919,
+      "step": 5273
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39710468701920015,
+      "learning_rate": 2.2805638132706696e-05,
+      "loss": 0.6931,
+      "step": 5274
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3736450206066529,
+      "learning_rate": 2.2799903267090877e-05,
+      "loss": 0.6511,
+      "step": 5275
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.38044008259408485,
+      "learning_rate": 2.2794168166650543e-05,
+      "loss": 0.6835,
+      "step": 5276
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3235890517299611,
+      "learning_rate": 2.2788432831866708e-05,
+      "loss": 0.6222,
+      "step": 5277
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39733124191087876,
+      "learning_rate": 2.2782697263220374e-05,
+      "loss": 0.652,
+      "step": 5278
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4069722632204904,
+      "learning_rate": 2.277696146119258e-05,
+      "loss": 0.6796,
+      "step": 5279
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.36772393956784266,
+      "learning_rate": 2.2771225426264382e-05,
+      "loss": 0.6762,
+      "step": 5280
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39216348546579805,
+      "learning_rate": 2.2765489158916853e-05,
+      "loss": 0.6577,
+      "step": 5281
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3452367299304362,
+      "learning_rate": 2.2759752659631093e-05,
+      "loss": 0.6516,
+      "step": 5282
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3523412832065915,
+      "learning_rate": 2.275401592888821e-05,
+      "loss": 0.6829,
+      "step": 5283
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4555651588402231,
+      "learning_rate": 2.2748278967169335e-05,
+      "loss": 0.7252,
+      "step": 5284
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3562832144232101,
+      "learning_rate": 2.274254177495562e-05,
+      "loss": 0.6587,
+      "step": 5285
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3377041170653142,
+      "learning_rate": 2.2736804352728243e-05,
+      "loss": 0.6426,
+      "step": 5286
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39599142426611933,
+      "learning_rate": 2.273106670096839e-05,
+      "loss": 0.6723,
+      "step": 5287
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3979756649199984,
+      "learning_rate": 2.2725328820157263e-05,
+      "loss": 0.6644,
+      "step": 5288
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.36346447555802214,
+      "learning_rate": 2.2719590710776108e-05,
+      "loss": 0.6952,
+      "step": 5289
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3382508542441172,
+      "learning_rate": 2.271385237330616e-05,
+      "loss": 0.6513,
+      "step": 5290
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.366693767943543,
+      "learning_rate": 2.2708113808228697e-05,
+      "loss": 0.6446,
+      "step": 5291
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.37496002271706086,
+      "learning_rate": 2.2702375016024992e-05,
+      "loss": 0.6679,
+      "step": 5292
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3250153542119779,
+      "learning_rate": 2.2696635997176362e-05,
+      "loss": 0.6613,
+      "step": 5293
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39444303148792664,
+      "learning_rate": 2.2690896752164128e-05,
+      "loss": 0.6932,
+      "step": 5294
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1357081937954483,
+      "learning_rate": 2.268515728146963e-05,
+      "loss": 7.0888,
+      "step": 5295
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.38970703324227585,
+      "learning_rate": 2.2679417585574237e-05,
+      "loss": 0.6935,
+      "step": 5296
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39108430714748516,
+      "learning_rate": 2.2673677664959322e-05,
+      "loss": 0.7011,
+      "step": 5297
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4149508960456668,
+      "learning_rate": 2.26679375201063e-05,
+      "loss": 0.6823,
+      "step": 5298
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3624310449695396,
+      "learning_rate": 2.266219715149657e-05,
+      "loss": 0.6419,
+      "step": 5299
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39909333331869323,
+      "learning_rate": 2.2656456559611587e-05,
+      "loss": 0.6894,
+      "step": 5300
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3731512140268845,
+      "learning_rate": 2.2650715744932796e-05,
+      "loss": 0.6885,
+      "step": 5301
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.38498536258157806,
+      "learning_rate": 2.2644974707941677e-05,
+      "loss": 0.6669,
+      "step": 5302
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4335641406336333,
+      "learning_rate": 2.263923344911973e-05,
+      "loss": 0.6861,
+      "step": 5303
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3688748497268878,
+      "learning_rate": 2.2633491968948454e-05,
+      "loss": 0.6686,
+      "step": 5304
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.36829210873736923,
+      "learning_rate": 2.262775026790939e-05,
+      "loss": 0.6295,
+      "step": 5305
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4489345140522624,
+      "learning_rate": 2.2622008346484082e-05,
+      "loss": 0.6919,
+      "step": 5306
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3337138453821074,
+      "learning_rate": 2.26162662051541e-05,
+      "loss": 0.6313,
+      "step": 5307
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.35723436729342034,
+      "learning_rate": 2.261052384440104e-05,
+      "loss": 0.682,
+      "step": 5308
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 3.126408618311136,
+      "learning_rate": 2.2604781264706482e-05,
+      "loss": 7.334,
+      "step": 5309
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39144177912832434,
+      "learning_rate": 2.2599038466552075e-05,
+      "loss": 0.7142,
+      "step": 5310
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4247730849566587,
+      "learning_rate": 2.2593295450419452e-05,
+      "loss": 0.6759,
+      "step": 5311
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3826084284757317,
+      "learning_rate": 2.2587552216790272e-05,
+      "loss": 0.7014,
+      "step": 5312
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.412530687935625,
+      "learning_rate": 2.258180876614621e-05,
+      "loss": 0.6566,
+      "step": 5313
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4070171946538816,
+      "learning_rate": 2.257606509896896e-05,
+      "loss": 0.6391,
+      "step": 5314
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3798059883711926,
+      "learning_rate": 2.257032121574025e-05,
+      "loss": 0.6675,
+      "step": 5315
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.376389595276396,
+      "learning_rate": 2.25645771169418e-05,
+      "loss": 0.6785,
+      "step": 5316
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.37558306261179136,
+      "learning_rate": 2.255883280305536e-05,
+      "loss": 0.7155,
+      "step": 5317
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4067275148313486,
+      "learning_rate": 2.255308827456271e-05,
+      "loss": 0.6914,
+      "step": 5318
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.45636169216485406,
+      "learning_rate": 2.2547343531945626e-05,
+      "loss": 0.7081,
+      "step": 5319
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4158499962359905,
+      "learning_rate": 2.2541598575685914e-05,
+      "loss": 0.7059,
+      "step": 5320
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.367933307672749,
+      "learning_rate": 2.2535853406265408e-05,
+      "loss": 0.675,
+      "step": 5321
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.35833109439373356,
+      "learning_rate": 2.2530108024165936e-05,
+      "loss": 0.6481,
+      "step": 5322
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3744489494643644,
+      "learning_rate": 2.2524362429869356e-05,
+      "loss": 0.6874,
+      "step": 5323
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.358724079253473,
+      "learning_rate": 2.2518616623857542e-05,
+      "loss": 0.6444,
+      "step": 5324
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.40409724490959764,
+      "learning_rate": 2.2512870606612407e-05,
+      "loss": 0.6525,
+      "step": 5325
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4312302204744032,
+      "learning_rate": 2.2507124378615842e-05,
+      "loss": 0.696,
+      "step": 5326
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.36859950741297126,
+      "learning_rate": 2.250137794034978e-05,
+      "loss": 0.7029,
+      "step": 5327
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.42199048950394785,
+      "learning_rate": 2.2495631292296176e-05,
+      "loss": 0.708,
+      "step": 5328
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.41784093324069127,
+      "learning_rate": 2.248988443493699e-05,
+      "loss": 0.7209,
+      "step": 5329
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.33152818000176837,
+      "learning_rate": 2.2484137368754203e-05,
+      "loss": 0.6645,
+      "step": 5330
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4942262606515672,
+      "learning_rate": 2.2478390094229807e-05,
+      "loss": 0.6716,
+      "step": 5331
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.36614810965632516,
+      "learning_rate": 2.2472642611845837e-05,
+      "loss": 0.6531,
+      "step": 5332
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.37022413706420054,
+      "learning_rate": 2.2466894922084315e-05,
+      "loss": 0.7186,
+      "step": 5333
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3562018531515345,
+      "learning_rate": 2.246114702542729e-05,
+      "loss": 0.668,
+      "step": 5334
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.38997870407432766,
+      "learning_rate": 2.2455398922356846e-05,
+      "loss": 0.7078,
+      "step": 5335
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3552830366132395,
+      "learning_rate": 2.2449650613355052e-05,
+      "loss": 0.6704,
+      "step": 5336
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3792755872294333,
+      "learning_rate": 2.2443902098904018e-05,
+      "loss": 0.664,
+      "step": 5337
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.36585596795428915,
+      "learning_rate": 2.243815337948587e-05,
+      "loss": 0.6706,
+      "step": 5338
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.37927342905218786,
+      "learning_rate": 2.2432404455582745e-05,
+      "loss": 0.6732,
+      "step": 5339
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3606873677623572,
+      "learning_rate": 2.2426655327676793e-05,
+      "loss": 0.6919,
+      "step": 5340
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4111600433795266,
+      "learning_rate": 2.242090599625019e-05,
+      "loss": 0.7183,
+      "step": 5341
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.34886200085789204,
+      "learning_rate": 2.2415156461785122e-05,
+      "loss": 0.6524,
+      "step": 5342
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5050518203576889,
+      "learning_rate": 2.2409406724763808e-05,
+      "loss": 0.7099,
+      "step": 5343
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3569558812375592,
+      "learning_rate": 2.240365678566845e-05,
+      "loss": 0.6814,
+      "step": 5344
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4068717387661615,
+      "learning_rate": 2.2397906644981306e-05,
+      "loss": 0.6916,
+      "step": 5345
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.36386804175106474,
+      "learning_rate": 2.2392156303184627e-05,
+      "loss": 0.6639,
+      "step": 5346
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3740550753935611,
+      "learning_rate": 2.2386405760760687e-05,
+      "loss": 0.6684,
+      "step": 5347
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.42650559116233594,
+      "learning_rate": 2.238065501819178e-05,
+      "loss": 0.6648,
+      "step": 5348
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3949453754979329,
+      "learning_rate": 2.237490407596021e-05,
+      "loss": 0.6676,
+      "step": 5349
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.41606149972893597,
+      "learning_rate": 2.23691529345483e-05,
+      "loss": 0.6843,
+      "step": 5350
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3961851008382366,
+      "learning_rate": 2.2363401594438398e-05,
+      "loss": 0.6892,
+      "step": 5351
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3204342729831388,
+      "learning_rate": 2.235765005611286e-05,
+      "loss": 0.622,
+      "step": 5352
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4244437000446791,
+      "learning_rate": 2.2351898320054062e-05,
+      "loss": 0.7083,
+      "step": 5353
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.320490389879709,
+      "learning_rate": 2.234614638674438e-05,
+      "loss": 0.6603,
+      "step": 5354
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.38411426687408406,
+      "learning_rate": 2.2340394256666244e-05,
+      "loss": 0.695,
+      "step": 5355
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.36603235000435247,
+      "learning_rate": 2.2334641930302066e-05,
+      "loss": 0.5863,
+      "step": 5356
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.41058884657830447,
+      "learning_rate": 2.232888940813429e-05,
+      "loss": 0.6832,
+      "step": 5357
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.406434711285097,
+      "learning_rate": 2.2323136690645368e-05,
+      "loss": 0.6383,
+      "step": 5358
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3710597426647069,
+      "learning_rate": 2.2317383778317783e-05,
+      "loss": 0.6548,
+      "step": 5359
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3897816647101547,
+      "learning_rate": 2.231163067163402e-05,
+      "loss": 0.6599,
+      "step": 5360
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3556561110775065,
+      "learning_rate": 2.2305877371076576e-05,
+      "loss": 0.6657,
+      "step": 5361
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3789769209975138,
+      "learning_rate": 2.2300123877127987e-05,
+      "loss": 0.6337,
+      "step": 5362
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.42090808363755,
+      "learning_rate": 2.2294370190270792e-05,
+      "loss": 0.6877,
+      "step": 5363
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4079302868962553,
+      "learning_rate": 2.228861631098753e-05,
+      "loss": 0.6553,
+      "step": 5364
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3641323054863593,
+      "learning_rate": 2.2282862239760786e-05,
+      "loss": 0.7182,
+      "step": 5365
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.35745799743049395,
+      "learning_rate": 2.227710797707314e-05,
+      "loss": 0.6485,
+      "step": 5366
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3683552152308115,
+      "learning_rate": 2.22713535234072e-05,
+      "loss": 0.6777,
+      "step": 5367
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.37125872009142985,
+      "learning_rate": 2.2265598879245583e-05,
+      "loss": 0.6648,
+      "step": 5368
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.5751361494761875,
+      "learning_rate": 2.225984404507093e-05,
+      "loss": 7.1468,
+      "step": 5369
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3765432530213137,
+      "learning_rate": 2.2254089021365882e-05,
+      "loss": 0.6681,
+      "step": 5370
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.45017830546241683,
+      "learning_rate": 2.22483338086131e-05,
+      "loss": 0.6908,
+      "step": 5371
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3932964434202794,
+      "learning_rate": 2.224257840729529e-05,
+      "loss": 0.6489,
+      "step": 5372
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3921777527599384,
+      "learning_rate": 2.2236822817895133e-05,
+      "loss": 0.6874,
+      "step": 5373
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.385523537017858,
+      "learning_rate": 2.2231067040895348e-05,
+      "loss": 0.6978,
+      "step": 5374
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.38395256039402087,
+      "learning_rate": 2.2225311076778663e-05,
+      "loss": 0.6638,
+      "step": 5375
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3512827450331755,
+      "learning_rate": 2.221955492602783e-05,
+      "loss": 0.6078,
+      "step": 5376
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4256255582199718,
+      "learning_rate": 2.2213798589125603e-05,
+      "loss": 0.6719,
+      "step": 5377
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4599119100063296,
+      "learning_rate": 2.220804206655476e-05,
+      "loss": 0.7208,
+      "step": 5378
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4065832569105626,
+      "learning_rate": 2.2202285358798096e-05,
+      "loss": 0.6558,
+      "step": 5379
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.37375371138894387,
+      "learning_rate": 2.2196528466338424e-05,
+      "loss": 0.672,
+      "step": 5380
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4392880129797536,
+      "learning_rate": 2.219077138965856e-05,
+      "loss": 0.7394,
+      "step": 5381
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.35058246755670375,
+      "learning_rate": 2.218501412924134e-05,
+      "loss": 0.6689,
+      "step": 5382
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3898098863234877,
+      "learning_rate": 2.217925668556963e-05,
+      "loss": 0.6729,
+      "step": 5383
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.27600372261888,
+      "learning_rate": 2.2173499059126285e-05,
+      "loss": 7.1118,
+      "step": 5384
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3879941802175919,
+      "learning_rate": 2.2167741250394208e-05,
+      "loss": 0.6338,
+      "step": 5385
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3523904167602684,
+      "learning_rate": 2.2161983259856287e-05,
+      "loss": 0.6513,
+      "step": 5386
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4079041238110573,
+      "learning_rate": 2.2156225087995446e-05,
+      "loss": 0.6593,
+      "step": 5387
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.45675671986462885,
+      "learning_rate": 2.2150466735294605e-05,
+      "loss": 0.7014,
+      "step": 5388
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3865070115863227,
+      "learning_rate": 2.2144708202236712e-05,
+      "loss": 0.6687,
+      "step": 5389
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.40638057050220777,
+      "learning_rate": 2.2138949489304743e-05,
+      "loss": 0.6179,
+      "step": 5390
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3889506978751414,
+      "learning_rate": 2.2133190596981655e-05,
+      "loss": 0.655,
+      "step": 5391
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.39049896166180836,
+      "learning_rate": 2.212743152575045e-05,
+      "loss": 0.6292,
+      "step": 5392
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4513030244848252,
+      "learning_rate": 2.2121672276094132e-05,
+      "loss": 0.7259,
+      "step": 5393
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.40504875282847846,
+      "learning_rate": 2.2115912848495725e-05,
+      "loss": 0.6275,
+      "step": 5394
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4270697197094342,
+      "learning_rate": 2.2110153243438264e-05,
+      "loss": 0.7074,
+      "step": 5395
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.39486916091465896,
+      "learning_rate": 2.210439346140479e-05,
+      "loss": 0.7074,
+      "step": 5396
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4365095194402751,
+      "learning_rate": 2.2098633502878388e-05,
+      "loss": 0.6324,
+      "step": 5397
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.41090757191448596,
+      "learning_rate": 2.2092873368342126e-05,
+      "loss": 0.7044,
+      "step": 5398
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4077183719619472,
+      "learning_rate": 2.20871130582791e-05,
+      "loss": 0.65,
+      "step": 5399
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.38758630070559613,
+      "learning_rate": 2.2081352573172424e-05,
+      "loss": 0.66,
+      "step": 5400
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3641579188509184,
+      "learning_rate": 2.2075591913505218e-05,
+      "loss": 0.6746,
+      "step": 5401
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.45350228636289147,
+      "learning_rate": 2.2069831079760625e-05,
+      "loss": 0.7359,
+      "step": 5402
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3125896835152207,
+      "learning_rate": 2.2064070072421804e-05,
+      "loss": 0.66,
+      "step": 5403
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3560767328426949,
+      "learning_rate": 2.205830889197192e-05,
+      "loss": 0.6774,
+      "step": 5404
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.393326373837524,
+      "learning_rate": 2.205254753889415e-05,
+      "loss": 0.6788,
+      "step": 5405
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4009772413975159,
+      "learning_rate": 2.2046786013671693e-05,
+      "loss": 0.6768,
+      "step": 5406
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4053707289065991,
+      "learning_rate": 2.2041024316787775e-05,
+      "loss": 0.6289,
+      "step": 5407
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.39687181466933047,
+      "learning_rate": 2.2035262448725606e-05,
+      "loss": 0.6716,
+      "step": 5408
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3871192832805118,
+      "learning_rate": 2.2029500409968435e-05,
+      "loss": 0.6929,
+      "step": 5409
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.36633777763081316,
+      "learning_rate": 2.2023738200999517e-05,
+      "loss": 0.6827,
+      "step": 5410
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.34956132505889753,
+      "learning_rate": 2.201797582230212e-05,
+      "loss": 0.6377,
+      "step": 5411
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.40466760947998703,
+      "learning_rate": 2.2012213274359526e-05,
+      "loss": 0.6622,
+      "step": 5412
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3746033993032864,
+      "learning_rate": 2.2006450557655034e-05,
+      "loss": 0.6571,
+      "step": 5413
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4058518272653098,
+      "learning_rate": 2.2000687672671963e-05,
+      "loss": 0.6821,
+      "step": 5414
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.35925136301727695,
+      "learning_rate": 2.1994924619893623e-05,
+      "loss": 0.6409,
+      "step": 5415
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.37878069646444273,
+      "learning_rate": 2.198916139980337e-05,
+      "loss": 0.6487,
+      "step": 5416
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.40300442397294833,
+      "learning_rate": 2.1983398012884552e-05,
+      "loss": 0.68,
+      "step": 5417
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3636350770680979,
+      "learning_rate": 2.197763445962054e-05,
+      "loss": 0.6589,
+      "step": 5418
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3753720340179966,
+      "learning_rate": 2.1971870740494712e-05,
+      "loss": 0.6647,
+      "step": 5419
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.41967776608808954,
+      "learning_rate": 2.196610685599047e-05,
+      "loss": 0.6538,
+      "step": 5420
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.36994793567327783,
+      "learning_rate": 2.196034280659122e-05,
+      "loss": 0.6311,
+      "step": 5421
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3966235216103406,
+      "learning_rate": 2.195457859278038e-05,
+      "loss": 0.6865,
+      "step": 5422
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3395630515093548,
+      "learning_rate": 2.1948814215041395e-05,
+      "loss": 0.6566,
+      "step": 5423
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.41431544018084,
+      "learning_rate": 2.194304967385772e-05,
+      "loss": 0.7228,
+      "step": 5424
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.39868213059103225,
+      "learning_rate": 2.1937284969712812e-05,
+      "loss": 0.6975,
+      "step": 5425
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4268948905554942,
+      "learning_rate": 2.193152010309016e-05,
+      "loss": 0.6421,
+      "step": 5426
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 5.062122320572271,
+      "learning_rate": 2.192575507447324e-05,
+      "loss": 7.2133,
+      "step": 5427
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3524412994541918,
+      "learning_rate": 2.1919989884345575e-05,
+      "loss": 0.6652,
+      "step": 5428
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.38012398319690294,
+      "learning_rate": 2.1914224533190676e-05,
+      "loss": 0.6573,
+      "step": 5429
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3948289045779377,
+      "learning_rate": 2.190845902149207e-05,
+      "loss": 0.6602,
+      "step": 5430
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4062647465887642,
+      "learning_rate": 2.190269334973332e-05,
+      "loss": 0.676,
+      "step": 5431
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.37515221842162355,
+      "learning_rate": 2.189692751839797e-05,
+      "loss": 0.6645,
+      "step": 5432
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.36945727739541445,
+      "learning_rate": 2.1891161527969603e-05,
+      "loss": 0.6746,
+      "step": 5433
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.43588131073212516,
+      "learning_rate": 2.1885395378931805e-05,
+      "loss": 0.7151,
+      "step": 5434
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3420030441392762,
+      "learning_rate": 2.1879629071768166e-05,
+      "loss": 0.6679,
+      "step": 5435
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.390590443262345,
+      "learning_rate": 2.1873862606962316e-05,
+      "loss": 0.6865,
+      "step": 5436
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.39900589873259745,
+      "learning_rate": 2.1868095984997865e-05,
+      "loss": 0.6898,
+      "step": 5437
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4415028706934783,
+      "learning_rate": 2.1862329206358465e-05,
+      "loss": 0.7212,
+      "step": 5438
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4027990554228184,
+      "learning_rate": 2.185656227152776e-05,
+      "loss": 0.6749,
+      "step": 5439
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.422131125316762,
+      "learning_rate": 2.1850795180989427e-05,
+      "loss": 0.6429,
+      "step": 5440
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3938405643656958,
+      "learning_rate": 2.1845027935227128e-05,
+      "loss": 0.6883,
+      "step": 5441
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.34374320355916543,
+      "learning_rate": 2.1839260534724564e-05,
+      "loss": 0.6292,
+      "step": 5442
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.41460947992023384,
+      "learning_rate": 2.1833492979965446e-05,
+      "loss": 0.6488,
+      "step": 5443
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.39790979094199336,
+      "learning_rate": 2.1827725271433483e-05,
+      "loss": 0.6755,
+      "step": 5444
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.38429289141591194,
+      "learning_rate": 2.1821957409612405e-05,
+      "loss": 0.661,
+      "step": 5445
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.386287199069429,
+      "learning_rate": 2.1816189394985967e-05,
+      "loss": 0.6754,
+      "step": 5446
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4267371491560869,
+      "learning_rate": 2.1810421228037913e-05,
+      "loss": 0.6775,
+      "step": 5447
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.37165686975663004,
+      "learning_rate": 2.1804652909252024e-05,
+      "loss": 0.7007,
+      "step": 5448
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4279754365168446,
+      "learning_rate": 2.1798884439112066e-05,
+      "loss": 0.6045,
+      "step": 5449
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3614972593376467,
+      "learning_rate": 2.1793115818101853e-05,
+      "loss": 0.6582,
+      "step": 5450
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.36244468320105727,
+      "learning_rate": 2.1787347046705172e-05,
+      "loss": 0.665,
+      "step": 5451
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4383610289923083,
+      "learning_rate": 2.1781578125405852e-05,
+      "loss": 0.7264,
+      "step": 5452
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4152382771832305,
+      "learning_rate": 2.1775809054687734e-05,
+      "loss": 0.6688,
+      "step": 5453
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3864530705100902,
+      "learning_rate": 2.177003983503465e-05,
+      "loss": 0.6781,
+      "step": 5454
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.37561997340665915,
+      "learning_rate": 2.1764270466930465e-05,
+      "loss": 0.659,
+      "step": 5455
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.353470386197914,
+      "learning_rate": 2.1758500950859046e-05,
+      "loss": 0.6312,
+      "step": 5456
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.39154947548729335,
+      "learning_rate": 2.175273128730428e-05,
+      "loss": 0.6946,
+      "step": 5457
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.42409704959329986,
+      "learning_rate": 2.174696147675005e-05,
+      "loss": 0.7216,
+      "step": 5458
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.43593310012666137,
+      "learning_rate": 2.1741191519680273e-05,
+      "loss": 0.6711,
+      "step": 5459
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 6.865948770996866,
+      "learning_rate": 2.1735421416578867e-05,
+      "loss": 7.262,
+      "step": 5460
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3836249168293492,
+      "learning_rate": 2.1729651167929763e-05,
+      "loss": 0.6654,
+      "step": 5461
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.41470468633383945,
+      "learning_rate": 2.1723880774216902e-05,
+      "loss": 0.622,
+      "step": 5462
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3419382032510597,
+      "learning_rate": 2.171811023592425e-05,
+      "loss": 0.6565,
+      "step": 5463
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3640805768098906,
+      "learning_rate": 2.1712339553535766e-05,
+      "loss": 0.6287,
+      "step": 5464
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.36198422419888115,
+      "learning_rate": 2.1706568727535422e-05,
+      "loss": 0.6236,
+      "step": 5465
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4277577415353558,
+      "learning_rate": 2.1700797758407225e-05,
+      "loss": 0.6522,
+      "step": 5466
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.42054013884243263,
+      "learning_rate": 2.1695026646635177e-05,
+      "loss": 0.6697,
+      "step": 5467
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.40888180011209674,
+      "learning_rate": 2.1689255392703288e-05,
+      "loss": 0.6627,
+      "step": 5468
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.386842776626424,
+      "learning_rate": 2.1683483997095588e-05,
+      "loss": 0.6479,
+      "step": 5469
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.40604453535683954,
+      "learning_rate": 2.1677712460296125e-05,
+      "loss": 0.7063,
+      "step": 5470
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3489694048228632,
+      "learning_rate": 2.167194078278894e-05,
+      "loss": 0.6866,
+      "step": 5471
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4088775749372993,
+      "learning_rate": 2.1666168965058096e-05,
+      "loss": 0.6867,
+      "step": 5472
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3973274996779672,
+      "learning_rate": 2.1660397007587682e-05,
+      "loss": 0.704,
+      "step": 5473
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.38320427466113266,
+      "learning_rate": 2.165462491086177e-05,
+      "loss": 0.6982,
+      "step": 5474
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3623867754460067,
+      "learning_rate": 2.164885267536447e-05,
+      "loss": 0.663,
+      "step": 5475
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.34047452160406294,
+      "learning_rate": 2.164308030157988e-05,
+      "loss": 0.6537,
+      "step": 5476
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3954570557264004,
+      "learning_rate": 2.1637307789992135e-05,
+      "loss": 0.677,
+      "step": 5477
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.34641137458693444,
+      "learning_rate": 2.1631535141085362e-05,
+      "loss": 0.6651,
+      "step": 5478
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3757812716444577,
+      "learning_rate": 2.1625762355343706e-05,
+      "loss": 0.6558,
+      "step": 5479
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.39749706279429853,
+      "learning_rate": 2.1619989433251325e-05,
+      "loss": 0.6616,
+      "step": 5480
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.39950155561282136,
+      "learning_rate": 2.1614216375292392e-05,
+      "loss": 0.7096,
+      "step": 5481
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3687454550179657,
+      "learning_rate": 2.160844318195107e-05,
+      "loss": 0.6594,
+      "step": 5482
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3862803279507181,
+      "learning_rate": 2.160266985371157e-05,
+      "loss": 0.6934,
+      "step": 5483
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3702336249071157,
+      "learning_rate": 2.1596896391058082e-05,
+      "loss": 0.6795,
+      "step": 5484
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3662203695052902,
+      "learning_rate": 2.159112279447482e-05,
+      "loss": 0.6335,
+      "step": 5485
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.349195326303672,
+      "learning_rate": 2.1585349064446013e-05,
+      "loss": 0.6418,
+      "step": 5486
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.39444901497058577,
+      "learning_rate": 2.1579575201455894e-05,
+      "loss": 0.6976,
+      "step": 5487
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3880440585070764,
+      "learning_rate": 2.157380120598872e-05,
+      "loss": 0.6658,
+      "step": 5488
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.42142350306958054,
+      "learning_rate": 2.156802707852872e-05,
+      "loss": 0.7065,
+      "step": 5489
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 6.378935795026131,
+      "learning_rate": 2.1562252819560197e-05,
+      "loss": 7.6121,
+      "step": 5490
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.457410804862877,
+      "learning_rate": 2.1556478429567415e-05,
+      "loss": 0.6826,
+      "step": 5491
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.48433316331024356,
+      "learning_rate": 2.1550703909034666e-05,
+      "loss": 0.6529,
+      "step": 5492
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3854726691985786,
+      "learning_rate": 2.154492925844625e-05,
+      "loss": 0.6709,
+      "step": 5493
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.391847359353783,
+      "learning_rate": 2.1539154478286488e-05,
+      "loss": 0.6704,
+      "step": 5494
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.423106863653123,
+      "learning_rate": 2.153337956903969e-05,
+      "loss": 0.6659,
+      "step": 5495
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3640945015133437,
+      "learning_rate": 2.1527604531190205e-05,
+      "loss": 0.6544,
+      "step": 5496
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3357468728445515,
+      "learning_rate": 2.1521829365222376e-05,
+      "loss": 0.6557,
+      "step": 5497
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3899522567387217,
+      "learning_rate": 2.1516054071620552e-05,
+      "loss": 0.6664,
+      "step": 5498
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3589434242051042,
+      "learning_rate": 2.1510278650869098e-05,
+      "loss": 0.6702,
+      "step": 5499
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4048127659906414,
+      "learning_rate": 2.150450310345241e-05,
+      "loss": 0.6928,
+      "step": 5500
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3900684613951847,
+      "learning_rate": 2.149872742985486e-05,
+      "loss": 0.6454,
+      "step": 5501
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4169382674499063,
+      "learning_rate": 2.149295163056085e-05,
+      "loss": 0.6991,
+      "step": 5502
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4557242430152251,
+      "learning_rate": 2.148717570605478e-05,
+      "loss": 0.6977,
+      "step": 5503
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.423129678493281,
+      "learning_rate": 2.1481399656821092e-05,
+      "loss": 0.643,
+      "step": 5504
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.45288586762582855,
+      "learning_rate": 2.1475623483344204e-05,
+      "loss": 0.7039,
+      "step": 5505
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.38861181642095827,
+      "learning_rate": 2.1469847186108542e-05,
+      "loss": 0.6515,
+      "step": 5506
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3331760459294499,
+      "learning_rate": 2.1464070765598584e-05,
+      "loss": 0.6203,
+      "step": 5507
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.38414987738044476,
+      "learning_rate": 2.1458294222298784e-05,
+      "loss": 0.6513,
+      "step": 5508
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3879562762070006,
+      "learning_rate": 2.14525175566936e-05,
+      "loss": 0.6623,
+      "step": 5509
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3981934709599509,
+      "learning_rate": 2.144674076926752e-05,
+      "loss": 0.6731,
+      "step": 5510
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.36385914692801496,
+      "learning_rate": 2.144096386050505e-05,
+      "loss": 0.6571,
+      "step": 5511
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3702914844004559,
+      "learning_rate": 2.143518683089067e-05,
+      "loss": 0.6619,
+      "step": 5512
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.37119416922895626,
+      "learning_rate": 2.1429409680908905e-05,
+      "loss": 0.6539,
+      "step": 5513
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.2684108878044107,
+      "learning_rate": 2.1423632411044283e-05,
+      "loss": 7.4136,
+      "step": 5514
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3387495627773984,
+      "learning_rate": 2.141785502178133e-05,
+      "loss": 0.6271,
+      "step": 5515
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0848570064740617,
+      "learning_rate": 2.1412077513604582e-05,
+      "loss": 7.4603,
+      "step": 5516
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3989953556396006,
+      "learning_rate": 2.14062998869986e-05,
+      "loss": 0.6725,
+      "step": 5517
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3637252093881462,
+      "learning_rate": 2.140052214244795e-05,
+      "loss": 0.6696,
+      "step": 5518
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.37650808286446447,
+      "learning_rate": 2.1394744280437194e-05,
+      "loss": 0.6684,
+      "step": 5519
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4169116967017954,
+      "learning_rate": 2.1388966301450916e-05,
+      "loss": 0.6421,
+      "step": 5520
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3723662238731368,
+      "learning_rate": 2.1383188205973723e-05,
+      "loss": 0.6921,
+      "step": 5521
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3792506332444056,
+      "learning_rate": 2.13774099944902e-05,
+      "loss": 0.6884,
+      "step": 5522
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.41986513284604016,
+      "learning_rate": 2.1371631667484963e-05,
+      "loss": 0.7018,
+      "step": 5523
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4416654523479833,
+      "learning_rate": 2.1365853225442633e-05,
+      "loss": 0.7159,
+      "step": 5524
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.41746119726272773,
+      "learning_rate": 2.136007466884785e-05,
+      "loss": 0.6685,
+      "step": 5525
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.36792515697803385,
+      "learning_rate": 2.1354295998185238e-05,
+      "loss": 0.6822,
+      "step": 5526
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.36771207030295633,
+      "learning_rate": 2.134851721393946e-05,
+      "loss": 0.6686,
+      "step": 5527
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3280907646468596,
+      "learning_rate": 2.134273831659517e-05,
+      "loss": 0.6589,
+      "step": 5528
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.36870398956106465,
+      "learning_rate": 2.1336959306637044e-05,
+      "loss": 0.6412,
+      "step": 5529
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3606163144386984,
+      "learning_rate": 2.133118018454975e-05,
+      "loss": 0.6543,
+      "step": 5530
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.39471862249314826,
+      "learning_rate": 2.1325400950817988e-05,
+      "loss": 0.6844,
+      "step": 5531
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3806158725041603,
+      "learning_rate": 2.1319621605926448e-05,
+      "loss": 0.654,
+      "step": 5532
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3322527960829883,
+      "learning_rate": 2.131384215035983e-05,
+      "loss": 0.6878,
+      "step": 5533
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3983164158219578,
+      "learning_rate": 2.1308062584602865e-05,
+      "loss": 0.6829,
+      "step": 5534
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.396461714742024,
+      "learning_rate": 2.130228290914027e-05,
+      "loss": 0.6895,
+      "step": 5535
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.30793992669319775,
+      "learning_rate": 2.1296503124456774e-05,
+      "loss": 0.6197,
+      "step": 5536
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3924671065092576,
+      "learning_rate": 2.1290723231037132e-05,
+      "loss": 0.7206,
+      "step": 5537
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.35675063143922175,
+      "learning_rate": 2.128494322936609e-05,
+      "loss": 0.6799,
+      "step": 5538
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.37553366404642763,
+      "learning_rate": 2.127916311992841e-05,
+      "loss": 0.6773,
+      "step": 5539
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.32102477711080185,
+      "learning_rate": 2.127338290320887e-05,
+      "loss": 0.6034,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4221668144925308,
+      "learning_rate": 2.126760257969224e-05,
+      "loss": 0.6729,
+      "step": 5541
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3511646106316305,
+      "learning_rate": 2.126182214986331e-05,
+      "loss": 0.6417,
+      "step": 5542
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.300613837939728,
+      "learning_rate": 2.125604161420688e-05,
+      "loss": 7.3437,
+      "step": 5543
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4468800978972018,
+      "learning_rate": 2.1250260973207757e-05,
+      "loss": 0.7106,
+      "step": 5544
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4105430809648359,
+      "learning_rate": 2.1244480227350763e-05,
+      "loss": 0.6779,
+      "step": 5545
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.47906520538335695,
+      "learning_rate": 2.123869937712071e-05,
+      "loss": 0.7042,
+      "step": 5546
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.42851788751335984,
+      "learning_rate": 2.1232918423002437e-05,
+      "loss": 0.681,
+      "step": 5547
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4219542534219292,
+      "learning_rate": 2.1227137365480785e-05,
+      "loss": 0.6869,
+      "step": 5548
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3982587690137404,
+      "learning_rate": 2.1221356205040604e-05,
+      "loss": 0.6543,
+      "step": 5549
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3608723726277492,
+      "learning_rate": 2.121557494216675e-05,
+      "loss": 0.6552,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4493959769071166,
+      "learning_rate": 2.1209793577344097e-05,
+      "loss": 0.7681,
+      "step": 5551
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4460467446901692,
+      "learning_rate": 2.120401211105752e-05,
+      "loss": 0.7079,
+      "step": 5552
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3588358219685667,
+      "learning_rate": 2.1198230543791895e-05,
+      "loss": 0.6603,
+      "step": 5553
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.40164564098985794,
+      "learning_rate": 2.1192448876032125e-05,
+      "loss": 0.7082,
+      "step": 5554
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3652891443619659,
+      "learning_rate": 2.1186667108263108e-05,
+      "loss": 0.6695,
+      "step": 5555
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3523899420361972,
+      "learning_rate": 2.118088524096975e-05,
+      "loss": 0.6646,
+      "step": 5556
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.41166833563600863,
+      "learning_rate": 2.117510327463698e-05,
+      "loss": 0.7048,
+      "step": 5557
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3598236262987743,
+      "learning_rate": 2.1169321209749713e-05,
+      "loss": 0.6512,
+      "step": 5558
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.41461484551275396,
+      "learning_rate": 2.116353904679289e-05,
+      "loss": 0.6989,
+      "step": 5559
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.40942191387110977,
+      "learning_rate": 2.1157756786251447e-05,
+      "loss": 0.6835,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.488909577827893,
+      "learning_rate": 2.115197442861035e-05,
+      "loss": 0.6879,
+      "step": 5561
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4130331710232056,
+      "learning_rate": 2.1146191974354542e-05,
+      "loss": 0.639,
+      "step": 5562
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3696170746136158,
+      "learning_rate": 2.1140409423969e-05,
+      "loss": 0.7103,
+      "step": 5563
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3895593410548085,
+      "learning_rate": 2.1134626777938698e-05,
+      "loss": 0.6767,
+      "step": 5564
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.39497368616162737,
+      "learning_rate": 2.1128844036748616e-05,
+      "loss": 0.6765,
+      "step": 5565
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.38006986996821224,
+      "learning_rate": 2.1123061200883748e-05,
+      "loss": 0.7036,
+      "step": 5566
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.35059490390954035,
+      "learning_rate": 2.111727827082909e-05,
+      "loss": 0.641,
+      "step": 5567
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3345181991018341,
+      "learning_rate": 2.111149524706966e-05,
+      "loss": 0.6468,
+      "step": 5568
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.933930857240372,
+      "learning_rate": 2.1105712130090464e-05,
+      "loss": 7.4761,
+      "step": 5569
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3450392549646932,
+      "learning_rate": 2.1099928920376525e-05,
+      "loss": 0.6619,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3534524924855958,
+      "learning_rate": 2.1094145618412875e-05,
+      "loss": 0.6491,
+      "step": 5571
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.396671446485618,
+      "learning_rate": 2.1088362224684555e-05,
+      "loss": 0.6779,
+      "step": 5572
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3678157431839391,
+      "learning_rate": 2.1082578739676607e-05,
+      "loss": 0.6582,
+      "step": 5573
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3758767180549978,
+      "learning_rate": 2.1076795163874096e-05,
+      "loss": 0.6788,
+      "step": 5574
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3974596001822545,
+      "learning_rate": 2.1071011497762065e-05,
+      "loss": 0.6877,
+      "step": 5575
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4021321705850788,
+      "learning_rate": 2.10652277418256e-05,
+      "loss": 0.7131,
+      "step": 5576
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.37390105279068997,
+      "learning_rate": 2.1059443896549766e-05,
+      "loss": 0.6639,
+      "step": 5577
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.37531652277567534,
+      "learning_rate": 2.1053659962419654e-05,
+      "loss": 0.6611,
+      "step": 5578
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.35689253635310203,
+      "learning_rate": 2.1047875939920353e-05,
+      "loss": 0.6409,
+      "step": 5579
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3977646771615265,
+      "learning_rate": 2.1042091829536965e-05,
+      "loss": 0.6603,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.850598004502237,
+      "learning_rate": 2.103630763175459e-05,
+      "loss": 7.2512,
+      "step": 5581
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.39191068645213706,
+      "learning_rate": 2.1030523347058348e-05,
+      "loss": 0.6889,
+      "step": 5582
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.41656730557335764,
+      "learning_rate": 2.1024738975933357e-05,
+      "loss": 0.6621,
+      "step": 5583
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3559556682257004,
+      "learning_rate": 2.1018954518864744e-05,
+      "loss": 0.6282,
+      "step": 5584
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.360452090656393,
+      "learning_rate": 2.101316997633765e-05,
+      "loss": 0.6805,
+      "step": 5585
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.37730361556492514,
+      "learning_rate": 2.1007385348837208e-05,
+      "loss": 0.6483,
+      "step": 5586
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.37947904271544586,
+      "learning_rate": 2.1001600636848575e-05,
+      "loss": 0.6847,
+      "step": 5587
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3494626016926944,
+      "learning_rate": 2.099581584085691e-05,
+      "loss": 0.6358,
+      "step": 5588
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4120958839707049,
+      "learning_rate": 2.0990030961347368e-05,
+      "loss": 0.6988,
+      "step": 5589
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.34940394115754364,
+      "learning_rate": 2.0984245998805125e-05,
+      "loss": 0.623,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.32648814767936607,
+      "learning_rate": 2.0978460953715364e-05,
+      "loss": 0.6216,
+      "step": 5591
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3911812163055521,
+      "learning_rate": 2.0972675826563266e-05,
+      "loss": 0.6857,
+      "step": 5592
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.38027981472107003,
+      "learning_rate": 2.0966890617834014e-05,
+      "loss": 0.7026,
+      "step": 5593
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.37015510719466876,
+      "learning_rate": 2.0961105328012813e-05,
+      "loss": 0.6576,
+      "step": 5594
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3244108873687041,
+      "learning_rate": 2.0955319957584876e-05,
+      "loss": 0.6456,
+      "step": 5595
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3467540172588793,
+      "learning_rate": 2.0949534507035402e-05,
+      "loss": 0.6827,
+      "step": 5596
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3703506147628915,
+      "learning_rate": 2.094374897684962e-05,
+      "loss": 0.6529,
+      "step": 5597
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4342487425510405,
+      "learning_rate": 2.093796336751275e-05,
+      "loss": 0.6936,
+      "step": 5598
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.342022637785697,
+      "learning_rate": 2.0932177679510027e-05,
+      "loss": 0.6478,
+      "step": 5599
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3981225746173784,
+      "learning_rate": 2.0926391913326678e-05,
+      "loss": 0.6548,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4232683435109346,
+      "learning_rate": 2.0920606069447968e-05,
+      "loss": 0.6903,
+      "step": 5601
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3466152454748177,
+      "learning_rate": 2.0914820148359135e-05,
+      "loss": 0.6386,
+      "step": 5602
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0790744787476225,
+      "learning_rate": 2.090903415054544e-05,
+      "loss": 7.1374,
+      "step": 5603
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4273010479911556,
+      "learning_rate": 2.0903248076492152e-05,
+      "loss": 0.6626,
+      "step": 5604
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4237229387801545,
+      "learning_rate": 2.089746192668454e-05,
+      "loss": 0.7165,
+      "step": 5605
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3725533782547384,
+      "learning_rate": 2.089167570160788e-05,
+      "loss": 0.7022,
+      "step": 5606
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4096396186180145,
+      "learning_rate": 2.0885889401747445e-05,
+      "loss": 0.7081,
+      "step": 5607
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.41314912981045476,
+      "learning_rate": 2.0880103027588542e-05,
+      "loss": 0.6747,
+      "step": 5608
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.42534100316776413,
+      "learning_rate": 2.087431657961647e-05,
+      "loss": 0.71,
+      "step": 5609
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3964766460041356,
+      "learning_rate": 2.0868530058316508e-05,
+      "loss": 0.6792,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.37504487452154894,
+      "learning_rate": 2.0862743464173988e-05,
+      "loss": 0.6673,
+      "step": 5611
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.343790435713193,
+      "learning_rate": 2.0856956797674216e-05,
+      "loss": 0.6978,
+      "step": 5612
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3708276328274381,
+      "learning_rate": 2.0851170059302507e-05,
+      "loss": 0.6885,
+      "step": 5613
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.38349238186699947,
+      "learning_rate": 2.0845383249544192e-05,
+      "loss": 0.7156,
+      "step": 5614
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3164182546491926,
+      "learning_rate": 2.0839596368884612e-05,
+      "loss": 0.6211,
+      "step": 5615
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3742889811112162,
+      "learning_rate": 2.08338094178091e-05,
+      "loss": 0.626,
+      "step": 5616
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4237494983076579,
+      "learning_rate": 2.0828022396802984e-05,
+      "loss": 0.6906,
+      "step": 5617
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3422568455571787,
+      "learning_rate": 2.0822235306351645e-05,
+      "loss": 0.6541,
+      "step": 5618
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4135628087306588,
+      "learning_rate": 2.081644814694042e-05,
+      "loss": 0.7213,
+      "step": 5619
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3949932412207581,
+      "learning_rate": 2.0810660919054672e-05,
+      "loss": 0.6649,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4320877095409823,
+      "learning_rate": 2.0804873623179772e-05,
+      "loss": 0.7238,
+      "step": 5621
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4076439983569877,
+      "learning_rate": 2.07990862598011e-05,
+      "loss": 0.7117,
+      "step": 5622
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3933446667285797,
+      "learning_rate": 2.0793298829404024e-05,
+      "loss": 0.6683,
+      "step": 5623
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.36815075893177035,
+      "learning_rate": 2.0787511332473924e-05,
+      "loss": 0.6301,
+      "step": 5624
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4575879836265929,
+      "learning_rate": 2.0781723769496213e-05,
+      "loss": 0.719,
+      "step": 5625
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3443541763387075,
+      "learning_rate": 2.077593614095627e-05,
+      "loss": 0.6973,
+      "step": 5626
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.36337366510951175,
+      "learning_rate": 2.0770148447339496e-05,
+      "loss": 0.623,
+      "step": 5627
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.42897334549050953,
+      "learning_rate": 2.0764360689131303e-05,
+      "loss": 0.7111,
+      "step": 5628
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4046960651110864,
+      "learning_rate": 2.0758572866817107e-05,
+      "loss": 0.6964,
+      "step": 5629
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.49129107232979485,
+      "learning_rate": 2.0752784980882315e-05,
+      "loss": 0.69,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.41512804386914515,
+      "learning_rate": 2.0746997031812353e-05,
+      "loss": 0.7068,
+      "step": 5631
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3908862674094683,
+      "learning_rate": 2.0741209020092664e-05,
+      "loss": 0.6626,
+      "step": 5632
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3649217318024735,
+      "learning_rate": 2.0735420946208663e-05,
+      "loss": 0.6949,
+      "step": 5633
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3950332443244868,
+      "learning_rate": 2.072963281064579e-05,
+      "loss": 0.6752,
+      "step": 5634
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3698094504362425,
+      "learning_rate": 2.0723844613889504e-05,
+      "loss": 0.6396,
+      "step": 5635
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.36080091123216257,
+      "learning_rate": 2.071805635642524e-05,
+      "loss": 0.6642,
+      "step": 5636
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.38372876176658666,
+      "learning_rate": 2.0712268038738456e-05,
+      "loss": 0.6572,
+      "step": 5637
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3657461016643928,
+      "learning_rate": 2.070647966131461e-05,
+      "loss": 0.6121,
+      "step": 5638
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.45392545153263725,
+      "learning_rate": 2.070069122463917e-05,
+      "loss": 0.7112,
+      "step": 5639
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.419947765964471,
+      "learning_rate": 2.069490272919761e-05,
+      "loss": 0.7045,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3171387889592577,
+      "learning_rate": 2.0689114175475384e-05,
+      "loss": 0.6209,
+      "step": 5641
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.44846482985717623,
+      "learning_rate": 2.0683325563957993e-05,
+      "loss": 0.7263,
+      "step": 5642
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.39440073032048006,
+      "learning_rate": 2.0677536895130915e-05,
+      "loss": 0.6578,
+      "step": 5643
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4105381980286797,
+      "learning_rate": 2.067174816947963e-05,
+      "loss": 0.6355,
+      "step": 5644
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.4906932779075674,
+      "learning_rate": 2.066595938748963e-05,
+      "loss": 7.0925,
+      "step": 5645
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3911257237608767,
+      "learning_rate": 2.0660170549646436e-05,
+      "loss": 0.6944,
+      "step": 5646
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.39987879432779627,
+      "learning_rate": 2.0654381656435526e-05,
+      "loss": 0.6864,
+      "step": 5647
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.42862483374241334,
+      "learning_rate": 2.0648592708342418e-05,
+      "loss": 0.6622,
+      "step": 5648
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4005821903433002,
+      "learning_rate": 2.0642803705852623e-05,
+      "loss": 0.6951,
+      "step": 5649
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.39847766626519565,
+      "learning_rate": 2.063701464945166e-05,
+      "loss": 0.7082,
+      "step": 5650
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.34852748255709826,
+      "learning_rate": 2.0631225539625048e-05,
+      "loss": 0.6551,
+      "step": 5651
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.38897568304721414,
+      "learning_rate": 2.0625436376858307e-05,
+      "loss": 0.6917,
+      "step": 5652
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4501930081837213,
+      "learning_rate": 2.061964716163698e-05,
+      "loss": 0.7159,
+      "step": 5653
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3328127303638535,
+      "learning_rate": 2.0613857894446587e-05,
+      "loss": 0.6557,
+      "step": 5654
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.40627880399916105,
+      "learning_rate": 2.0608068575772675e-05,
+      "loss": 0.7033,
+      "step": 5655
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.42543891694202846,
+      "learning_rate": 2.060227920610079e-05,
+      "loss": 0.6559,
+      "step": 5656
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.38377332789378127,
+      "learning_rate": 2.0596489785916472e-05,
+      "loss": 0.6657,
+      "step": 5657
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.36380007971082895,
+      "learning_rate": 2.0590700315705276e-05,
+      "loss": 0.6688,
+      "step": 5658
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.39408822281611666,
+      "learning_rate": 2.0584910795952758e-05,
+      "loss": 0.6419,
+      "step": 5659
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.37460647320682955,
+      "learning_rate": 2.057912122714448e-05,
+      "loss": 0.6846,
+      "step": 5660
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.38856758283646603,
+      "learning_rate": 2.0573331609766e-05,
+      "loss": 0.6849,
+      "step": 5661
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4027075947079001,
+      "learning_rate": 2.0567541944302892e-05,
+      "loss": 0.6806,
+      "step": 5662
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3924852401928964,
+      "learning_rate": 2.0561752231240732e-05,
+      "loss": 0.6934,
+      "step": 5663
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.39025283147698425,
+      "learning_rate": 2.0555962471065082e-05,
+      "loss": 0.684,
+      "step": 5664
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.37117799575338656,
+      "learning_rate": 2.0550172664261535e-05,
+      "loss": 0.6322,
+      "step": 5665
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.36674766029976263,
+      "learning_rate": 2.0544382811315672e-05,
+      "loss": 0.6791,
+      "step": 5666
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.635331442296016,
+      "learning_rate": 2.0538592912713083e-05,
+      "loss": 7.1049,
+      "step": 5667
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9910387701966954,
+      "learning_rate": 2.053280296893935e-05,
+      "loss": 7.0988,
+      "step": 5668
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.40643724973967155,
+      "learning_rate": 2.0527012980480076e-05,
+      "loss": 0.7039,
+      "step": 5669
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7414405888427356,
+      "learning_rate": 2.052122294782087e-05,
+      "loss": 7.0958,
+      "step": 5670
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.38925641887739554,
+      "learning_rate": 2.0515432871447314e-05,
+      "loss": 0.6416,
+      "step": 5671
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3840518224548239,
+      "learning_rate": 2.0509642751845024e-05,
+      "loss": 0.6292,
+      "step": 5672
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3742373027807576,
+      "learning_rate": 2.050385258949962e-05,
+      "loss": 0.6616,
+      "step": 5673
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.45798685513219844,
+      "learning_rate": 2.0498062384896703e-05,
+      "loss": 0.6929,
+      "step": 5674
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4486473522606196,
+      "learning_rate": 2.0492272138521902e-05,
+      "loss": 0.6705,
+      "step": 5675
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.49657568468392266,
+      "learning_rate": 2.048648185086082e-05,
+      "loss": 0.6994,
+      "step": 5676
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.35927815982700495,
+      "learning_rate": 2.0480691522399104e-05,
+      "loss": 0.6458,
+      "step": 5677
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3652875212789878,
+      "learning_rate": 2.0474901153622363e-05,
+      "loss": 0.6448,
+      "step": 5678
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.37138666964755535,
+      "learning_rate": 2.046911074501624e-05,
+      "loss": 0.6593,
+      "step": 5679
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4378377726978975,
+      "learning_rate": 2.0463320297066366e-05,
+      "loss": 0.6999,
+      "step": 5680
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0796087199112407,
+      "learning_rate": 2.045752981025837e-05,
+      "loss": 7.1071,
+      "step": 5681
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.41741645968023033,
+      "learning_rate": 2.045173928507791e-05,
+      "loss": 0.6795,
+      "step": 5682
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.37630703594085807,
+      "learning_rate": 2.0445948722010618e-05,
+      "loss": 0.6464,
+      "step": 5683
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.40592367204481206,
+      "learning_rate": 2.044015812154215e-05,
+      "loss": 0.681,
+      "step": 5684
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.398040873472039,
+      "learning_rate": 2.0434367484158146e-05,
+      "loss": 0.6792,
+      "step": 5685
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4196618054345081,
+      "learning_rate": 2.0428576810344265e-05,
+      "loss": 0.6767,
+      "step": 5686
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3616104881989019,
+      "learning_rate": 2.0422786100586167e-05,
+      "loss": 0.6665,
+      "step": 5687
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3672904339083125,
+      "learning_rate": 2.041699535536951e-05,
+      "loss": 0.6253,
+      "step": 5688
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.384155629787126,
+      "learning_rate": 2.041120457517996e-05,
+      "loss": 0.6692,
+      "step": 5689
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.39026567520183775,
+      "learning_rate": 2.040541376050317e-05,
+      "loss": 0.66,
+      "step": 5690
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3907529629986264,
+      "learning_rate": 2.0399622911824822e-05,
+      "loss": 0.7155,
+      "step": 5691
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3887307201759257,
+      "learning_rate": 2.0393832029630583e-05,
+      "loss": 0.5987,
+      "step": 5692
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3531881673474553,
+      "learning_rate": 2.038804111440613e-05,
+      "loss": 0.6496,
+      "step": 5693
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.40192249674629255,
+      "learning_rate": 2.0382250166637133e-05,
+      "loss": 0.6427,
+      "step": 5694
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.33441337473055466,
+      "learning_rate": 2.037645918680928e-05,
+      "loss": 0.6702,
+      "step": 5695
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3504783018701928,
+      "learning_rate": 2.037066817540825e-05,
+      "loss": 0.6524,
+      "step": 5696
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.35596228461885604,
+      "learning_rate": 2.036487713291972e-05,
+      "loss": 0.6435,
+      "step": 5697
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4413529513918519,
+      "learning_rate": 2.035908605982939e-05,
+      "loss": 0.7284,
+      "step": 5698
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.40490414876169817,
+      "learning_rate": 2.0353294956622948e-05,
+      "loss": 0.6657,
+      "step": 5699
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.35475206742761,
+      "learning_rate": 2.0347503823786083e-05,
+      "loss": 0.6406,
+      "step": 5700
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.38208065205956454,
+      "learning_rate": 2.0341712661804495e-05,
+      "loss": 0.6873,
+      "step": 5701
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4639151265805279,
+      "learning_rate": 2.0335921471163873e-05,
+      "loss": 0.7005,
+      "step": 5702
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4267010941610006,
+      "learning_rate": 2.0330130252349925e-05,
+      "loss": 0.6363,
+      "step": 5703
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3806981928595967,
+      "learning_rate": 2.0324339005848354e-05,
+      "loss": 0.6703,
+      "step": 5704
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3833126605133572,
+      "learning_rate": 2.031854773214486e-05,
+      "loss": 0.6766,
+      "step": 5705
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.36448466865268214,
+      "learning_rate": 2.0312756431725157e-05,
+      "loss": 0.6774,
+      "step": 5706
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.401637798975463,
+      "learning_rate": 2.0306965105074943e-05,
+      "loss": 0.7209,
+      "step": 5707
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3822655349107779,
+      "learning_rate": 2.030117375267994e-05,
+      "loss": 0.6665,
+      "step": 5708
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4227386995416471,
+      "learning_rate": 2.029538237502586e-05,
+      "loss": 0.6605,
+      "step": 5709
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3880327541603075,
+      "learning_rate": 2.028959097259841e-05,
+      "loss": 0.6414,
+      "step": 5710
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.41520865696455733,
+      "learning_rate": 2.0283799545883323e-05,
+      "loss": 0.6514,
+      "step": 5711
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3655955983835748,
+      "learning_rate": 2.0278008095366304e-05,
+      "loss": 0.6691,
+      "step": 5712
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3239882743938296,
+      "learning_rate": 2.0272216621533087e-05,
+      "loss": 0.6355,
+      "step": 5713
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3806063883584348,
+      "learning_rate": 2.0266425124869385e-05,
+      "loss": 0.6729,
+      "step": 5714
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.44737561526693675,
+      "learning_rate": 2.0260633605860933e-05,
+      "loss": 0.6189,
+      "step": 5715
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4362954698677588,
+      "learning_rate": 2.0254842064993458e-05,
+      "loss": 0.6673,
+      "step": 5716
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4216129562135438,
+      "learning_rate": 2.0249050502752685e-05,
+      "loss": 0.6878,
+      "step": 5717
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3815608233710015,
+      "learning_rate": 2.0243258919624343e-05,
+      "loss": 0.6789,
+      "step": 5718
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.44067886698320774,
+      "learning_rate": 2.0237467316094177e-05,
+      "loss": 0.6569,
+      "step": 5719
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.404391733778676,
+      "learning_rate": 2.023167569264791e-05,
+      "loss": 0.6869,
+      "step": 5720
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.42697447342859013,
+      "learning_rate": 2.022588404977128e-05,
+      "loss": 0.7029,
+      "step": 5721
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.39974961044932394,
+      "learning_rate": 2.022009238795003e-05,
+      "loss": 0.6759,
+      "step": 5722
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.33709285657176663,
+      "learning_rate": 2.02143007076699e-05,
+      "loss": 0.653,
+      "step": 5723
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.39019012312019324,
+      "learning_rate": 2.020850900941662e-05,
+      "loss": 0.653,
+      "step": 5724
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.42343083320203934,
+      "learning_rate": 2.020271729367594e-05,
+      "loss": 0.6745,
+      "step": 5725
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.39152095139421267,
+      "learning_rate": 2.019692556093362e-05,
+      "loss": 0.625,
+      "step": 5726
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3615799681825127,
+      "learning_rate": 2.0191133811675384e-05,
+      "loss": 7.0445,
+      "step": 5727
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.37375625348516794,
+      "learning_rate": 2.0185342046386983e-05,
+      "loss": 0.6467,
+      "step": 5728
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.43858860661063076,
+      "learning_rate": 2.017955026555417e-05,
+      "loss": 0.698,
+      "step": 5729
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3608877176792534,
+      "learning_rate": 2.017375846966269e-05,
+      "loss": 0.6541,
+      "step": 5730
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.35544010277645643,
+      "learning_rate": 2.0167966659198303e-05,
+      "loss": 0.6287,
+      "step": 5731
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3420655805475388,
+      "learning_rate": 2.0162174834646748e-05,
+      "loss": 0.6374,
+      "step": 5732
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4344107208431377,
+      "learning_rate": 2.0156382996493794e-05,
+      "loss": 0.6619,
+      "step": 5733
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3650832290557812,
+      "learning_rate": 2.0150591145225186e-05,
+      "loss": 0.6615,
+      "step": 5734
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.40687219974840005,
+      "learning_rate": 2.014479928132667e-05,
+      "loss": 0.6579,
+      "step": 5735
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.36190886270362005,
+      "learning_rate": 2.0139007405284022e-05,
+      "loss": 0.6295,
+      "step": 5736
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3800076640195948,
+      "learning_rate": 2.0133215517582993e-05,
+      "loss": 0.6498,
+      "step": 5737
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.44145204106726005,
+      "learning_rate": 2.012742361870933e-05,
+      "loss": 0.6888,
+      "step": 5738
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.34959088305441316,
+      "learning_rate": 2.0121631709148806e-05,
+      "loss": 0.688,
+      "step": 5739
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.35841875893432573,
+      "learning_rate": 2.0115839789387186e-05,
+      "loss": 0.6542,
+      "step": 5740
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3959492207986419,
+      "learning_rate": 2.011004785991021e-05,
+      "loss": 0.7112,
+      "step": 5741
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3562567436482011,
+      "learning_rate": 2.010425592120366e-05,
+      "loss": 0.6409,
+      "step": 5742
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.994833639529342,
+      "learning_rate": 2.0098463973753294e-05,
+      "loss": 7.1645,
+      "step": 5743
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3942023224649611,
+      "learning_rate": 2.009267201804487e-05,
+      "loss": 0.6601,
+      "step": 5744
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3682837871646261,
+      "learning_rate": 2.0086880054564156e-05,
+      "loss": 0.6774,
+      "step": 5745
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.37313348282447956,
+      "learning_rate": 2.0081088083796912e-05,
+      "loss": 0.6532,
+      "step": 5746
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3775957912687512,
+      "learning_rate": 2.007529610622892e-05,
+      "loss": 0.6917,
+      "step": 5747
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.41723103848233817,
+      "learning_rate": 2.0069504122345923e-05,
+      "loss": 0.7002,
+      "step": 5748
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3943279462374005,
+      "learning_rate": 2.0063712132633703e-05,
+      "loss": 0.6589,
+      "step": 5749
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3563011430799528,
+      "learning_rate": 2.0057920137578027e-05,
+      "loss": 0.6228,
+      "step": 5750
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4846915378545768,
+      "learning_rate": 2.0052128137664657e-05,
+      "loss": 0.7248,
+      "step": 5751
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4283275543814761,
+      "learning_rate": 2.0046336133379354e-05,
+      "loss": 0.685,
+      "step": 5752
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.35447285115694444,
+      "learning_rate": 2.0040544125207905e-05,
+      "loss": 0.6524,
+      "step": 5753
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.39759739665261945,
+      "learning_rate": 2.0034752113636072e-05,
+      "loss": 0.6645,
+      "step": 5754
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.40658892484147513,
+      "learning_rate": 2.0028960099149616e-05,
+      "loss": 0.6515,
+      "step": 5755
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.38112980316245876,
+      "learning_rate": 2.002316808223431e-05,
+      "loss": 0.666,
+      "step": 5756
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3724199789940745,
+      "learning_rate": 2.001737606337593e-05,
+      "loss": 0.6947,
+      "step": 5757
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4222873837448207,
+      "learning_rate": 2.001158404306024e-05,
+      "loss": 0.6609,
+      "step": 5758
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3621516368680209,
+      "learning_rate": 2.0005792021773e-05,
+      "loss": 0.661,
+      "step": 5759
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33806766751439915,
+      "learning_rate": 2e-05,
+      "loss": 0.6189,
+      "step": 5760
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.35903861582323765,
+      "learning_rate": 1.9994207978227002e-05,
+      "loss": 0.6643,
+      "step": 5761
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.35358065162377994,
+      "learning_rate": 1.998841595693977e-05,
+      "loss": 0.6803,
+      "step": 5762
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3560360875479464,
+      "learning_rate": 1.998262393662408e-05,
+      "loss": 0.6576,
+      "step": 5763
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.35614513580666485,
+      "learning_rate": 1.997683191776569e-05,
+      "loss": 0.6297,
+      "step": 5764
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3579428038051199,
+      "learning_rate": 1.997103990085039e-05,
+      "loss": 0.6471,
+      "step": 5765
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3620422762271326,
+      "learning_rate": 1.996524788636393e-05,
+      "loss": 0.6783,
+      "step": 5766
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.36801732476063903,
+      "learning_rate": 1.99594558747921e-05,
+      "loss": 0.6711,
+      "step": 5767
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.5375848029442505,
+      "learning_rate": 1.995366386662065e-05,
+      "loss": 7.1512,
+      "step": 5768
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4079845876512994,
+      "learning_rate": 1.9947871862335353e-05,
+      "loss": 0.6808,
+      "step": 5769
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3864926790729311,
+      "learning_rate": 1.9942079862421987e-05,
+      "loss": 0.6408,
+      "step": 5770
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40860046194303895,
+      "learning_rate": 1.99362878673663e-05,
+      "loss": 0.6989,
+      "step": 5771
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3924905707717434,
+      "learning_rate": 1.9930495877654084e-05,
+      "loss": 0.6223,
+      "step": 5772
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3950131869476831,
+      "learning_rate": 1.9924703893771088e-05,
+      "loss": 0.6549,
+      "step": 5773
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40045546494893786,
+      "learning_rate": 1.991891191620309e-05,
+      "loss": 0.7204,
+      "step": 5774
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.42513511954991845,
+      "learning_rate": 1.991311994543585e-05,
+      "loss": 0.7054,
+      "step": 5775
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40765510204857147,
+      "learning_rate": 1.9907327981955137e-05,
+      "loss": 0.6806,
+      "step": 5776
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4025198476698416,
+      "learning_rate": 1.9901536026246716e-05,
+      "loss": 0.6872,
+      "step": 5777
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.46085543333844725,
+      "learning_rate": 1.989574407879634e-05,
+      "loss": 0.6789,
+      "step": 5778
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.34043922029844437,
+      "learning_rate": 1.9889952140089786e-05,
+      "loss": 0.653,
+      "step": 5779
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3559292784412687,
+      "learning_rate": 1.988416021061282e-05,
+      "loss": 0.6756,
+      "step": 5780
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.37919711601144906,
+      "learning_rate": 1.9878368290851197e-05,
+      "loss": 0.7078,
+      "step": 5781
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.35054483057514285,
+      "learning_rate": 1.9872576381290673e-05,
+      "loss": 0.6627,
+      "step": 5782
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.35669216137079873,
+      "learning_rate": 1.9866784482417018e-05,
+      "loss": 0.6522,
+      "step": 5783
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4239936146729435,
+      "learning_rate": 1.9860992594715988e-05,
+      "loss": 0.6786,
+      "step": 5784
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.44043776816544516,
+      "learning_rate": 1.985520071867333e-05,
+      "loss": 0.6931,
+      "step": 5785
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4520647820159047,
+      "learning_rate": 1.984940885477482e-05,
+      "loss": 0.6498,
+      "step": 5786
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40870832012960373,
+      "learning_rate": 1.9843617003506212e-05,
+      "loss": 0.6358,
+      "step": 5787
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.42193553051305444,
+      "learning_rate": 1.9837825165353256e-05,
+      "loss": 0.675,
+      "step": 5788
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3652255503261703,
+      "learning_rate": 1.9832033340801703e-05,
+      "loss": 0.6618,
+      "step": 5789
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3760451624595482,
+      "learning_rate": 1.9826241530337315e-05,
+      "loss": 0.6383,
+      "step": 5790
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.38021014746497156,
+      "learning_rate": 1.9820449734445844e-05,
+      "loss": 0.6809,
+      "step": 5791
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.39435377431909474,
+      "learning_rate": 1.9814657953613024e-05,
+      "loss": 0.6678,
+      "step": 5792
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3897504304975721,
+      "learning_rate": 1.9808866188324623e-05,
+      "loss": 0.6682,
+      "step": 5793
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4068988196052713,
+      "learning_rate": 1.9803074439066388e-05,
+      "loss": 0.6421,
+      "step": 5794
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.37122363299503086,
+      "learning_rate": 1.9797282706324063e-05,
+      "loss": 0.6704,
+      "step": 5795
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3945731225329753,
+      "learning_rate": 1.9791490990583386e-05,
+      "loss": 0.6975,
+      "step": 5796
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3882088433593043,
+      "learning_rate": 1.9785699292330115e-05,
+      "loss": 0.6401,
+      "step": 5797
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.43289112019364956,
+      "learning_rate": 1.977990761204998e-05,
+      "loss": 0.6907,
+      "step": 5798
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3958590285095311,
+      "learning_rate": 1.9774115950228725e-05,
+      "loss": 0.6374,
+      "step": 5799
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 8.403021192913652,
+      "learning_rate": 1.9768324307352096e-05,
+      "loss": 7.1634,
+      "step": 5800
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3904634548831611,
+      "learning_rate": 1.976253268390583e-05,
+      "loss": 0.6738,
+      "step": 5801
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4023852775616037,
+      "learning_rate": 1.9756741080375663e-05,
+      "loss": 0.6479,
+      "step": 5802
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3959426555934778,
+      "learning_rate": 1.9750949497247322e-05,
+      "loss": 0.6987,
+      "step": 5803
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.38139244773085834,
+      "learning_rate": 1.9745157935006552e-05,
+      "loss": 0.7215,
+      "step": 5804
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3879189661907089,
+      "learning_rate": 1.9739366394139067e-05,
+      "loss": 0.661,
+      "step": 5805
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4181294790634482,
+      "learning_rate": 1.9733574875130618e-05,
+      "loss": 0.6897,
+      "step": 5806
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.36485523396878117,
+      "learning_rate": 1.972778337846692e-05,
+      "loss": 0.6402,
+      "step": 5807
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3914180667679309,
+      "learning_rate": 1.9721991904633702e-05,
+      "loss": 0.6367,
+      "step": 5808
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4301100755068248,
+      "learning_rate": 1.9716200454116683e-05,
+      "loss": 0.6976,
+      "step": 5809
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3650174455109415,
+      "learning_rate": 1.9710409027401592e-05,
+      "loss": 0.6828,
+      "step": 5810
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4079311622007888,
+      "learning_rate": 1.970461762497415e-05,
+      "loss": 0.6878,
+      "step": 5811
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3605862846711083,
+      "learning_rate": 1.969882624732006e-05,
+      "loss": 0.6614,
+      "step": 5812
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40852850704011917,
+      "learning_rate": 1.9693034894925056e-05,
+      "loss": 0.6886,
+      "step": 5813
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3821363322262271,
+      "learning_rate": 1.968724356827485e-05,
+      "loss": 0.6471,
+      "step": 5814
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40224739676451693,
+      "learning_rate": 1.9681452267855142e-05,
+      "loss": 0.6784,
+      "step": 5815
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.39254279098211053,
+      "learning_rate": 1.967566099415165e-05,
+      "loss": 0.6621,
+      "step": 5816
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3914575320870067,
+      "learning_rate": 1.9669869747650078e-05,
+      "loss": 0.6656,
+      "step": 5817
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3376967971451896,
+      "learning_rate": 1.9664078528836134e-05,
+      "loss": 0.6677,
+      "step": 5818
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.37425045090413084,
+      "learning_rate": 1.965828733819551e-05,
+      "loss": 0.6899,
+      "step": 5819
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.37734911982713903,
+      "learning_rate": 1.9652496176213916e-05,
+      "loss": 0.6809,
+      "step": 5820
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.43532354286997443,
+      "learning_rate": 1.9646705043377055e-05,
+      "loss": 0.6716,
+      "step": 5821
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33573852019607064,
+      "learning_rate": 1.9640913940170612e-05,
+      "loss": 0.6726,
+      "step": 5822
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4302739672369836,
+      "learning_rate": 1.9635122867080284e-05,
+      "loss": 0.6912,
+      "step": 5823
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.39390653736970777,
+      "learning_rate": 1.962933182459176e-05,
+      "loss": 0.6835,
+      "step": 5824
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3765704795485958,
+      "learning_rate": 1.962354081319073e-05,
+      "loss": 0.7055,
+      "step": 5825
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.34772480984415627,
+      "learning_rate": 1.961774983336287e-05,
+      "loss": 0.6578,
+      "step": 5826
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4293829835868863,
+      "learning_rate": 1.9611958885593875e-05,
+      "loss": 0.7057,
+      "step": 5827
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.394892543400124,
+      "learning_rate": 1.960616797036942e-05,
+      "loss": 0.7048,
+      "step": 5828
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3695182989172334,
+      "learning_rate": 1.9600377088175185e-05,
+      "loss": 0.6831,
+      "step": 5829
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3807999631794022,
+      "learning_rate": 1.9594586239496837e-05,
+      "loss": 0.6539,
+      "step": 5830
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3348289300902839,
+      "learning_rate": 1.958879542482005e-05,
+      "loss": 0.6391,
+      "step": 5831
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3508481435447869,
+      "learning_rate": 1.9583004644630497e-05,
+      "loss": 0.663,
+      "step": 5832
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.36807589426637144,
+      "learning_rate": 1.9577213899413836e-05,
+      "loss": 0.6965,
+      "step": 5833
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.388868830372282,
+      "learning_rate": 1.9571423189655735e-05,
+      "loss": 0.6859,
+      "step": 5834
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.34061104029814954,
+      "learning_rate": 1.956563251584186e-05,
+      "loss": 0.6636,
+      "step": 5835
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3605235247009034,
+      "learning_rate": 1.9559841878457857e-05,
+      "loss": 0.6429,
+      "step": 5836
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3612688492376653,
+      "learning_rate": 1.955405127798939e-05,
+      "loss": 0.6428,
+      "step": 5837
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4011423318790025,
+      "learning_rate": 1.9548260714922097e-05,
+      "loss": 0.6518,
+      "step": 5838
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.43243668726178613,
+      "learning_rate": 1.9542470189741634e-05,
+      "loss": 0.7408,
+      "step": 5839
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4227759844168636,
+      "learning_rate": 1.9536679702933637e-05,
+      "loss": 0.6985,
+      "step": 5840
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.35457077686816624,
+      "learning_rate": 1.9530889254983764e-05,
+      "loss": 0.659,
+      "step": 5841
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3784382293785197,
+      "learning_rate": 1.952509884637764e-05,
+      "loss": 0.6781,
+      "step": 5842
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.39506034697189035,
+      "learning_rate": 1.9519308477600903e-05,
+      "loss": 0.7017,
+      "step": 5843
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3646058757240521,
+      "learning_rate": 1.9513518149139183e-05,
+      "loss": 0.6426,
+      "step": 5844
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40780203572316726,
+      "learning_rate": 1.950772786147811e-05,
+      "loss": 0.7061,
+      "step": 5845
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3439853023160825,
+      "learning_rate": 1.95019376151033e-05,
+      "loss": 0.6554,
+      "step": 5846
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.39952835070598963,
+      "learning_rate": 1.949614741050038e-05,
+      "loss": 0.6554,
+      "step": 5847
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.37815873104572206,
+      "learning_rate": 1.9490357248154976e-05,
+      "loss": 0.6671,
+      "step": 5848
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3485479858365938,
+      "learning_rate": 1.9484567128552692e-05,
+      "loss": 0.6416,
+      "step": 5849
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3518444727096962,
+      "learning_rate": 1.9478777052179138e-05,
+      "loss": 0.6381,
+      "step": 5850
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3750872738364074,
+      "learning_rate": 1.9472987019519928e-05,
+      "loss": 0.6611,
+      "step": 5851
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3624596662534853,
+      "learning_rate": 1.9467197031060658e-05,
+      "loss": 0.6698,
+      "step": 5852
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3726574705980596,
+      "learning_rate": 1.9461407087286927e-05,
+      "loss": 0.6795,
+      "step": 5853
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3849020992079385,
+      "learning_rate": 1.9455617188684328e-05,
+      "loss": 0.6835,
+      "step": 5854
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3667417968501865,
+      "learning_rate": 1.9449827335738468e-05,
+      "loss": 0.6729,
+      "step": 5855
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3788300967099965,
+      "learning_rate": 1.944403752893492e-05,
+      "loss": 0.6735,
+      "step": 5856
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4085282214399018,
+      "learning_rate": 1.9438247768759275e-05,
+      "loss": 0.6829,
+      "step": 5857
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.39976407394782515,
+      "learning_rate": 1.9432458055697115e-05,
+      "loss": 0.6618,
+      "step": 5858
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6789159883513696,
+      "learning_rate": 1.942666839023401e-05,
+      "loss": 0.7169,
+      "step": 5859
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.36446309431447355,
+      "learning_rate": 1.942087877285553e-05,
+      "loss": 0.6858,
+      "step": 5860
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3667013421490002,
+      "learning_rate": 1.9415089204047245e-05,
+      "loss": 0.6714,
+      "step": 5861
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3882773540552353,
+      "learning_rate": 1.9409299684294728e-05,
+      "loss": 0.7013,
+      "step": 5862
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.41282778503638,
+      "learning_rate": 1.9403510214083538e-05,
+      "loss": 0.6442,
+      "step": 5863
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.39421646716317044,
+      "learning_rate": 1.9397720793899218e-05,
+      "loss": 0.7093,
+      "step": 5864
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3656260450036199,
+      "learning_rate": 1.9391931424227332e-05,
+      "loss": 0.6548,
+      "step": 5865
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.43059134957145134,
+      "learning_rate": 1.938614210555342e-05,
+      "loss": 0.7078,
+      "step": 5866
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3666299849092604,
+      "learning_rate": 1.9380352838363033e-05,
+      "loss": 0.699,
+      "step": 5867
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3916252905606723,
+      "learning_rate": 1.9374563623141693e-05,
+      "loss": 0.7034,
+      "step": 5868
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.36915786769582215,
+      "learning_rate": 1.936877446037496e-05,
+      "loss": 0.6676,
+      "step": 5869
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.34278028221376267,
+      "learning_rate": 1.9362985350548342e-05,
+      "loss": 0.6554,
+      "step": 5870
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.38355685515104077,
+      "learning_rate": 1.935719629414738e-05,
+      "loss": 0.6959,
+      "step": 5871
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37018653261487827,
+      "learning_rate": 1.935140729165759e-05,
+      "loss": 0.6508,
+      "step": 5872
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.38095719352346774,
+      "learning_rate": 1.934561834356448e-05,
+      "loss": 0.6201,
+      "step": 5873
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3656893919256307,
+      "learning_rate": 1.9339829450353574e-05,
+      "loss": 0.6449,
+      "step": 5874
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37041896745702524,
+      "learning_rate": 1.933404061251037e-05,
+      "loss": 0.6503,
+      "step": 5875
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37776870558842296,
+      "learning_rate": 1.932825183052038e-05,
+      "loss": 0.6675,
+      "step": 5876
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.31756150931656346,
+      "learning_rate": 1.9322463104869095e-05,
+      "loss": 0.6146,
+      "step": 5877
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3600405277414511,
+      "learning_rate": 1.9316674436042014e-05,
+      "loss": 0.685,
+      "step": 5878
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.38713369099296274,
+      "learning_rate": 1.9310885824524623e-05,
+      "loss": 0.6965,
+      "step": 5879
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4164555361161508,
+      "learning_rate": 1.93050972708024e-05,
+      "loss": 0.6632,
+      "step": 5880
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3947621555860475,
+      "learning_rate": 1.929930877536084e-05,
+      "loss": 0.7072,
+      "step": 5881
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3728285138679661,
+      "learning_rate": 1.929352033868539e-05,
+      "loss": 0.6461,
+      "step": 5882
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3405208490062477,
+      "learning_rate": 1.928773196126155e-05,
+      "loss": 0.6943,
+      "step": 5883
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.354763672538472,
+      "learning_rate": 1.9281943643574766e-05,
+      "loss": 0.622,
+      "step": 5884
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3750081834650353,
+      "learning_rate": 1.9276155386110503e-05,
+      "loss": 0.6696,
+      "step": 5885
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3927397896190539,
+      "learning_rate": 1.9270367189354216e-05,
+      "loss": 0.6542,
+      "step": 5886
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.39709517135671063,
+      "learning_rate": 1.9264579053791347e-05,
+      "loss": 0.6702,
+      "step": 5887
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37789774495178297,
+      "learning_rate": 1.925879097990734e-05,
+      "loss": 0.6721,
+      "step": 5888
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.389123012920186,
+      "learning_rate": 1.9253002968187647e-05,
+      "loss": 0.6557,
+      "step": 5889
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.39648783136057053,
+      "learning_rate": 1.924721501911769e-05,
+      "loss": 0.6775,
+      "step": 5890
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3946560355003891,
+      "learning_rate": 1.92414271331829e-05,
+      "loss": 0.6887,
+      "step": 5891
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.45943159273054135,
+      "learning_rate": 1.92356393108687e-05,
+      "loss": 0.7309,
+      "step": 5892
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3758210267595532,
+      "learning_rate": 1.9229851552660507e-05,
+      "loss": 0.6258,
+      "step": 5893
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.35362764330079693,
+      "learning_rate": 1.922406385904374e-05,
+      "loss": 0.6731,
+      "step": 5894
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3581800297776818,
+      "learning_rate": 1.921827623050379e-05,
+      "loss": 0.6242,
+      "step": 5895
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.34753128711619075,
+      "learning_rate": 1.9212488667526076e-05,
+      "loss": 0.6657,
+      "step": 5896
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.39822148290166054,
+      "learning_rate": 1.9206701170595983e-05,
+      "loss": 0.7052,
+      "step": 5897
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3461805610769296,
+      "learning_rate": 1.9200913740198908e-05,
+      "loss": 0.6534,
+      "step": 5898
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.35564077819848644,
+      "learning_rate": 1.919512637682023e-05,
+      "loss": 0.6724,
+      "step": 5899
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3871134137759464,
+      "learning_rate": 1.9189339080945334e-05,
+      "loss": 0.6395,
+      "step": 5900
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3558918311867568,
+      "learning_rate": 1.9183551853059586e-05,
+      "loss": 0.6603,
+      "step": 5901
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.38806130903795527,
+      "learning_rate": 1.9177764693648355e-05,
+      "loss": 0.6169,
+      "step": 5902
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.39691523996105543,
+      "learning_rate": 1.9171977603197013e-05,
+      "loss": 0.63,
+      "step": 5903
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3968785487436537,
+      "learning_rate": 1.9166190582190907e-05,
+      "loss": 0.6634,
+      "step": 5904
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.39729741480877745,
+      "learning_rate": 1.916040363111539e-05,
+      "loss": 0.69,
+      "step": 5905
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.36336425538070144,
+      "learning_rate": 1.915461675045581e-05,
+      "loss": 0.6883,
+      "step": 5906
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3929138685347205,
+      "learning_rate": 1.91488299406975e-05,
+      "loss": 0.707,
+      "step": 5907
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.41003392462011096,
+      "learning_rate": 1.9143043202325794e-05,
+      "loss": 0.7104,
+      "step": 5908
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3748154613801716,
+      "learning_rate": 1.9137256535826016e-05,
+      "loss": 0.6153,
+      "step": 5909
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.34671187685098204,
+      "learning_rate": 1.9131469941683495e-05,
+      "loss": 0.645,
+      "step": 5910
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3158743337585783,
+      "learning_rate": 1.912568342038354e-05,
+      "loss": 0.6318,
+      "step": 5911
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3708569968028817,
+      "learning_rate": 1.911989697241146e-05,
+      "loss": 0.6756,
+      "step": 5912
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4143229608037123,
+      "learning_rate": 1.9114110598252562e-05,
+      "loss": 0.6873,
+      "step": 5913
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.38907322748869255,
+      "learning_rate": 1.9108324298392133e-05,
+      "loss": 0.6707,
+      "step": 5914
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3594838657013841,
+      "learning_rate": 1.910253807331547e-05,
+      "loss": 0.6382,
+      "step": 5915
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 10.511319763057223,
+      "learning_rate": 1.909675192350785e-05,
+      "loss": 6.973,
+      "step": 5916
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.38857001550040526,
+      "learning_rate": 1.9090965849454562e-05,
+      "loss": 0.7141,
+      "step": 5917
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4255725352879854,
+      "learning_rate": 1.9085179851640868e-05,
+      "loss": 0.6966,
+      "step": 5918
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3926886747433605,
+      "learning_rate": 1.907939393055204e-05,
+      "loss": 0.6931,
+      "step": 5919
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3989418081041926,
+      "learning_rate": 1.907360808667333e-05,
+      "loss": 0.6619,
+      "step": 5920
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3814628124920036,
+      "learning_rate": 1.9067822320489987e-05,
+      "loss": 0.6826,
+      "step": 5921
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4341823025748612,
+      "learning_rate": 1.906203663248726e-05,
+      "loss": 0.7111,
+      "step": 5922
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.39448092055964895,
+      "learning_rate": 1.9056251023150385e-05,
+      "loss": 0.6456,
+      "step": 5923
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.35497247417877453,
+      "learning_rate": 1.9050465492964604e-05,
+      "loss": 0.6482,
+      "step": 5924
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3700631144360202,
+      "learning_rate": 1.904468004241513e-05,
+      "loss": 0.6527,
+      "step": 5925
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4167131006848698,
+      "learning_rate": 1.903889467198719e-05,
+      "loss": 0.6621,
+      "step": 5926
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3367536517988797,
+      "learning_rate": 1.903310938216599e-05,
+      "loss": 0.6158,
+      "step": 5927
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3597327704467679,
+      "learning_rate": 1.9027324173436744e-05,
+      "loss": 0.6553,
+      "step": 5928
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.39913040100266095,
+      "learning_rate": 1.9021539046284646e-05,
+      "loss": 0.6539,
+      "step": 5929
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.40721635380269006,
+      "learning_rate": 1.9015754001194875e-05,
+      "loss": 0.6728,
+      "step": 5930
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.456499184628863,
+      "learning_rate": 1.9009969038652635e-05,
+      "loss": 0.694,
+      "step": 5931
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4142081487094415,
+      "learning_rate": 1.9004184159143094e-05,
+      "loss": 0.6549,
+      "step": 5932
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4307753405673029,
+      "learning_rate": 1.8998399363151432e-05,
+      "loss": 0.6786,
+      "step": 5933
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3959153801832691,
+      "learning_rate": 1.89926146511628e-05,
+      "loss": 0.667,
+      "step": 5934
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.406533037494611,
+      "learning_rate": 1.898683002366236e-05,
+      "loss": 0.6497,
+      "step": 5935
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.365578117134459,
+      "learning_rate": 1.8981045481135266e-05,
+      "loss": 0.657,
+      "step": 5936
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4970018804459071,
+      "learning_rate": 1.897526102406665e-05,
+      "loss": 0.6772,
+      "step": 5937
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3594297390227337,
+      "learning_rate": 1.8969476652941656e-05,
+      "loss": 0.6645,
+      "step": 5938
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.38770334498708264,
+      "learning_rate": 1.8963692368245413e-05,
+      "loss": 0.656,
+      "step": 5939
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.35950341470166375,
+      "learning_rate": 1.8957908170463045e-05,
+      "loss": 0.647,
+      "step": 5940
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3480849219876529,
+      "learning_rate": 1.895212406007965e-05,
+      "loss": 0.633,
+      "step": 5941
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.38788778222503273,
+      "learning_rate": 1.8946340037580353e-05,
+      "loss": 0.6624,
+      "step": 5942
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.45436787249214644,
+      "learning_rate": 1.8940556103450244e-05,
+      "loss": 0.6908,
+      "step": 5943
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.42311264640025226,
+      "learning_rate": 1.8934772258174406e-05,
+      "loss": 0.6994,
+      "step": 5944
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.34324712415588926,
+      "learning_rate": 1.892898850223794e-05,
+      "loss": 0.6257,
+      "step": 5945
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.38296428896880286,
+      "learning_rate": 1.8923204836125914e-05,
+      "loss": 0.6633,
+      "step": 5946
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.39620999112482697,
+      "learning_rate": 1.89174212603234e-05,
+      "loss": 0.6532,
+      "step": 5947
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37787816627462967,
+      "learning_rate": 1.891163777531545e-05,
+      "loss": 0.6984,
+      "step": 5948
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.43335136057764506,
+      "learning_rate": 1.8905854381587132e-05,
+      "loss": 0.6893,
+      "step": 5949
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.39046350303060606,
+      "learning_rate": 1.8900071079623482e-05,
+      "loss": 0.6828,
+      "step": 5950
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4167896519987439,
+      "learning_rate": 1.8894287869909543e-05,
+      "loss": 0.643,
+      "step": 5951
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.34922645447379325,
+      "learning_rate": 1.8888504752930343e-05,
+      "loss": 0.6531,
+      "step": 5952
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4005995895879954,
+      "learning_rate": 1.8882721729170913e-05,
+      "loss": 0.6902,
+      "step": 5953
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3705497546182399,
+      "learning_rate": 1.8876938799116256e-05,
+      "loss": 0.6525,
+      "step": 5954
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4150032071371418,
+      "learning_rate": 1.8871155963251394e-05,
+      "loss": 0.6814,
+      "step": 5955
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.35171550766833887,
+      "learning_rate": 1.8865373222061315e-05,
+      "loss": 0.6391,
+      "step": 5956
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3369890595346905,
+      "learning_rate": 1.8859590576031008e-05,
+      "loss": 0.6559,
+      "step": 5957
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.40331681444983564,
+      "learning_rate": 1.8853808025645458e-05,
+      "loss": 0.6396,
+      "step": 5958
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.41803098875811967,
+      "learning_rate": 1.8848025571389654e-05,
+      "loss": 0.7,
+      "step": 5959
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37602347976207606,
+      "learning_rate": 1.8842243213748556e-05,
+      "loss": 0.6324,
+      "step": 5960
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37512272289645043,
+      "learning_rate": 1.8836460953207117e-05,
+      "loss": 0.6435,
+      "step": 5961
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.40693046498189706,
+      "learning_rate": 1.8830678790250293e-05,
+      "loss": 0.6545,
+      "step": 5962
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3543535050950025,
+      "learning_rate": 1.882489672536303e-05,
+      "loss": 0.6477,
+      "step": 5963
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.35351906740549494,
+      "learning_rate": 1.8819114759030255e-05,
+      "loss": 0.6553,
+      "step": 5964
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4029211674053556,
+      "learning_rate": 1.8813332891736896e-05,
+      "loss": 0.6738,
+      "step": 5965
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3764816049272621,
+      "learning_rate": 1.8807551123967878e-05,
+      "loss": 0.6643,
+      "step": 5966
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3763601637756536,
+      "learning_rate": 1.8801769456208112e-05,
+      "loss": 0.6652,
+      "step": 5967
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3629360055050509,
+      "learning_rate": 1.879598788894249e-05,
+      "loss": 0.6641,
+      "step": 5968
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3665663229374289,
+      "learning_rate": 1.879020642265591e-05,
+      "loss": 0.6294,
+      "step": 5969
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.36374472941703123,
+      "learning_rate": 1.878442505783326e-05,
+      "loss": 0.6537,
+      "step": 5970
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37447026497288427,
+      "learning_rate": 1.8778643794959406e-05,
+      "loss": 0.6384,
+      "step": 5971
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4020396672634978,
+      "learning_rate": 1.877286263451922e-05,
+      "loss": 0.6537,
+      "step": 5972
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.42639149386325187,
+      "learning_rate": 1.8767081576997566e-05,
+      "loss": 0.7005,
+      "step": 5973
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3671426133064761,
+      "learning_rate": 1.8761300622879297e-05,
+      "loss": 0.6819,
+      "step": 5974
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.38644397290455257,
+      "learning_rate": 1.8755519772649244e-05,
+      "loss": 0.6881,
+      "step": 5975
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37326227297102826,
+      "learning_rate": 1.8749739026792246e-05,
+      "loss": 0.7101,
+      "step": 5976
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3413628016165125,
+      "learning_rate": 1.874395838579313e-05,
+      "loss": 0.6767,
+      "step": 5977
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37832198735259914,
+      "learning_rate": 1.8738177850136693e-05,
+      "loss": 0.6834,
+      "step": 5978
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.35125855377725324,
+      "learning_rate": 1.8732397420307763e-05,
+      "loss": 0.6642,
+      "step": 5979
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.43569422182124823,
+      "learning_rate": 1.8726617096791137e-05,
+      "loss": 0.6478,
+      "step": 5980
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.414279588875255,
+      "learning_rate": 1.8720836880071593e-05,
+      "loss": 0.648,
+      "step": 5981
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3372819231065177,
+      "learning_rate": 1.8715056770633916e-05,
+      "loss": 0.6821,
+      "step": 5982
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.36427352520115824,
+      "learning_rate": 1.8709276768962878e-05,
+      "loss": 0.6111,
+      "step": 5983
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.386419116915848,
+      "learning_rate": 1.870349687554323e-05,
+      "loss": 0.6625,
+      "step": 5984
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3914845665183141,
+      "learning_rate": 1.8697717090859736e-05,
+      "loss": 0.6585,
+      "step": 5985
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4436602764434437,
+      "learning_rate": 1.869193741539714e-05,
+      "loss": 0.67,
+      "step": 5986
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.41129746448526877,
+      "learning_rate": 1.8686157849640175e-05,
+      "loss": 0.6757,
+      "step": 5987
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.40879134827286684,
+      "learning_rate": 1.868037839407356e-05,
+      "loss": 0.6655,
+      "step": 5988
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3255170423807809,
+      "learning_rate": 1.867459904918202e-05,
+      "loss": 0.6269,
+      "step": 5989
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.42014684943702196,
+      "learning_rate": 1.8668819815450255e-05,
+      "loss": 0.6573,
+      "step": 5990
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.34722721304765974,
+      "learning_rate": 1.8663040693362963e-05,
+      "loss": 0.6486,
+      "step": 5991
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.39557913876831013,
+      "learning_rate": 1.865726168340483e-05,
+      "loss": 0.6682,
+      "step": 5992
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38044625744489907,
+      "learning_rate": 1.865148278606054e-05,
+      "loss": 0.673,
+      "step": 5993
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3806079830376094,
+      "learning_rate": 1.8645704001814766e-05,
+      "loss": 0.6743,
+      "step": 5994
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.44060045975450846,
+      "learning_rate": 1.8639925331152158e-05,
+      "loss": 0.6778,
+      "step": 5995
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.42997835566885834,
+      "learning_rate": 1.863414677455737e-05,
+      "loss": 0.6876,
+      "step": 5996
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3945704000053389,
+      "learning_rate": 1.8628368332515047e-05,
+      "loss": 0.6345,
+      "step": 5997
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3805615803554362,
+      "learning_rate": 1.862259000550981e-05,
+      "loss": 0.6917,
+      "step": 5998
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.37725812042715484,
+      "learning_rate": 1.861681179402628e-05,
+      "loss": 0.6968,
+      "step": 5999
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.41370487765581576,
+      "learning_rate": 1.8611033698549083e-05,
+      "loss": 0.6309,
+      "step": 6000
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3875373264485141,
+      "learning_rate": 1.8605255719562813e-05,
+      "loss": 0.6499,
+      "step": 6001
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.39986226726354096,
+      "learning_rate": 1.8599477857552057e-05,
+      "loss": 0.6484,
+      "step": 6002
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3766046287755045,
+      "learning_rate": 1.8593700113001406e-05,
+      "loss": 0.6949,
+      "step": 6003
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.41416216690141505,
+      "learning_rate": 1.8587922486395428e-05,
+      "loss": 0.6986,
+      "step": 6004
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4365860748321308,
+      "learning_rate": 1.8582144978218683e-05,
+      "loss": 0.6718,
+      "step": 6005
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.39170924872685886,
+      "learning_rate": 1.857636758895572e-05,
+      "loss": 0.6466,
+      "step": 6006
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.32562479087564145,
+      "learning_rate": 1.8570590319091095e-05,
+      "loss": 0.6661,
+      "step": 6007
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38528158426904924,
+      "learning_rate": 1.8564813169109336e-05,
+      "loss": 0.6785,
+      "step": 6008
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3761767351143225,
+      "learning_rate": 1.855903613949496e-05,
+      "loss": 0.6901,
+      "step": 6009
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3489318411767773,
+      "learning_rate": 1.8553259230732485e-05,
+      "loss": 0.6294,
+      "step": 6010
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.40296897252365976,
+      "learning_rate": 1.8547482443306408e-05,
+      "loss": 0.653,
+      "step": 6011
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3813878880651801,
+      "learning_rate": 1.854170577770123e-05,
+      "loss": 0.6723,
+      "step": 6012
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3827751349510595,
+      "learning_rate": 1.8535929234401412e-05,
+      "loss": 0.6492,
+      "step": 6013
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.41014425618186306,
+      "learning_rate": 1.8530152813891455e-05,
+      "loss": 0.6629,
+      "step": 6014
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.40967820340572947,
+      "learning_rate": 1.8524376516655803e-05,
+      "loss": 0.6966,
+      "step": 6015
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38413814674940644,
+      "learning_rate": 1.851860034317891e-05,
+      "loss": 0.6551,
+      "step": 6016
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.37956806783774355,
+      "learning_rate": 1.8512824293945223e-05,
+      "loss": 0.6333,
+      "step": 6017
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.36544447581306977,
+      "learning_rate": 1.850704836943916e-05,
+      "loss": 0.6717,
+      "step": 6018
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.35122731502923904,
+      "learning_rate": 1.850127257014515e-05,
+      "loss": 0.6579,
+      "step": 6019
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3875920103594927,
+      "learning_rate": 1.8495496896547595e-05,
+      "loss": 0.6724,
+      "step": 6020
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3797112656501925,
+      "learning_rate": 1.8489721349130902e-05,
+      "loss": 0.6573,
+      "step": 6021
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.36946790871053237,
+      "learning_rate": 1.848394592837945e-05,
+      "loss": 0.6276,
+      "step": 6022
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38659802109943897,
+      "learning_rate": 1.847817063477763e-05,
+      "loss": 0.6559,
+      "step": 6023
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3664195587259628,
+      "learning_rate": 1.84723954688098e-05,
+      "loss": 0.6832,
+      "step": 6024
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3898227832674989,
+      "learning_rate": 1.8466620430960316e-05,
+      "loss": 0.677,
+      "step": 6025
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4092141050261573,
+      "learning_rate": 1.8460845521713523e-05,
+      "loss": 0.6562,
+      "step": 6026
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3716100148104808,
+      "learning_rate": 1.8455070741553754e-05,
+      "loss": 0.6733,
+      "step": 6027
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3560396222195697,
+      "learning_rate": 1.8449296090965344e-05,
+      "loss": 0.6081,
+      "step": 6028
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.36999332229560644,
+      "learning_rate": 1.844352157043259e-05,
+      "loss": 0.6384,
+      "step": 6029
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.39435797548108026,
+      "learning_rate": 1.843774718043981e-05,
+      "loss": 0.6969,
+      "step": 6030
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3864331472049591,
+      "learning_rate": 1.8431972921471283e-05,
+      "loss": 0.667,
+      "step": 6031
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3429910691436734,
+      "learning_rate": 1.8426198794011295e-05,
+      "loss": 0.6764,
+      "step": 6032
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.33762776230233926,
+      "learning_rate": 1.8420424798544112e-05,
+      "loss": 0.6239,
+      "step": 6033
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3523846857310831,
+      "learning_rate": 1.8414650935553987e-05,
+      "loss": 0.6621,
+      "step": 6034
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.43771005641357863,
+      "learning_rate": 1.8408877205525185e-05,
+      "loss": 0.6909,
+      "step": 6035
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4156865629913077,
+      "learning_rate": 1.8403103608941925e-05,
+      "loss": 0.6658,
+      "step": 6036
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.37209951631631655,
+      "learning_rate": 1.8397330146288437e-05,
+      "loss": 0.6507,
+      "step": 6037
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3992513032529473,
+      "learning_rate": 1.8391556818048938e-05,
+      "loss": 0.6804,
+      "step": 6038
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38941978958353596,
+      "learning_rate": 1.838578362470762e-05,
+      "loss": 0.6277,
+      "step": 6039
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3567354534064885,
+      "learning_rate": 1.8380010566748685e-05,
+      "loss": 0.6717,
+      "step": 6040
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3846786069022261,
+      "learning_rate": 1.8374237644656297e-05,
+      "loss": 0.6902,
+      "step": 6041
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3938910729571962,
+      "learning_rate": 1.8368464858914645e-05,
+      "loss": 0.6852,
+      "step": 6042
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.37686175635964875,
+      "learning_rate": 1.836269221000787e-05,
+      "loss": 0.6371,
+      "step": 6043
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3519403902217929,
+      "learning_rate": 1.8356919698420125e-05,
+      "loss": 0.6691,
+      "step": 6044
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38292557963901985,
+      "learning_rate": 1.8351147324635538e-05,
+      "loss": 0.6761,
+      "step": 6045
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.39692599678057633,
+      "learning_rate": 1.8345375089138238e-05,
+      "loss": 0.6514,
+      "step": 6046
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3877703409814465,
+      "learning_rate": 1.833960299241233e-05,
+      "loss": 0.6803,
+      "step": 6047
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4213690521586744,
+      "learning_rate": 1.8333831034941907e-05,
+      "loss": 0.7301,
+      "step": 6048
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3299208591523752,
+      "learning_rate": 1.8328059217211065e-05,
+      "loss": 0.63,
+      "step": 6049
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3503154355217774,
+      "learning_rate": 1.832228753970388e-05,
+      "loss": 0.6563,
+      "step": 6050
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38683046662280446,
+      "learning_rate": 1.8316516002904415e-05,
+      "loss": 0.696,
+      "step": 6051
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3615374724953376,
+      "learning_rate": 1.831074460729672e-05,
+      "loss": 0.6567,
+      "step": 6052
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3772486553037395,
+      "learning_rate": 1.8304973353364833e-05,
+      "loss": 0.6899,
+      "step": 6053
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3989623307448315,
+      "learning_rate": 1.8299202241592785e-05,
+      "loss": 0.6818,
+      "step": 6054
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3854010688618826,
+      "learning_rate": 1.829343127246458e-05,
+      "loss": 0.6861,
+      "step": 6055
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3851328126287681,
+      "learning_rate": 1.828766044646424e-05,
+      "loss": 0.6981,
+      "step": 6056
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.35521438794699217,
+      "learning_rate": 1.828188976407576e-05,
+      "loss": 0.6391,
+      "step": 6057
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3451113333558127,
+      "learning_rate": 1.82761192257831e-05,
+      "loss": 0.6338,
+      "step": 6058
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4057002425775362,
+      "learning_rate": 1.827034883207024e-05,
+      "loss": 0.6839,
+      "step": 6059
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3766111196988251,
+      "learning_rate": 1.826457858342114e-05,
+      "loss": 0.7027,
+      "step": 6060
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3518262605861944,
+      "learning_rate": 1.825880848031973e-05,
+      "loss": 0.649,
+      "step": 6061
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38641218961110224,
+      "learning_rate": 1.8253038523249955e-05,
+      "loss": 0.6918,
+      "step": 6062
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.32669737338862453,
+      "learning_rate": 1.8247268712695727e-05,
+      "loss": 0.6298,
+      "step": 6063
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.37457170688927083,
+      "learning_rate": 1.824149904914096e-05,
+      "loss": 0.6746,
+      "step": 6064
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3530215547744065,
+      "learning_rate": 1.8235729533069545e-05,
+      "loss": 0.6215,
+      "step": 6065
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.32995421384492574,
+      "learning_rate": 1.8229960164965356e-05,
+      "loss": 0.658,
+      "step": 6066
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3894614302676267,
+      "learning_rate": 1.8224190945312276e-05,
+      "loss": 0.6895,
+      "step": 6067
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38833801466169143,
+      "learning_rate": 1.8218421874594148e-05,
+      "loss": 0.6706,
+      "step": 6068
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4056167147044769,
+      "learning_rate": 1.8212652953294834e-05,
+      "loss": 0.6619,
+      "step": 6069
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3353844876014096,
+      "learning_rate": 1.8206884181898153e-05,
+      "loss": 0.6219,
+      "step": 6070
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3853708200806084,
+      "learning_rate": 1.820111556088794e-05,
+      "loss": 0.6833,
+      "step": 6071
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.35499363636752496,
+      "learning_rate": 1.819534709074798e-05,
+      "loss": 0.6335,
+      "step": 6072
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3627189339241273,
+      "learning_rate": 1.818957877196209e-05,
+      "loss": 0.6917,
+      "step": 6073
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.37566663565853087,
+      "learning_rate": 1.818381060501404e-05,
+      "loss": 0.6431,
+      "step": 6074
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.39443233310178016,
+      "learning_rate": 1.817804259038759e-05,
+      "loss": 0.6568,
+      "step": 6075
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3877980503760176,
+      "learning_rate": 1.817227472856652e-05,
+      "loss": 0.6303,
+      "step": 6076
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.45964536022210467,
+      "learning_rate": 1.8166507020034557e-05,
+      "loss": 0.6822,
+      "step": 6077
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.32996492251727805,
+      "learning_rate": 1.8160739465275442e-05,
+      "loss": 0.6281,
+      "step": 6078
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3411246394844653,
+      "learning_rate": 1.815497206477288e-05,
+      "loss": 0.6606,
+      "step": 6079
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3716931861992988,
+      "learning_rate": 1.8149204819010587e-05,
+      "loss": 0.6686,
+      "step": 6080
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.40051171098751626,
+      "learning_rate": 1.814343772847225e-05,
+      "loss": 0.6882,
+      "step": 6081
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3994555437114774,
+      "learning_rate": 1.813767079364154e-05,
+      "loss": 0.6828,
+      "step": 6082
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.37646512100098173,
+      "learning_rate": 1.8131904015002135e-05,
+      "loss": 0.6504,
+      "step": 6083
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4320624319879037,
+      "learning_rate": 1.8126137393037687e-05,
+      "loss": 0.6903,
+      "step": 6084
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.36911433555890183,
+      "learning_rate": 1.8120370928231838e-05,
+      "loss": 0.7034,
+      "step": 6085
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.40715356557021515,
+      "learning_rate": 1.81146046210682e-05,
+      "loss": 0.67,
+      "step": 6086
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4420447138337445,
+      "learning_rate": 1.8108838472030403e-05,
+      "loss": 0.6656,
+      "step": 6087
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4153549354208904,
+      "learning_rate": 1.810307248160204e-05,
+      "loss": 0.669,
+      "step": 6088
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4052640629512302,
+      "learning_rate": 1.8097306650266685e-05,
+      "loss": 0.6594,
+      "step": 6089
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.38482904830713954,
+      "learning_rate": 1.809154097850793e-05,
+      "loss": 0.6802,
+      "step": 6090
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.40960272565701955,
+      "learning_rate": 1.808577546680933e-05,
+      "loss": 0.6496,
+      "step": 6091
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3619249774231085,
+      "learning_rate": 1.8080010115654435e-05,
+      "loss": 0.6458,
+      "step": 6092
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3973766565016022,
+      "learning_rate": 1.8074244925526766e-05,
+      "loss": 0.6791,
+      "step": 6093
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.37982764774403216,
+      "learning_rate": 1.8068479896909853e-05,
+      "loss": 0.6575,
+      "step": 6094
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3832794475961476,
+      "learning_rate": 1.8062715030287198e-05,
+      "loss": 0.6387,
+      "step": 6095
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3659798224274141,
+      "learning_rate": 1.8056950326142284e-05,
+      "loss": 0.6454,
+      "step": 6096
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.365323420656852,
+      "learning_rate": 1.805118578495861e-05,
+      "loss": 0.6537,
+      "step": 6097
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.39215227147772114,
+      "learning_rate": 1.8045421407219627e-05,
+      "loss": 0.6432,
+      "step": 6098
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3483368903426654,
+      "learning_rate": 1.8039657193408788e-05,
+      "loss": 0.6496,
+      "step": 6099
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.39428180128850826,
+      "learning_rate": 1.803389314400954e-05,
+      "loss": 0.6921,
+      "step": 6100
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.35011918233020006,
+      "learning_rate": 1.8028129259505295e-05,
+      "loss": 0.6492,
+      "step": 6101
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.42888856473033926,
+      "learning_rate": 1.8022365540379468e-05,
+      "loss": 0.6944,
+      "step": 6102
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.37737981979023927,
+      "learning_rate": 1.8016601987115448e-05,
+      "loss": 0.6688,
+      "step": 6103
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4125532495447383,
+      "learning_rate": 1.801083860019663e-05,
+      "loss": 0.6587,
+      "step": 6104
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.37706241719996664,
+      "learning_rate": 1.8005075380106384e-05,
+      "loss": 0.6631,
+      "step": 6105
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.37639508416575257,
+      "learning_rate": 1.7999312327328048e-05,
+      "loss": 0.6918,
+      "step": 6106
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3336323802774409,
+      "learning_rate": 1.7993549442344973e-05,
+      "loss": 0.6478,
+      "step": 6107
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.35870599115697377,
+      "learning_rate": 1.7987786725640484e-05,
+      "loss": 0.672,
+      "step": 6108
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3502653037356752,
+      "learning_rate": 1.798202417769789e-05,
+      "loss": 0.6512,
+      "step": 6109
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3632184368876394,
+      "learning_rate": 1.7976261799000486e-05,
+      "loss": 0.6669,
+      "step": 6110
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.41624942206506355,
+      "learning_rate": 1.7970499590031568e-05,
+      "loss": 0.6711,
+      "step": 6111
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3667144393819964,
+      "learning_rate": 1.79647375512744e-05,
+      "loss": 0.6688,
+      "step": 6112
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.43076361023204063,
+      "learning_rate": 1.795897568321223e-05,
+      "loss": 0.6777,
+      "step": 6113
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3417250856717122,
+      "learning_rate": 1.795321398632831e-05,
+      "loss": 0.6431,
+      "step": 6114
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5735993382370823,
+      "learning_rate": 1.794745246110586e-05,
+      "loss": 0.6664,
+      "step": 6115
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3587900768994039,
+      "learning_rate": 1.794169110802809e-05,
+      "loss": 0.6432,
+      "step": 6116
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.336948907206154,
+      "learning_rate": 1.7935929927578196e-05,
+      "loss": 0.661,
+      "step": 6117
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.37172941980504554,
+      "learning_rate": 1.7930168920239375e-05,
+      "loss": 0.6628,
+      "step": 6118
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.40837098894098617,
+      "learning_rate": 1.7924408086494786e-05,
+      "loss": 0.7147,
+      "step": 6119
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.41505568491485834,
+      "learning_rate": 1.7918647426827583e-05,
+      "loss": 0.6612,
+      "step": 6120
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3872069425223913,
+      "learning_rate": 1.7912886941720908e-05,
+      "loss": 0.6366,
+      "step": 6121
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3830913431119368,
+      "learning_rate": 1.7907126631657884e-05,
+      "loss": 0.6762,
+      "step": 6122
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.35514182943317074,
+      "learning_rate": 1.790136649712162e-05,
+      "loss": 0.6404,
+      "step": 6123
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3536901272068983,
+      "learning_rate": 1.789560653859521e-05,
+      "loss": 0.6586,
+      "step": 6124
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4836308327234327,
+      "learning_rate": 1.7889846756561743e-05,
+      "loss": 0.7328,
+      "step": 6125
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.36964257649457766,
+      "learning_rate": 1.7884087151504282e-05,
+      "loss": 0.6667,
+      "step": 6126
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.33822051551900073,
+      "learning_rate": 1.787832772390587e-05,
+      "loss": 0.6039,
+      "step": 6127
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.42390606329349784,
+      "learning_rate": 1.7872568474249557e-05,
+      "loss": 0.6846,
+      "step": 6128
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.37308700530860367,
+      "learning_rate": 1.7866809403018352e-05,
+      "loss": 0.6388,
+      "step": 6129
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.38121979314990156,
+      "learning_rate": 1.7861050510695267e-05,
+      "loss": 0.6942,
+      "step": 6130
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.36535157547624547,
+      "learning_rate": 1.7855291797763287e-05,
+      "loss": 0.6413,
+      "step": 6131
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.33696054938957887,
+      "learning_rate": 1.7849533264705402e-05,
+      "loss": 0.6634,
+      "step": 6132
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.39255284637226817,
+      "learning_rate": 1.784377491200456e-05,
+      "loss": 0.6724,
+      "step": 6133
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.34453038912454303,
+      "learning_rate": 1.7838016740143716e-05,
+      "loss": 0.6583,
+      "step": 6134
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.376750618929586,
+      "learning_rate": 1.78322587496058e-05,
+      "loss": 0.6491,
+      "step": 6135
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4086338381912816,
+      "learning_rate": 1.7826500940873718e-05,
+      "loss": 0.6743,
+      "step": 6136
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4152030044557798,
+      "learning_rate": 1.782074331443038e-05,
+      "loss": 0.6392,
+      "step": 6137
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3573284408045095,
+      "learning_rate": 1.7814985870758662e-05,
+      "loss": 0.6285,
+      "step": 6138
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.40262707826617444,
+      "learning_rate": 1.780922861034145e-05,
+      "loss": 0.6668,
+      "step": 6139
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.39816930299236664,
+      "learning_rate": 1.780347153366158e-05,
+      "loss": 0.6934,
+      "step": 6140
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3818288097037952,
+      "learning_rate": 1.7797714641201907e-05,
+      "loss": 0.6571,
+      "step": 6141
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.37484953690839995,
+      "learning_rate": 1.7791957933445247e-05,
+      "loss": 0.6409,
+      "step": 6142
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4134908417800225,
+      "learning_rate": 1.7786201410874404e-05,
+      "loss": 0.7064,
+      "step": 6143
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.40387363124963416,
+      "learning_rate": 1.778044507397217e-05,
+      "loss": 0.6676,
+      "step": 6144
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.40630528244405434,
+      "learning_rate": 1.7774688923221336e-05,
+      "loss": 0.6838,
+      "step": 6145
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.46395034393701395,
+      "learning_rate": 1.7768932959104655e-05,
+      "loss": 0.6646,
+      "step": 6146
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.43146291853708235,
+      "learning_rate": 1.776317718210487e-05,
+      "loss": 0.6693,
+      "step": 6147
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.312708897937671,
+      "learning_rate": 1.7757421592704716e-05,
+      "loss": 7.022,
+      "step": 6148
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3660718963617434,
+      "learning_rate": 1.7751666191386903e-05,
+      "loss": 0.6402,
+      "step": 6149
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3451263022574987,
+      "learning_rate": 1.7745910978634128e-05,
+      "loss": 0.6838,
+      "step": 6150
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4045290860967485,
+      "learning_rate": 1.7740155954929075e-05,
+      "loss": 0.6786,
+      "step": 6151
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3793453175961619,
+      "learning_rate": 1.773440112075442e-05,
+      "loss": 0.6421,
+      "step": 6152
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3322819852424164,
+      "learning_rate": 1.7728646476592802e-05,
+      "loss": 0.6699,
+      "step": 6153
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3693184296081721,
+      "learning_rate": 1.7722892022926862e-05,
+      "loss": 0.648,
+      "step": 6154
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.36125152119240517,
+      "learning_rate": 1.7717137760239224e-05,
+      "loss": 0.6478,
+      "step": 6155
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3876331795913414,
+      "learning_rate": 1.7711383689012474e-05,
+      "loss": 0.6608,
+      "step": 6156
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.37303713407653355,
+      "learning_rate": 1.770562980972922e-05,
+      "loss": 0.6898,
+      "step": 6157
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4231260317274098,
+      "learning_rate": 1.7699876122872013e-05,
+      "loss": 0.6615,
+      "step": 6158
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3485934056994257,
+      "learning_rate": 1.7694122628923428e-05,
+      "loss": 0.65,
+      "step": 6159
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3756644041971924,
+      "learning_rate": 1.7688369328365986e-05,
+      "loss": 0.6336,
+      "step": 6160
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3564552517944668,
+      "learning_rate": 1.768261622168222e-05,
+      "loss": 0.632,
+      "step": 6161
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3757157094228882,
+      "learning_rate": 1.7676863309354636e-05,
+      "loss": 0.6502,
+      "step": 6162
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.36081553468357463,
+      "learning_rate": 1.7671110591865713e-05,
+      "loss": 0.6646,
+      "step": 6163
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4395246314798202,
+      "learning_rate": 1.766535806969794e-05,
+      "loss": 0.6865,
+      "step": 6164
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4071944742302026,
+      "learning_rate": 1.7659605743333756e-05,
+      "loss": 0.6779,
+      "step": 6165
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3604203508686588,
+      "learning_rate": 1.765385361325562e-05,
+      "loss": 0.6497,
+      "step": 6166
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3684090300786254,
+      "learning_rate": 1.7648101679945945e-05,
+      "loss": 0.6318,
+      "step": 6167
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4134900163985627,
+      "learning_rate": 1.7642349943887146e-05,
+      "loss": 0.6468,
+      "step": 6168
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.40613499172060774,
+      "learning_rate": 1.763659840556161e-05,
+      "loss": 0.6603,
+      "step": 6169
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.36994425702547074,
+      "learning_rate": 1.7630847065451705e-05,
+      "loss": 0.6625,
+      "step": 6170
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3424989931122598,
+      "learning_rate": 1.76250959240398e-05,
+      "loss": 0.638,
+      "step": 6171
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3567427243452327,
+      "learning_rate": 1.7619344981808225e-05,
+      "loss": 0.6681,
+      "step": 6172
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.36767551019081746,
+      "learning_rate": 1.7613594239239317e-05,
+      "loss": 0.6266,
+      "step": 6173
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3792502545948233,
+      "learning_rate": 1.7607843696815376e-05,
+      "loss": 0.6954,
+      "step": 6174
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3681480847042529,
+      "learning_rate": 1.76020933550187e-05,
+      "loss": 0.6714,
+      "step": 6175
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.38985497340506237,
+      "learning_rate": 1.7596343214331557e-05,
+      "loss": 0.7177,
+      "step": 6176
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.40975181511105,
+      "learning_rate": 1.7590593275236202e-05,
+      "loss": 0.6783,
+      "step": 6177
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.37670606820091135,
+      "learning_rate": 1.7584843538214885e-05,
+      "loss": 0.6837,
+      "step": 6178
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3286789690309884,
+      "learning_rate": 1.7579094003749814e-05,
+      "loss": 0.6039,
+      "step": 6179
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3319667039934395,
+      "learning_rate": 1.757334467232321e-05,
+      "loss": 0.6431,
+      "step": 6180
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3838160644527377,
+      "learning_rate": 1.7567595544417258e-05,
+      "loss": 0.6787,
+      "step": 6181
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.42960037331534534,
+      "learning_rate": 1.7561846620514132e-05,
+      "loss": 0.6932,
+      "step": 6182
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.42270287869585016,
+      "learning_rate": 1.755609790109599e-05,
+      "loss": 0.7036,
+      "step": 6183
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3965796208449693,
+      "learning_rate": 1.7550349386644954e-05,
+      "loss": 0.6934,
+      "step": 6184
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3437851086927114,
+      "learning_rate": 1.7544601077643167e-05,
+      "loss": 0.6416,
+      "step": 6185
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.405609331595519,
+      "learning_rate": 1.7538852974572712e-05,
+      "loss": 0.674,
+      "step": 6186
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.39111223464069217,
+      "learning_rate": 1.7533105077915695e-05,
+      "loss": 0.6535,
+      "step": 6187
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4203436499185545,
+      "learning_rate": 1.7527357388154166e-05,
+      "loss": 0.7047,
+      "step": 6188
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.677304039141015,
+      "learning_rate": 1.7521609905770196e-05,
+      "loss": 6.9299,
+      "step": 6189
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.36394896313224273,
+      "learning_rate": 1.7515862631245807e-05,
+      "loss": 0.6492,
+      "step": 6190
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3892918060066155,
+      "learning_rate": 1.751011556506302e-05,
+      "loss": 0.6655,
+      "step": 6191
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.381256271020503,
+      "learning_rate": 1.7504368707703834e-05,
+      "loss": 0.6885,
+      "step": 6192
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3110862500682283,
+      "learning_rate": 1.7498622059650224e-05,
+      "loss": 0.6335,
+      "step": 6193
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3839232023224397,
+      "learning_rate": 1.749287562138416e-05,
+      "loss": 0.7006,
+      "step": 6194
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.41166127024693094,
+      "learning_rate": 1.74871293933876e-05,
+      "loss": 0.6613,
+      "step": 6195
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3513588026151969,
+      "learning_rate": 1.748138337614246e-05,
+      "loss": 0.6342,
+      "step": 6196
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.36903953743708573,
+      "learning_rate": 1.7475637570130654e-05,
+      "loss": 0.6384,
+      "step": 6197
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3641318365209394,
+      "learning_rate": 1.7469891975834077e-05,
+      "loss": 0.6498,
+      "step": 6198
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4618480857030218,
+      "learning_rate": 1.7464146593734606e-05,
+      "loss": 0.6766,
+      "step": 6199
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4104264448306001,
+      "learning_rate": 1.7458401424314085e-05,
+      "loss": 0.6804,
+      "step": 6200
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.37531959178680996,
+      "learning_rate": 1.7452656468054377e-05,
+      "loss": 0.6195,
+      "step": 6201
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4149893432360101,
+      "learning_rate": 1.7446911725437296e-05,
+      "loss": 0.6869,
+      "step": 6202
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3887917601398308,
+      "learning_rate": 1.7441167196944645e-05,
+      "loss": 0.6507,
+      "step": 6203
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3812366256535385,
+      "learning_rate": 1.743542288305821e-05,
+      "loss": 0.6896,
+      "step": 6204
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4067679071253796,
+      "learning_rate": 1.742967878425976e-05,
+      "loss": 0.683,
+      "step": 6205
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.34399729008051483,
+      "learning_rate": 1.7423934901031048e-05,
+      "loss": 0.6642,
+      "step": 6206
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3541725053075815,
+      "learning_rate": 1.7418191233853797e-05,
+      "loss": 0.6568,
+      "step": 6207
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3330608180575759,
+      "learning_rate": 1.741244778320973e-05,
+      "loss": 0.6548,
+      "step": 6208
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38619913130989325,
+      "learning_rate": 1.7406704549580555e-05,
+      "loss": 0.6742,
+      "step": 6209
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4014032650017743,
+      "learning_rate": 1.7400961533447928e-05,
+      "loss": 0.6837,
+      "step": 6210
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4238920283566734,
+      "learning_rate": 1.739521873529352e-05,
+      "loss": 0.618,
+      "step": 6211
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3835621797377135,
+      "learning_rate": 1.7389476155598974e-05,
+      "loss": 0.6238,
+      "step": 6212
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.34499433850758815,
+      "learning_rate": 1.7383733794845904e-05,
+      "loss": 0.624,
+      "step": 6213
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37665406014782016,
+      "learning_rate": 1.7377991653515925e-05,
+      "loss": 0.6604,
+      "step": 6214
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.36683417070509167,
+      "learning_rate": 1.7372249732090617e-05,
+      "loss": 0.6392,
+      "step": 6215
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3843932707279337,
+      "learning_rate": 1.7366508031051553e-05,
+      "loss": 0.6569,
+      "step": 6216
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3606392728865644,
+      "learning_rate": 1.7360766550880276e-05,
+      "loss": 0.67,
+      "step": 6217
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3525801100028506,
+      "learning_rate": 1.7355025292058326e-05,
+      "loss": 0.6835,
+      "step": 6218
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38557775782286674,
+      "learning_rate": 1.734928425506721e-05,
+      "loss": 0.6526,
+      "step": 6219
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.39369972270618514,
+      "learning_rate": 1.7343543440388423e-05,
+      "loss": 0.6578,
+      "step": 6220
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3607493317709914,
+      "learning_rate": 1.7337802848503428e-05,
+      "loss": 0.697,
+      "step": 6221
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38562002376812066,
+      "learning_rate": 1.7332062479893705e-05,
+      "loss": 0.6717,
+      "step": 6222
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3743166104818402,
+      "learning_rate": 1.732632233504068e-05,
+      "loss": 0.6575,
+      "step": 6223
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37835483435595585,
+      "learning_rate": 1.7320582414425766e-05,
+      "loss": 0.6789,
+      "step": 6224
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37632669654225553,
+      "learning_rate": 1.7314842718530376e-05,
+      "loss": 0.6557,
+      "step": 6225
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3567042809434403,
+      "learning_rate": 1.7309103247835882e-05,
+      "loss": 0.6548,
+      "step": 6226
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.35765367628715705,
+      "learning_rate": 1.730336400282364e-05,
+      "loss": 0.6382,
+      "step": 6227
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.39443385818219684,
+      "learning_rate": 1.7297624983975008e-05,
+      "loss": 0.7095,
+      "step": 6228
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.36357527696282643,
+      "learning_rate": 1.729188619177131e-05,
+      "loss": 0.6612,
+      "step": 6229
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.34093212384524046,
+      "learning_rate": 1.7286147626693845e-05,
+      "loss": 0.6615,
+      "step": 6230
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38155357227253367,
+      "learning_rate": 1.7280409289223895e-05,
+      "loss": 0.6779,
+      "step": 6231
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37046648696946877,
+      "learning_rate": 1.727467117984274e-05,
+      "loss": 0.6937,
+      "step": 6232
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4075030100592945,
+      "learning_rate": 1.7268933299031623e-05,
+      "loss": 0.6808,
+      "step": 6233
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.2881314243800395,
+      "learning_rate": 1.726319564727176e-05,
+      "loss": 7.0938,
+      "step": 6234
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3560220576676446,
+      "learning_rate": 1.725745822504438e-05,
+      "loss": 0.6729,
+      "step": 6235
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38401790659409757,
+      "learning_rate": 1.7251721032830672e-05,
+      "loss": 0.676,
+      "step": 6236
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.40384308066020846,
+      "learning_rate": 1.7245984071111797e-05,
+      "loss": 0.6815,
+      "step": 6237
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.42696277466321014,
+      "learning_rate": 1.724024734036891e-05,
+      "loss": 0.6993,
+      "step": 6238
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3956321562171736,
+      "learning_rate": 1.723451084108315e-05,
+      "loss": 0.6699,
+      "step": 6239
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3926358689095438,
+      "learning_rate": 1.7228774573735628e-05,
+      "loss": 0.6729,
+      "step": 6240
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3658888984833933,
+      "learning_rate": 1.7223038538807424e-05,
+      "loss": 0.6268,
+      "step": 6241
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4341920105456103,
+      "learning_rate": 1.721730273677963e-05,
+      "loss": 0.71,
+      "step": 6242
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.36206317364460977,
+      "learning_rate": 1.72115671681333e-05,
+      "loss": 0.6572,
+      "step": 6243
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.351001033482235,
+      "learning_rate": 1.720583183334946e-05,
+      "loss": 0.6262,
+      "step": 6244
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.34936647731928816,
+      "learning_rate": 1.720009673290913e-05,
+      "loss": 0.6572,
+      "step": 6245
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37933102686964754,
+      "learning_rate": 1.7194361867293307e-05,
+      "loss": 0.6573,
+      "step": 6246
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4360444935973272,
+      "learning_rate": 1.7188627236982963e-05,
+      "loss": 0.6833,
+      "step": 6247
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4151496216313952,
+      "learning_rate": 1.7182892842459057e-05,
+      "loss": 0.6656,
+      "step": 6248
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3819892443451882,
+      "learning_rate": 1.7177158684202526e-05,
+      "loss": 0.6406,
+      "step": 6249
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.2437004098613136,
+      "learning_rate": 1.7171424762694292e-05,
+      "loss": 7.0129,
+      "step": 6250
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37665960965865136,
+      "learning_rate": 1.716569107841524e-05,
+      "loss": 0.6739,
+      "step": 6251
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3573225508962722,
+      "learning_rate": 1.7159957631846264e-05,
+      "loss": 0.644,
+      "step": 6252
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3695925523387414,
+      "learning_rate": 1.7154224423468208e-05,
+      "loss": 0.6633,
+      "step": 6253
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3854200131926975,
+      "learning_rate": 1.7148491453761905e-05,
+      "loss": 0.6702,
+      "step": 6254
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37158429290107947,
+      "learning_rate": 1.7142758723208182e-05,
+      "loss": 0.6504,
+      "step": 6255
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.35499760293249055,
+      "learning_rate": 1.713702623228784e-05,
+      "loss": 0.6213,
+      "step": 6256
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37415786162459996,
+      "learning_rate": 1.713129398148165e-05,
+      "loss": 0.5983,
+      "step": 6257
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.34619009296888575,
+      "learning_rate": 1.7125561971270367e-05,
+      "loss": 0.6352,
+      "step": 6258
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.39586169272912325,
+      "learning_rate": 1.711983020213473e-05,
+      "loss": 0.6699,
+      "step": 6259
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3522851575541752,
+      "learning_rate": 1.711409867455546e-05,
+      "loss": 0.668,
+      "step": 6260
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.35471917257470875,
+      "learning_rate": 1.7108367389013244e-05,
+      "loss": 0.6539,
+      "step": 6261
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38370407225256115,
+      "learning_rate": 1.7102636345988757e-05,
+      "loss": 0.6729,
+      "step": 6262
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3906950440701388,
+      "learning_rate": 1.7096905545962668e-05,
+      "loss": 0.6382,
+      "step": 6263
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.35238450067011917,
+      "learning_rate": 1.7091174989415607e-05,
+      "loss": 0.647,
+      "step": 6264
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.40339370216003384,
+      "learning_rate": 1.7085444676828183e-05,
+      "loss": 0.6896,
+      "step": 6265
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3802690463440936,
+      "learning_rate": 1.7079714608680996e-05,
+      "loss": 0.6619,
+      "step": 6266
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.40764947978713034,
+      "learning_rate": 1.707398478545462e-05,
+      "loss": 0.7081,
+      "step": 6267
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.40771641543478343,
+      "learning_rate": 1.70682552076296e-05,
+      "loss": 0.6923,
+      "step": 6268
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3597525183659057,
+      "learning_rate": 1.7062525875686472e-05,
+      "loss": 0.6255,
+      "step": 6269
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4306386714478443,
+      "learning_rate": 1.705679679010576e-05,
+      "loss": 0.6971,
+      "step": 6270
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3564521981276586,
+      "learning_rate": 1.7051067951367945e-05,
+      "loss": 0.6616,
+      "step": 6271
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4077274370720037,
+      "learning_rate": 1.70453393599535e-05,
+      "loss": 0.7306,
+      "step": 6272
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37790216940521537,
+      "learning_rate": 1.7039611016342873e-05,
+      "loss": 0.6786,
+      "step": 6273
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3542712955250376,
+      "learning_rate": 1.7033882921016495e-05,
+      "loss": 0.6663,
+      "step": 6274
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.34453912487978905,
+      "learning_rate": 1.7028155074454776e-05,
+      "loss": 0.64,
+      "step": 6275
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4160157215397023,
+      "learning_rate": 1.7022427477138093e-05,
+      "loss": 0.6671,
+      "step": 6276
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.40809026142959787,
+      "learning_rate": 1.701670012954683e-05,
+      "loss": 0.7115,
+      "step": 6277
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4392074137014451,
+      "learning_rate": 1.7010973032161322e-05,
+      "loss": 0.6709,
+      "step": 6278
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3841514493627408,
+      "learning_rate": 1.70052461854619e-05,
+      "loss": 0.6365,
+      "step": 6279
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3944871684864094,
+      "learning_rate": 1.699951958992886e-05,
+      "loss": 0.6561,
+      "step": 6280
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3802549022427526,
+      "learning_rate": 1.699379324604249e-05,
+      "loss": 0.6774,
+      "step": 6281
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3756920137358451,
+      "learning_rate": 1.698806715428305e-05,
+      "loss": 0.6488,
+      "step": 6282
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.36672730567397815,
+      "learning_rate": 1.6982341315130772e-05,
+      "loss": 0.6357,
+      "step": 6283
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3769015659248492,
+      "learning_rate": 1.6976615729065893e-05,
+      "loss": 0.6587,
+      "step": 6284
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3740844026990775,
+      "learning_rate": 1.6970890396568598e-05,
+      "loss": 0.6682,
+      "step": 6285
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8497626719085087,
+      "learning_rate": 1.696516531811907e-05,
+      "loss": 7.0809,
+      "step": 6286
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3620793399526604,
+      "learning_rate": 1.695944049419746e-05,
+      "loss": 0.6822,
+      "step": 6287
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.40773528064217934,
+      "learning_rate": 1.6953715925283902e-05,
+      "loss": 0.6914,
+      "step": 6288
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38861199086328757,
+      "learning_rate": 1.6947991611858515e-05,
+      "loss": 0.6686,
+      "step": 6289
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3769170729749784,
+      "learning_rate": 1.694226755440138e-05,
+      "loss": 0.6569,
+      "step": 6290
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3365366230366306,
+      "learning_rate": 1.6936543753392575e-05,
+      "loss": 0.6477,
+      "step": 6291
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3363070035236765,
+      "learning_rate": 1.6930820209312143e-05,
+      "loss": 0.6295,
+      "step": 6292
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.39238575314968865,
+      "learning_rate": 1.6925096922640115e-05,
+      "loss": 0.6347,
+      "step": 6293
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4265836461436611,
+      "learning_rate": 1.69193738938565e-05,
+      "loss": 0.6932,
+      "step": 6294
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.36457437738134013,
+      "learning_rate": 1.691365112344127e-05,
+      "loss": 0.6532,
+      "step": 6295
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3636752382109582,
+      "learning_rate": 1.6907928611874396e-05,
+      "loss": 0.6632,
+      "step": 6296
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3740863188304835,
+      "learning_rate": 1.6902206359635805e-05,
+      "loss": 0.7016,
+      "step": 6297
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3482279093308694,
+      "learning_rate": 1.6896484367205438e-05,
+      "loss": 0.6549,
+      "step": 6298
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4164299674015566,
+      "learning_rate": 1.6890762635063173e-05,
+      "loss": 0.6836,
+      "step": 6299
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.372610109828466,
+      "learning_rate": 1.6885041163688894e-05,
+      "loss": 0.6983,
+      "step": 6300
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4146082358475099,
+      "learning_rate": 1.6879319953562453e-05,
+      "loss": 0.6716,
+      "step": 6301
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3586831125202301,
+      "learning_rate": 1.6873599005163673e-05,
+      "loss": 0.6768,
+      "step": 6302
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3800601924971378,
+      "learning_rate": 1.6867878318972372e-05,
+      "loss": 0.6412,
+      "step": 6303
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3630324052728003,
+      "learning_rate": 1.6862157895468326e-05,
+      "loss": 0.6599,
+      "step": 6304
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4146344458684669,
+      "learning_rate": 1.6856437735131317e-05,
+      "loss": 0.6633,
+      "step": 6305
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38906464156870807,
+      "learning_rate": 1.6850717838441076e-05,
+      "loss": 0.6913,
+      "step": 6306
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.36843140922288053,
+      "learning_rate": 1.6844998205877328e-05,
+      "loss": 0.6226,
+      "step": 6307
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38595815574673026,
+      "learning_rate": 1.6839278837919767e-05,
+      "loss": 0.7097,
+      "step": 6308
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4014339742427698,
+      "learning_rate": 1.683355973504808e-05,
+      "loss": 0.6849,
+      "step": 6309
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4436381486305327,
+      "learning_rate": 1.6827840897741907e-05,
+      "loss": 0.687,
+      "step": 6310
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3513280321008188,
+      "learning_rate": 1.6822122326480884e-05,
+      "loss": 0.6276,
+      "step": 6311
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4288997940578963,
+      "learning_rate": 1.6816404021744625e-05,
+      "loss": 0.6586,
+      "step": 6312
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.41429957998217337,
+      "learning_rate": 1.681068598401272e-05,
+      "loss": 0.6681,
+      "step": 6313
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4058961935314111,
+      "learning_rate": 1.680496821376473e-05,
+      "loss": 0.6465,
+      "step": 6314
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.39008941129580255,
+      "learning_rate": 1.6799250711480195e-05,
+      "loss": 0.6681,
+      "step": 6315
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3852512200589718,
+      "learning_rate": 1.6793533477638638e-05,
+      "loss": 0.6585,
+      "step": 6316
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3650466786459179,
+      "learning_rate": 1.6787816512719552e-05,
+      "loss": 0.6709,
+      "step": 6317
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3773797876793646,
+      "learning_rate": 1.6782099817202423e-05,
+      "loss": 0.6583,
+      "step": 6318
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.38112192755847285,
+      "learning_rate": 1.677638339156669e-05,
+      "loss": 0.7052,
+      "step": 6319
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4077208483092955,
+      "learning_rate": 1.6770667236291792e-05,
+      "loss": 0.7046,
+      "step": 6320
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.386672700100387,
+      "learning_rate": 1.6764951351857136e-05,
+      "loss": 0.6614,
+      "step": 6321
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4124341383365625,
+      "learning_rate": 1.6759235738742095e-05,
+      "loss": 0.6864,
+      "step": 6322
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.34338877079918223,
+      "learning_rate": 1.675352039742605e-05,
+      "loss": 0.6238,
+      "step": 6323
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3872605534996954,
+      "learning_rate": 1.6747805328388314e-05,
+      "loss": 0.6685,
+      "step": 6324
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3950369835345636,
+      "learning_rate": 1.6742090532108228e-05,
+      "loss": 0.6895,
+      "step": 6325
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3987216412835802,
+      "learning_rate": 1.673637600906507e-05,
+      "loss": 0.6588,
+      "step": 6326
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.37636275539816816,
+      "learning_rate": 1.6730661759738117e-05,
+      "loss": 0.664,
+      "step": 6327
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.38014294802570664,
+      "learning_rate": 1.6724947784606617e-05,
+      "loss": 0.6676,
+      "step": 6328
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4019616454788246,
+      "learning_rate": 1.6719234084149786e-05,
+      "loss": 0.6813,
+      "step": 6329
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.41442964262144566,
+      "learning_rate": 1.6713520658846834e-05,
+      "loss": 0.6518,
+      "step": 6330
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.34318661082268737,
+      "learning_rate": 1.6707807509176924e-05,
+      "loss": 0.6851,
+      "step": 6331
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.35726355439066365,
+      "learning_rate": 1.6702094635619233e-05,
+      "loss": 0.6771,
+      "step": 6332
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3921272125163979,
+      "learning_rate": 1.6696382038652877e-05,
+      "loss": 0.6739,
+      "step": 6333
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.42249505483768085,
+      "learning_rate": 1.6690669718756978e-05,
+      "loss": 0.6855,
+      "step": 6334
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3539108618825002,
+      "learning_rate": 1.6684957676410604e-05,
+      "loss": 0.6343,
+      "step": 6335
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.36732445473173303,
+      "learning_rate": 1.6679245912092834e-05,
+      "loss": 0.6773,
+      "step": 6336
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3349971050810837,
+      "learning_rate": 1.6673534426282695e-05,
+      "loss": 0.6356,
+      "step": 6337
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.33779498597735225,
+      "learning_rate": 1.6667823219459202e-05,
+      "loss": 0.6085,
+      "step": 6338
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.38873934023600143,
+      "learning_rate": 1.6662112292101352e-05,
+      "loss": 0.6376,
+      "step": 6339
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.39516020329828205,
+      "learning_rate": 1.6656401644688118e-05,
+      "loss": 0.6261,
+      "step": 6340
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4224060198432209,
+      "learning_rate": 1.6650691277698442e-05,
+      "loss": 0.6856,
+      "step": 6341
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.36028522366189747,
+      "learning_rate": 1.664498119161124e-05,
+      "loss": 0.6847,
+      "step": 6342
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3176984057635746,
+      "learning_rate": 1.6639271386905415e-05,
+      "loss": 0.6346,
+      "step": 6343
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3802979416177736,
+      "learning_rate": 1.663356186405984e-05,
+      "loss": 0.6482,
+      "step": 6344
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.39895061741017407,
+      "learning_rate": 1.662785262355336e-05,
+      "loss": 0.6696,
+      "step": 6345
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3626545545893596,
+      "learning_rate": 1.662214366586481e-05,
+      "loss": 0.6172,
+      "step": 6346
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.40549368766660304,
+      "learning_rate": 1.6616434991472995e-05,
+      "loss": 0.6808,
+      "step": 6347
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3727392842626607,
+      "learning_rate": 1.6610726600856693e-05,
+      "loss": 0.6859,
+      "step": 6348
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3738001427746335,
+      "learning_rate": 1.660501849449465e-05,
+      "loss": 0.6418,
+      "step": 6349
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3759195516686619,
+      "learning_rate": 1.6599310672865615e-05,
+      "loss": 0.6676,
+      "step": 6350
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.343261360658499,
+      "learning_rate": 1.6593603136448285e-05,
+      "loss": 0.6797,
+      "step": 6351
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.45509693384924493,
+      "learning_rate": 1.6587895885721334e-05,
+      "loss": 0.721,
+      "step": 6352
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3617667505772969,
+      "learning_rate": 1.6582188921163444e-05,
+      "loss": 0.675,
+      "step": 6353
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4152241931554719,
+      "learning_rate": 1.6576482243253243e-05,
+      "loss": 0.673,
+      "step": 6354
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.32568076254765826,
+      "learning_rate": 1.657077585246934e-05,
+      "loss": 0.6339,
+      "step": 6355
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.398798298942432,
+      "learning_rate": 1.6565069749290323e-05,
+      "loss": 0.6954,
+      "step": 6356
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.364435933699849,
+      "learning_rate": 1.6559363934194763e-05,
+      "loss": 0.6216,
+      "step": 6357
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.38501127385195616,
+      "learning_rate": 1.655365840766119e-05,
+      "loss": 0.648,
+      "step": 6358
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3613058963845012,
+      "learning_rate": 1.654795317016812e-05,
+      "loss": 0.6291,
+      "step": 6359
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3873621101098607,
+      "learning_rate": 1.6542248222194052e-05,
+      "loss": 0.641,
+      "step": 6360
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.42497059663086995,
+      "learning_rate": 1.6536543564217456e-05,
+      "loss": 0.7103,
+      "step": 6361
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4249739013897492,
+      "learning_rate": 1.6530839196716762e-05,
+      "loss": 0.6828,
+      "step": 6362
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.39544073450899797,
+      "learning_rate": 1.65251351201704e-05,
+      "loss": 0.6349,
+      "step": 6363
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.38170392227250366,
+      "learning_rate": 1.6519431335056755e-05,
+      "loss": 0.6543,
+      "step": 6364
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3907733554272182,
+      "learning_rate": 1.65137278418542e-05,
+      "loss": 0.7221,
+      "step": 6365
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4183125534677653,
+      "learning_rate": 1.6508024641041078e-05,
+      "loss": 0.7062,
+      "step": 6366
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3817462555408398,
+      "learning_rate": 1.6502321733095713e-05,
+      "loss": 0.6652,
+      "step": 6367
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.37588253428002577,
+      "learning_rate": 1.6496619118496405e-05,
+      "loss": 0.6438,
+      "step": 6368
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4232308895015148,
+      "learning_rate": 1.6490916797721415e-05,
+      "loss": 0.6623,
+      "step": 6369
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3793146180388642,
+      "learning_rate": 1.6485214771249002e-05,
+      "loss": 0.6665,
+      "step": 6370
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3781722977177904,
+      "learning_rate": 1.6479513039557383e-05,
+      "loss": 0.6609,
+      "step": 6371
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3820138320200648,
+      "learning_rate": 1.6473811603124746e-05,
+      "loss": 0.6347,
+      "step": 6372
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.36815673280689837,
+      "learning_rate": 1.646811046242927e-05,
+      "loss": 0.6577,
+      "step": 6373
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.37198890938697604,
+      "learning_rate": 1.646240961794911e-05,
+      "loss": 0.6538,
+      "step": 6374
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4329616433206994,
+      "learning_rate": 1.645670907016238e-05,
+      "loss": 0.665,
+      "step": 6375
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3772992073406258,
+      "learning_rate": 1.6451008819547183e-05,
+      "loss": 0.6611,
+      "step": 6376
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.37586203481469194,
+      "learning_rate": 1.6445308866581592e-05,
+      "loss": 0.6524,
+      "step": 6377
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.38704939484217954,
+      "learning_rate": 1.6439609211743653e-05,
+      "loss": 0.6362,
+      "step": 6378
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.41369319794708626,
+      "learning_rate": 1.6433909855511384e-05,
+      "loss": 0.6918,
+      "step": 6379
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.43168041284425507,
+      "learning_rate": 1.6428210798362787e-05,
+      "loss": 0.7327,
+      "step": 6380
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3781713702887533,
+      "learning_rate": 1.642251204077584e-05,
+      "loss": 0.6522,
+      "step": 6381
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4243049019541015,
+      "learning_rate": 1.6416813583228494e-05,
+      "loss": 0.7071,
+      "step": 6382
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.42131959161670607,
+      "learning_rate": 1.641111542619866e-05,
+      "loss": 0.6763,
+      "step": 6383
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3792952373685859,
+      "learning_rate": 1.6405417570164244e-05,
+      "loss": 0.6926,
+      "step": 6384
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.36842100407255873,
+      "learning_rate": 1.6399720015603112e-05,
+      "loss": 0.6553,
+      "step": 6385
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.42136773993340715,
+      "learning_rate": 1.6394022762993116e-05,
+      "loss": 0.6408,
+      "step": 6386
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3786695283891925,
+      "learning_rate": 1.6388325812812066e-05,
+      "loss": 0.6568,
+      "step": 6387
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.40269141822730903,
+      "learning_rate": 1.638262916553778e-05,
+      "loss": 0.6585,
+      "step": 6388
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.39423854622176174,
+      "learning_rate": 1.637693282164802e-05,
+      "loss": 0.6894,
+      "step": 6389
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.34076280607287174,
+      "learning_rate": 1.637123678162052e-05,
+      "loss": 0.6249,
+      "step": 6390
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3574871062756524,
+      "learning_rate": 1.6365541045933014e-05,
+      "loss": 0.6535,
+      "step": 6391
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.42601702957218524,
+      "learning_rate": 1.6359845615063188e-05,
+      "loss": 0.6566,
+      "step": 6392
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3845002986142557,
+      "learning_rate": 1.635415048948872e-05,
+      "loss": 0.643,
+      "step": 6393
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.34109082551768427,
+      "learning_rate": 1.6348455669687235e-05,
+      "loss": 0.6598,
+      "step": 6394
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3453129526650697,
+      "learning_rate": 1.634276115613637e-05,
+      "loss": 0.6408,
+      "step": 6395
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.43333044249673636,
+      "learning_rate": 1.633706694931371e-05,
+      "loss": 0.6564,
+      "step": 6396
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3792195403985722,
+      "learning_rate": 1.6331373049696823e-05,
+      "loss": 0.6841,
+      "step": 6397
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3929588828007873,
+      "learning_rate": 1.632567945776325e-05,
+      "loss": 0.6326,
+      "step": 6398
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.354154941016977,
+      "learning_rate": 1.63199861739905e-05,
+      "loss": 0.6539,
+      "step": 6399
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4484592787231865,
+      "learning_rate": 1.6314293198856056e-05,
+      "loss": 0.6661,
+      "step": 6400
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3777414473283298,
+      "learning_rate": 1.6308600532837404e-05,
+      "loss": 0.6749,
+      "step": 6401
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3629779202582488,
+      "learning_rate": 1.6302908176411963e-05,
+      "loss": 0.6542,
+      "step": 6402
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4274323146172456,
+      "learning_rate": 1.6297216130057146e-05,
+      "loss": 0.6669,
+      "step": 6403
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.41664035757117984,
+      "learning_rate": 1.6291524394250347e-05,
+      "loss": 0.6883,
+      "step": 6404
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3924878391422984,
+      "learning_rate": 1.6285832969468917e-05,
+      "loss": 0.6702,
+      "step": 6405
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.35923239478544466,
+      "learning_rate": 1.6280141856190182e-05,
+      "loss": 0.6611,
+      "step": 6406
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3835050939346079,
+      "learning_rate": 1.6274451054891456e-05,
+      "loss": 0.6871,
+      "step": 6407
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.38001134592423286,
+      "learning_rate": 1.6268760566050028e-05,
+      "loss": 0.6975,
+      "step": 6408
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.35256995036319366,
+      "learning_rate": 1.6263070390143147e-05,
+      "loss": 0.6318,
+      "step": 6409
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.350660719276044,
+      "learning_rate": 1.6257380527648034e-05,
+      "loss": 0.6503,
+      "step": 6410
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3465063114174042,
+      "learning_rate": 1.6251690979041902e-05,
+      "loss": 0.635,
+      "step": 6411
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.394099525582191,
+      "learning_rate": 1.6246001744801918e-05,
+      "loss": 0.7154,
+      "step": 6412
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3776051299599249,
+      "learning_rate": 1.6240312825405232e-05,
+      "loss": 0.6732,
+      "step": 6413
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.36485907003057105,
+      "learning_rate": 1.623462422132896e-05,
+      "loss": 0.6276,
+      "step": 6414
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.37503524414663914,
+      "learning_rate": 1.6228935933050223e-05,
+      "loss": 0.6462,
+      "step": 6415
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3791597044249326,
+      "learning_rate": 1.622324796104607e-05,
+      "loss": 0.6821,
+      "step": 6416
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.43294732436941763,
+      "learning_rate": 1.6217560305793547e-05,
+      "loss": 0.6548,
+      "step": 6417
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3937323482790647,
+      "learning_rate": 1.6211872967769677e-05,
+      "loss": 0.7018,
+      "step": 6418
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.31382250225712044,
+      "learning_rate": 1.6206185947451447e-05,
+      "loss": 0.6614,
+      "step": 6419
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.38555127867288375,
+      "learning_rate": 1.6200499245315814e-05,
+      "loss": 0.6317,
+      "step": 6420
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3855360355673916,
+      "learning_rate": 1.6194812861839716e-05,
+      "loss": 0.6781,
+      "step": 6421
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3802197219317759,
+      "learning_rate": 1.618912679750008e-05,
+      "loss": 0.6842,
+      "step": 6422
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3757049197716606,
+      "learning_rate": 1.6183441052773775e-05,
+      "loss": 0.6609,
+      "step": 6423
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.398617012300218,
+      "learning_rate": 1.6177755628137655e-05,
+      "loss": 0.6828,
+      "step": 6424
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.40240394255284934,
+      "learning_rate": 1.6172070524068556e-05,
+      "loss": 0.6885,
+      "step": 6425
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.39057032249485313,
+      "learning_rate": 1.6166385741043278e-05,
+      "loss": 0.6695,
+      "step": 6426
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.37024067166532965,
+      "learning_rate": 1.61607012795386e-05,
+      "loss": 0.6688,
+      "step": 6427
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3295915126586618,
+      "learning_rate": 1.6155017140031263e-05,
+      "loss": 0.6479,
+      "step": 6428
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3843869249205382,
+      "learning_rate": 1.6149333322998005e-05,
+      "loss": 0.6487,
+      "step": 6429
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3548135343870049,
+      "learning_rate": 1.6143649828915502e-05,
+      "loss": 0.6722,
+      "step": 6430
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.419474422043759,
+      "learning_rate": 1.6137966658260438e-05,
+      "loss": 0.634,
+      "step": 6431
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3677208707093734,
+      "learning_rate": 1.6132283811509447e-05,
+      "loss": 0.6857,
+      "step": 6432
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.36697928112208605,
+      "learning_rate": 1.6126601289139133e-05,
+      "loss": 0.6477,
+      "step": 6433
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.672953013548011,
+      "learning_rate": 1.61209190916261e-05,
+      "loss": 7.0325,
+      "step": 6434
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.37495521948712385,
+      "learning_rate": 1.6115237219446886e-05,
+      "loss": 0.6798,
+      "step": 6435
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3856176714340566,
+      "learning_rate": 1.6109555673078044e-05,
+      "loss": 0.6926,
+      "step": 6436
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39640078500846454,
+      "learning_rate": 1.6103874452996065e-05,
+      "loss": 0.6615,
+      "step": 6437
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.33309416711876555,
+      "learning_rate": 1.609819355967744e-05,
+      "loss": 0.6338,
+      "step": 6438
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3726206943371615,
+      "learning_rate": 1.60925129935986e-05,
+      "loss": 7.2355,
+      "step": 6439
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.41137159670936296,
+      "learning_rate": 1.6086832755235976e-05,
+      "loss": 0.6755,
+      "step": 6440
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4664853869613852,
+      "learning_rate": 1.6081152845065967e-05,
+      "loss": 0.6326,
+      "step": 6441
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4043606632150121,
+      "learning_rate": 1.607547326356493e-05,
+      "loss": 0.6777,
+      "step": 6442
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3944655428461289,
+      "learning_rate": 1.6069794011209217e-05,
+      "loss": 0.621,
+      "step": 6443
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.470475710986876,
+      "learning_rate": 1.606411508847513e-05,
+      "loss": 0.6989,
+      "step": 6444
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4467762729474391,
+      "learning_rate": 1.605843649583896e-05,
+      "loss": 0.6952,
+      "step": 6445
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3964554371683558,
+      "learning_rate": 1.6052758233776965e-05,
+      "loss": 0.6532,
+      "step": 6446
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.446794225273102,
+      "learning_rate": 1.604708030276536e-05,
+      "loss": 0.6783,
+      "step": 6447
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3538084641499618,
+      "learning_rate": 1.6041402703280364e-05,
+      "loss": 0.647,
+      "step": 6448
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4074727819411609,
+      "learning_rate": 1.6035725435798134e-05,
+      "loss": 0.6887,
+      "step": 6449
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4459049128145627,
+      "learning_rate": 1.6030048500794832e-05,
+      "loss": 0.6052,
+      "step": 6450
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3925366039297191,
+      "learning_rate": 1.602437189874656e-05,
+      "loss": 0.6265,
+      "step": 6451
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3550753421319945,
+      "learning_rate": 1.6018695630129425e-05,
+      "loss": 0.6683,
+      "step": 6452
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3674051803259855,
+      "learning_rate": 1.601301969541947e-05,
+      "loss": 0.6942,
+      "step": 6453
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4269994264597275,
+      "learning_rate": 1.600734409509275e-05,
+      "loss": 0.6779,
+      "step": 6454
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.42949720198618174,
+      "learning_rate": 1.6001668829625253e-05,
+      "loss": 0.6612,
+      "step": 6455
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3511563685347508,
+      "learning_rate": 1.5995993899492955e-05,
+      "loss": 0.6693,
+      "step": 6456
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3821013558804599,
+      "learning_rate": 1.599031930517182e-05,
+      "loss": 0.6395,
+      "step": 6457
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3724600095367016,
+      "learning_rate": 1.5984645047137762e-05,
+      "loss": 0.6641,
+      "step": 6458
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4285869426991325,
+      "learning_rate": 1.5978971125866675e-05,
+      "loss": 0.6675,
+      "step": 6459
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.379035489154202,
+      "learning_rate": 1.597329754183442e-05,
+      "loss": 0.6334,
+      "step": 6460
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3519873789587891,
+      "learning_rate": 1.5967624295516844e-05,
+      "loss": 0.6572,
+      "step": 6461
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3687915867419615,
+      "learning_rate": 1.5961951387389746e-05,
+      "loss": 0.6474,
+      "step": 6462
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4392605129764944,
+      "learning_rate": 1.59562788179289e-05,
+      "loss": 0.6776,
+      "step": 6463
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4162786927183854,
+      "learning_rate": 1.5950606587610072e-05,
+      "loss": 0.6858,
+      "step": 6464
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39763450531462585,
+      "learning_rate": 1.594493469690898e-05,
+      "loss": 0.6501,
+      "step": 6465
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39736721326405827,
+      "learning_rate": 1.593926314630132e-05,
+      "loss": 0.7182,
+      "step": 6466
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.36097525116279544,
+      "learning_rate": 1.5933591936262754e-05,
+      "loss": 0.6597,
+      "step": 6467
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4004739865603725,
+      "learning_rate": 1.5927921067268922e-05,
+      "loss": 0.675,
+      "step": 6468
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3930549748023967,
+      "learning_rate": 1.592225053979543e-05,
+      "loss": 0.625,
+      "step": 6469
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38294249232146943,
+      "learning_rate": 1.5916580354317857e-05,
+      "loss": 0.629,
+      "step": 6470
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3419980609030875,
+      "learning_rate": 1.591091051131176e-05,
+      "loss": 0.6397,
+      "step": 6471
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3675038825846377,
+      "learning_rate": 1.5905241011252667e-05,
+      "loss": 0.6443,
+      "step": 6472
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39736671705451426,
+      "learning_rate": 1.5899571854616064e-05,
+      "loss": 0.678,
+      "step": 6473
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.35827332676565826,
+      "learning_rate": 1.5893903041877413e-05,
+      "loss": 0.6662,
+      "step": 6474
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.32648993868007375,
+      "learning_rate": 1.588823457351216e-05,
+      "loss": 0.61,
+      "step": 6475
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.32368722945307227,
+      "learning_rate": 1.5882566449995702e-05,
+      "loss": 0.6592,
+      "step": 6476
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.35142301249060726,
+      "learning_rate": 1.587689867180342e-05,
+      "loss": 0.6393,
+      "step": 6477
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3994598079496711,
+      "learning_rate": 1.587123123941067e-05,
+      "loss": 0.6641,
+      "step": 6478
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3843612695948116,
+      "learning_rate": 1.5865564153292774e-05,
+      "loss": 0.6616,
+      "step": 6479
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38171797889227477,
+      "learning_rate": 1.5859897413925017e-05,
+      "loss": 0.6676,
+      "step": 6480
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.40989968651616654,
+      "learning_rate": 1.5854231021782665e-05,
+      "loss": 0.642,
+      "step": 6481
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3939213756478566,
+      "learning_rate": 1.584856497734095e-05,
+      "loss": 0.6693,
+      "step": 6482
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.2397292985530015,
+      "learning_rate": 1.584289928107507e-05,
+      "loss": 7.0207,
+      "step": 6483
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39855506832262555,
+      "learning_rate": 1.5837233933460217e-05,
+      "loss": 0.6483,
+      "step": 6484
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4041801900305552,
+      "learning_rate": 1.583156893497152e-05,
+      "loss": 0.6547,
+      "step": 6485
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.42637097297444926,
+      "learning_rate": 1.5825904286084105e-05,
+      "loss": 0.6894,
+      "step": 6486
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39079547025472394,
+      "learning_rate": 1.5820239987273056e-05,
+      "loss": 0.6655,
+      "step": 6487
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4031593828754524,
+      "learning_rate": 1.5814576039013433e-05,
+      "loss": 0.7139,
+      "step": 6488
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38001842466655356,
+      "learning_rate": 1.5808912441780266e-05,
+      "loss": 0.684,
+      "step": 6489
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3861754985563638,
+      "learning_rate": 1.580324919604854e-05,
+      "loss": 0.6583,
+      "step": 6490
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.33873050880679384,
+      "learning_rate": 1.579758630229324e-05,
+      "loss": 0.6515,
+      "step": 6491
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.40030206850262906,
+      "learning_rate": 1.5791923760989304e-05,
+      "loss": 0.69,
+      "step": 6492
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.40014104389222815,
+      "learning_rate": 1.578626157261164e-05,
+      "loss": 0.6606,
+      "step": 6493
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3886898945622587,
+      "learning_rate": 1.578059973763513e-05,
+      "loss": 0.7041,
+      "step": 6494
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.31662777452754476,
+      "learning_rate": 1.5774938256534625e-05,
+      "loss": 0.6556,
+      "step": 6495
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3458054459718706,
+      "learning_rate": 1.5769277129784945e-05,
+      "loss": 0.6178,
+      "step": 6496
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38242676641645096,
+      "learning_rate": 1.5763616357860876e-05,
+      "loss": 0.6352,
+      "step": 6497
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4319741928332459,
+      "learning_rate": 1.575795594123719e-05,
+      "loss": 0.6894,
+      "step": 6498
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.359710365883035,
+      "learning_rate": 1.575229588038862e-05,
+      "loss": 0.6653,
+      "step": 6499
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3813719906081777,
+      "learning_rate": 1.5746636175789868e-05,
+      "loss": 0.6704,
+      "step": 6500
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3621173058642799,
+      "learning_rate": 1.5740976827915595e-05,
+      "loss": 0.6189,
+      "step": 6501
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39360487929673577,
+      "learning_rate": 1.573531783724046e-05,
+      "loss": 0.6173,
+      "step": 6502
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4082390936129999,
+      "learning_rate": 1.572965920423906e-05,
+      "loss": 0.6802,
+      "step": 6503
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4281811219658629,
+      "learning_rate": 1.5724000929385984e-05,
+      "loss": 0.652,
+      "step": 6504
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.37732115913117503,
+      "learning_rate": 1.5718343013155786e-05,
+      "loss": 0.6454,
+      "step": 6505
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3934026426207048,
+      "learning_rate": 1.5712685456022995e-05,
+      "loss": 0.6534,
+      "step": 6506
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 3.683804511711949,
+      "learning_rate": 1.57070282584621e-05,
+      "loss": 7.019,
+      "step": 6507
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4037001808593296,
+      "learning_rate": 1.570137142094755e-05,
+      "loss": 0.7231,
+      "step": 6508
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.415134179169504,
+      "learning_rate": 1.5695714943953796e-05,
+      "loss": 0.6212,
+      "step": 6509
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3583188328089315,
+      "learning_rate": 1.5690058827955227e-05,
+      "loss": 0.6759,
+      "step": 6510
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.401309244319307,
+      "learning_rate": 1.5684403073426217e-05,
+      "loss": 0.6709,
+      "step": 6511
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4160944243850799,
+      "learning_rate": 1.567874768084111e-05,
+      "loss": 0.6473,
+      "step": 6512
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3422986070280779,
+      "learning_rate": 1.567309265067422e-05,
+      "loss": 0.6428,
+      "step": 6513
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3685034262972861,
+      "learning_rate": 1.566743798339982e-05,
+      "loss": 0.6531,
+      "step": 6514
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3586598679427419,
+      "learning_rate": 1.5661783679492168e-05,
+      "loss": 0.6543,
+      "step": 6515
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.41914047191444326,
+      "learning_rate": 1.565612973942548e-05,
+      "loss": 0.6351,
+      "step": 6516
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.41323240405967326,
+      "learning_rate": 1.565047616367394e-05,
+      "loss": 0.6977,
+      "step": 6517
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.37041397515043173,
+      "learning_rate": 1.564482295271171e-05,
+      "loss": 0.6859,
+      "step": 6518
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.409400628078884,
+      "learning_rate": 1.5639170107012922e-05,
+      "loss": 0.6529,
+      "step": 6519
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3836754160389019,
+      "learning_rate": 1.5633517627051672e-05,
+      "loss": 0.6497,
+      "step": 6520
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.43399038062479767,
+      "learning_rate": 1.5627865513302024e-05,
+      "loss": 0.7071,
+      "step": 6521
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3979917679512521,
+      "learning_rate": 1.5622213766238018e-05,
+      "loss": 0.6935,
+      "step": 6522
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.36807525351826653,
+      "learning_rate": 1.5616562386333657e-05,
+      "loss": 0.6794,
+      "step": 6523
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4186093600697934,
+      "learning_rate": 1.561091137406291e-05,
+      "loss": 0.6484,
+      "step": 6524
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.36992721858156424,
+      "learning_rate": 1.560526072989972e-05,
+      "loss": 0.6323,
+      "step": 6525
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.63567342374351,
+      "learning_rate": 1.5599610454318017e-05,
+      "loss": 7.1374,
+      "step": 6526
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3536841562421173,
+      "learning_rate": 1.5593960547791668e-05,
+      "loss": 0.6235,
+      "step": 6527
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3456967251404736,
+      "learning_rate": 1.558831101079452e-05,
+      "loss": 0.663,
+      "step": 6528
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.357810927779728,
+      "learning_rate": 1.5582661843800408e-05,
+      "loss": 0.6638,
+      "step": 6529
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.31628947357935405,
+      "learning_rate": 1.5577013047283113e-05,
+      "loss": 0.6495,
+      "step": 6530
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3505878432267866,
+      "learning_rate": 1.5571364621716386e-05,
+      "loss": 0.6608,
+      "step": 6531
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4033262674110464,
+      "learning_rate": 1.5565716567573955e-05,
+      "loss": 0.6497,
+      "step": 6532
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3831703530007854,
+      "learning_rate": 1.556006888532953e-05,
+      "loss": 0.6453,
+      "step": 6533
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38656721419083956,
+      "learning_rate": 1.5554421575456765e-05,
+      "loss": 0.6593,
+      "step": 6534
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3821926175227122,
+      "learning_rate": 1.5548774638429286e-05,
+      "loss": 0.6321,
+      "step": 6535
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.43393071149500734,
+      "learning_rate": 1.554312807472071e-05,
+      "loss": 0.6738,
+      "step": 6536
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38704128559100115,
+      "learning_rate": 1.5537481884804602e-05,
+      "loss": 0.6624,
+      "step": 6537
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39540957275927463,
+      "learning_rate": 1.553183606915449e-05,
+      "loss": 0.7282,
+      "step": 6538
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3680404990950684,
+      "learning_rate": 1.552619062824389e-05,
+      "loss": 0.6842,
+      "step": 6539
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.36000541717681095,
+      "learning_rate": 1.5520545562546284e-05,
+      "loss": 0.673,
+      "step": 6540
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.40266249245983876,
+      "learning_rate": 1.5514900872535117e-05,
+      "loss": 0.6925,
+      "step": 6541
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.37295322823165705,
+      "learning_rate": 1.550925655868379e-05,
+      "loss": 0.6563,
+      "step": 6542
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.40779719537161935,
+      "learning_rate": 1.5503612621465696e-05,
+      "loss": 0.6685,
+      "step": 6543
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3843128445743332,
+      "learning_rate": 1.5497969061354176e-05,
+      "loss": 0.6579,
+      "step": 6544
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.37205227152960907,
+      "learning_rate": 1.5492325878822557e-05,
+      "loss": 0.6485,
+      "step": 6545
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3434358868710379,
+      "learning_rate": 1.548668307434412e-05,
+      "loss": 0.6287,
+      "step": 6546
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.35890308924332537,
+      "learning_rate": 1.5481040648392126e-05,
+      "loss": 0.6824,
+      "step": 6547
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3551997869328371,
+      "learning_rate": 1.5475398601439792e-05,
+      "loss": 0.6419,
+      "step": 6548
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.353885453412661,
+      "learning_rate": 1.5469756933960322e-05,
+      "loss": 0.6873,
+      "step": 6549
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.37832182517410007,
+      "learning_rate": 1.546411564642686e-05,
+      "loss": 0.6609,
+      "step": 6550
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4045709061772316,
+      "learning_rate": 1.545847473931254e-05,
+      "loss": 0.7198,
+      "step": 6551
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.36386682445431484,
+      "learning_rate": 1.5452834213090462e-05,
+      "loss": 0.6515,
+      "step": 6552
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3946592978499596,
+      "learning_rate": 1.5447194068233678e-05,
+      "loss": 0.6785,
+      "step": 6553
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.36748018861226894,
+      "learning_rate": 1.544155430521524e-05,
+      "loss": 0.7026,
+      "step": 6554
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.36770614770141347,
+      "learning_rate": 1.543591492450813e-05,
+      "loss": 0.6468,
+      "step": 6555
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3470075504593301,
+      "learning_rate": 1.5430275926585328e-05,
+      "loss": 0.6494,
+      "step": 6556
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4294110015800666,
+      "learning_rate": 1.5424637311919767e-05,
+      "loss": 0.6931,
+      "step": 6557
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3531585568853318,
+      "learning_rate": 1.5418999080984342e-05,
+      "loss": 0.6665,
+      "step": 6558
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3535404256428967,
+      "learning_rate": 1.541336123425194e-05,
+      "loss": 0.6329,
+      "step": 6559
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3876326133759575,
+      "learning_rate": 1.540772377219538e-05,
+      "loss": 0.7038,
+      "step": 6560
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3692479824459234,
+      "learning_rate": 1.5402086695287494e-05,
+      "loss": 0.6179,
+      "step": 6561
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.36925367897396816,
+      "learning_rate": 1.5396450004001032e-05,
+      "loss": 0.6888,
+      "step": 6562
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3850348233339253,
+      "learning_rate": 1.539081369880876e-05,
+      "loss": 0.6647,
+      "step": 6563
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3895753644890491,
+      "learning_rate": 1.5385177780183372e-05,
+      "loss": 0.6453,
+      "step": 6564
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.314131444132162,
+      "learning_rate": 1.537954224859755e-05,
+      "loss": 0.6297,
+      "step": 6565
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.35802781406977197,
+      "learning_rate": 1.5373907104523935e-05,
+      "loss": 0.6489,
+      "step": 6566
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.36728638548418185,
+      "learning_rate": 1.5368272348435148e-05,
+      "loss": 0.6605,
+      "step": 6567
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3899867492682113,
+      "learning_rate": 1.536263798080377e-05,
+      "loss": 0.6821,
+      "step": 6568
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3751291654657991,
+      "learning_rate": 1.535700400210234e-05,
+      "loss": 0.6642,
+      "step": 6569
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.37189478987022767,
+      "learning_rate": 1.535137041280338e-05,
+      "loss": 0.6097,
+      "step": 6570
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.40438529073316143,
+      "learning_rate": 1.5345737213379362e-05,
+      "loss": 0.6656,
+      "step": 6571
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3815259600896756,
+      "learning_rate": 1.5340104404302748e-05,
+      "loss": 0.6623,
+      "step": 6572
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.40650936412510935,
+      "learning_rate": 1.5334471986045946e-05,
+      "loss": 0.6988,
+      "step": 6573
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3777125111805543,
+      "learning_rate": 1.5328839959081347e-05,
+      "loss": 0.6464,
+      "step": 6574
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.40752816629775146,
+      "learning_rate": 1.53232083238813e-05,
+      "loss": 0.6729,
+      "step": 6575
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3997292296968249,
+      "learning_rate": 1.5317577080918126e-05,
+      "loss": 0.6825,
+      "step": 6576
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3574393893056031,
+      "learning_rate": 1.5311946230664104e-05,
+      "loss": 0.6202,
+      "step": 6577
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.39091518006548903,
+      "learning_rate": 1.5306315773591488e-05,
+      "loss": 0.678,
+      "step": 6578
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3300087855818448,
+      "learning_rate": 1.5300685710172504e-05,
+      "loss": 0.6447,
+      "step": 6579
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4015886443357325,
+      "learning_rate": 1.5295056040879323e-05,
+      "loss": 0.6532,
+      "step": 6580
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.33965633166088177,
+      "learning_rate": 1.528942676618412e-05,
+      "loss": 0.6579,
+      "step": 6581
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4079702198119784,
+      "learning_rate": 1.5283797886559004e-05,
+      "loss": 0.6983,
+      "step": 6582
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.38096235884372986,
+      "learning_rate": 1.5278169402476064e-05,
+      "loss": 0.6478,
+      "step": 6583
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3988395970125079,
+      "learning_rate": 1.5272541314407353e-05,
+      "loss": 0.6533,
+      "step": 6584
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3810984082758671,
+      "learning_rate": 1.526691362282489e-05,
+      "loss": 0.635,
+      "step": 6585
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3894862330140788,
+      "learning_rate": 1.526128632820067e-05,
+      "loss": 0.6387,
+      "step": 6586
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3959721571449702,
+      "learning_rate": 1.5255659431006635e-05,
+      "loss": 0.6881,
+      "step": 6587
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3847050389041505,
+      "learning_rate": 1.5250032931714723e-05,
+      "loss": 0.6952,
+      "step": 6588
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.37012620773952853,
+      "learning_rate": 1.5244406830796805e-05,
+      "loss": 0.6667,
+      "step": 6589
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.438929085735723,
+      "learning_rate": 1.523878112872475e-05,
+      "loss": 0.7065,
+      "step": 6590
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.9540258849079377,
+      "learning_rate": 1.5233155825970369e-05,
+      "loss": 7.0184,
+      "step": 6591
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.37254304295352636,
+      "learning_rate": 1.522753092300545e-05,
+      "loss": 0.6859,
+      "step": 6592
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.396793787464638,
+      "learning_rate": 1.522190642030175e-05,
+      "loss": 0.6567,
+      "step": 6593
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.40976880200567967,
+      "learning_rate": 1.5216282318330983e-05,
+      "loss": 0.6884,
+      "step": 6594
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4067555229937303,
+      "learning_rate": 1.5210658617564846e-05,
+      "loss": 0.6841,
+      "step": 6595
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5390954657089868,
+      "learning_rate": 1.5205035318474986e-05,
+      "loss": 6.8798,
+      "step": 6596
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3949306882306112,
+      "learning_rate": 1.5199412421533026e-05,
+      "loss": 0.6409,
+      "step": 6597
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4143629245714057,
+      "learning_rate": 1.5193789927210547e-05,
+      "loss": 0.6638,
+      "step": 6598
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.35536904883207476,
+      "learning_rate": 1.5188167835979099e-05,
+      "loss": 0.6443,
+      "step": 6599
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.33035447846834093,
+      "learning_rate": 1.5182546148310209e-05,
+      "loss": 0.6707,
+      "step": 6600
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.32508854159459505,
+      "learning_rate": 1.5176924864675343e-05,
+      "loss": 0.6755,
+      "step": 6601
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.40487606546390753,
+      "learning_rate": 1.5171303985545978e-05,
+      "loss": 0.6583,
+      "step": 6602
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.579713517238499,
+      "learning_rate": 1.5165683511393507e-05,
+      "loss": 6.8631,
+      "step": 6603
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.38872689182434794,
+      "learning_rate": 1.5160063442689328e-05,
+      "loss": 0.6412,
+      "step": 6604
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3822308920728207,
+      "learning_rate": 1.515444377990478e-05,
+      "loss": 0.6353,
+      "step": 6605
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.37516621892399526,
+      "learning_rate": 1.5148824523511182e-05,
+      "loss": 0.6732,
+      "step": 6606
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3787511017717751,
+      "learning_rate": 1.5143205673979813e-05,
+      "loss": 0.6562,
+      "step": 6607
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.34110087291586516,
+      "learning_rate": 1.513758723178191e-05,
+      "loss": 0.6237,
+      "step": 6608
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3814916015265152,
+      "learning_rate": 1.5131969197388698e-05,
+      "loss": 0.6665,
+      "step": 6609
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.38059217693933023,
+      "learning_rate": 1.5126351571271357e-05,
+      "loss": 0.672,
+      "step": 6610
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.44352639472776234,
+      "learning_rate": 1.5120734353901021e-05,
+      "loss": 0.6995,
+      "step": 6611
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.37463408551619287,
+      "learning_rate": 1.5115117545748797e-05,
+      "loss": 0.6418,
+      "step": 6612
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3734511230455771,
+      "learning_rate": 1.5109501147285773e-05,
+      "loss": 0.6896,
+      "step": 6613
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3326961584852661,
+      "learning_rate": 1.5103885158982978e-05,
+      "loss": 0.6755,
+      "step": 6614
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3639060119315376,
+      "learning_rate": 1.5098269581311416e-05,
+      "loss": 0.6478,
+      "step": 6615
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.34053734323805795,
+      "learning_rate": 1.5092654414742066e-05,
+      "loss": 0.6242,
+      "step": 6616
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3702502102803854,
+      "learning_rate": 1.5087039659745868e-05,
+      "loss": 0.6369,
+      "step": 6617
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3423414999884656,
+      "learning_rate": 1.5081425316793722e-05,
+      "loss": 0.6435,
+      "step": 6618
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.41479633511025943,
+      "learning_rate": 1.507581138635649e-05,
+      "loss": 0.6701,
+      "step": 6619
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4359976404288786,
+      "learning_rate": 1.5070197868905014e-05,
+      "loss": 0.7112,
+      "step": 6620
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.35794094891142897,
+      "learning_rate": 1.5064584764910089e-05,
+      "loss": 0.6757,
+      "step": 6621
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.34541447625504884,
+      "learning_rate": 1.505897207484247e-05,
+      "loss": 0.648,
+      "step": 6622
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.34950301673505674,
+      "learning_rate": 1.5053359799172902e-05,
+      "loss": 0.6534,
+      "step": 6623
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3827144780123269,
+      "learning_rate": 1.5047747938372078e-05,
+      "loss": 0.6507,
+      "step": 6624
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.38454160924512937,
+      "learning_rate": 1.504213649291065e-05,
+      "loss": 0.6517,
+      "step": 6625
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4239700771118604,
+      "learning_rate": 1.5036525463259247e-05,
+      "loss": 0.6887,
+      "step": 6626
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.5773143012931787,
+      "learning_rate": 1.503091484988846e-05,
+      "loss": 6.9651,
+      "step": 6627
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4117834453278801,
+      "learning_rate": 1.5025304653268842e-05,
+      "loss": 0.6865,
+      "step": 6628
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3764518935542538,
+      "learning_rate": 1.5019694873870912e-05,
+      "loss": 0.6561,
+      "step": 6629
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.35306985767326465,
+      "learning_rate": 1.501408551216516e-05,
+      "loss": 0.6318,
+      "step": 6630
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3912887381533654,
+      "learning_rate": 1.5008476568622038e-05,
+      "loss": 0.6755,
+      "step": 6631
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3917687112158564,
+      "learning_rate": 1.5002868043711956e-05,
+      "loss": 0.7021,
+      "step": 6632
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3798747545323898,
+      "learning_rate": 1.4997259937905297e-05,
+      "loss": 0.6513,
+      "step": 6633
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4067347444215427,
+      "learning_rate": 1.4991652251672406e-05,
+      "loss": 0.658,
+      "step": 6634
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3227793102753424,
+      "learning_rate": 1.498604498548359e-05,
+      "loss": 0.6032,
+      "step": 6635
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.43728609365076815,
+      "learning_rate": 1.4980438139809119e-05,
+      "loss": 0.6945,
+      "step": 6636
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3852474704504083,
+      "learning_rate": 1.4974831715119249e-05,
+      "loss": 0.6549,
+      "step": 6637
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.38808805467975166,
+      "learning_rate": 1.4969225711884174e-05,
+      "loss": 0.6785,
+      "step": 6638
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3659409505359935,
+      "learning_rate": 1.4963620130574059e-05,
+      "loss": 0.6195,
+      "step": 6639
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 3.923224792274254,
+      "learning_rate": 1.4958014971659046e-05,
+      "loss": 7.1573,
+      "step": 6640
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3544871135764454,
+      "learning_rate": 1.4952410235609228e-05,
+      "loss": 0.6772,
+      "step": 6641
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3690788552340987,
+      "learning_rate": 1.494680592289466e-05,
+      "loss": 0.6611,
+      "step": 6642
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.36653951507800664,
+      "learning_rate": 1.4941202033985373e-05,
+      "loss": 0.6481,
+      "step": 6643
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.40073752567382664,
+      "learning_rate": 1.4935598569351376e-05,
+      "loss": 0.6648,
+      "step": 6644
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4306667659599583,
+      "learning_rate": 1.4929995529462607e-05,
+      "loss": 0.6564,
+      "step": 6645
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.355141727354634,
+      "learning_rate": 1.4924392914788985e-05,
+      "loss": 0.6792,
+      "step": 6646
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3352355478402791,
+      "learning_rate": 1.4918790725800402e-05,
+      "loss": 0.6467,
+      "step": 6647
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3861089849686773,
+      "learning_rate": 1.491318896296671e-05,
+      "loss": 0.6551,
+      "step": 6648
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.35634035899645594,
+      "learning_rate": 1.4907587626757707e-05,
+      "loss": 0.6699,
+      "step": 6649
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3685985383004805,
+      "learning_rate": 1.4901986717643178e-05,
+      "loss": 0.6699,
+      "step": 6650
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3511218371764044,
+      "learning_rate": 1.4896386236092872e-05,
+      "loss": 0.6716,
+      "step": 6651
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.39630969479953587,
+      "learning_rate": 1.4890786182576493e-05,
+      "loss": 0.6794,
+      "step": 6652
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.377831351025697,
+      "learning_rate": 1.4885186557563698e-05,
+      "loss": 0.6818,
+      "step": 6653
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3503427886360221,
+      "learning_rate": 1.4879587361524136e-05,
+      "loss": 0.6371,
+      "step": 6654
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.33946730362858646,
+      "learning_rate": 1.4873988594927398e-05,
+      "loss": 0.6263,
+      "step": 6655
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.41781377190687347,
+      "learning_rate": 1.4868390258243035e-05,
+      "loss": 0.6576,
+      "step": 6656
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39787491283282683,
+      "learning_rate": 1.4862792351940591e-05,
+      "loss": 0.6766,
+      "step": 6657
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.38902533130099887,
+      "learning_rate": 1.485719487648955e-05,
+      "loss": 0.7142,
+      "step": 6658
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3699397124891859,
+      "learning_rate": 1.4851597832359366e-05,
+      "loss": 0.6742,
+      "step": 6659
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.33736452780962983,
+      "learning_rate": 1.484600122001945e-05,
+      "loss": 0.6344,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35127825176550453,
+      "learning_rate": 1.4840405039939189e-05,
+      "loss": 0.6489,
+      "step": 6661
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3457527499640352,
+      "learning_rate": 1.4834809292587923e-05,
+      "loss": 0.6703,
+      "step": 6662
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.958561542503577,
+      "learning_rate": 1.4829213978434965e-05,
+      "loss": 7.0305,
+      "step": 6663
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.34255357376942747,
+      "learning_rate": 1.4823619097949584e-05,
+      "loss": 0.6396,
+      "step": 6664
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.32582314991523903,
+      "learning_rate": 1.4818024651601022e-05,
+      "loss": 0.652,
+      "step": 6665
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39895842372388335,
+      "learning_rate": 1.4812430639858471e-05,
+      "loss": 0.6591,
+      "step": 6666
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3871202165936758,
+      "learning_rate": 1.4806837063191102e-05,
+      "loss": 0.6694,
+      "step": 6667
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.43892300166060877,
+      "learning_rate": 1.4801243922068036e-05,
+      "loss": 0.7187,
+      "step": 6668
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3773678971430715,
+      "learning_rate": 1.479565121695836e-05,
+      "loss": 0.6307,
+      "step": 6669
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3684760972595106,
+      "learning_rate": 1.4790058948331124e-05,
+      "loss": 0.6794,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35980043646626575,
+      "learning_rate": 1.4784467116655365e-05,
+      "loss": 0.663,
+      "step": 6671
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.33979751579694384,
+      "learning_rate": 1.4778875722400048e-05,
+      "loss": 0.6696,
+      "step": 6672
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.33213003789079915,
+      "learning_rate": 1.4773284766034115e-05,
+      "loss": 0.64,
+      "step": 6673
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3374587605401341,
+      "learning_rate": 1.4767694248026478e-05,
+      "loss": 0.626,
+      "step": 6674
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35137790172763606,
+      "learning_rate": 1.476210416884601e-05,
+      "loss": 0.6248,
+      "step": 6675
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4286965948764432,
+      "learning_rate": 1.4756514528961532e-05,
+      "loss": 0.6534,
+      "step": 6676
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3730252214205873,
+      "learning_rate": 1.4750925328841846e-05,
+      "loss": 0.6682,
+      "step": 6677
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3317168357204729,
+      "learning_rate": 1.474533656895572e-05,
+      "loss": 0.651,
+      "step": 6678
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3507585272385673,
+      "learning_rate": 1.4739748249771868e-05,
+      "loss": 0.6321,
+      "step": 6679
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.36874614232939207,
+      "learning_rate": 1.4734160371758979e-05,
+      "loss": 0.6433,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.37511497470121286,
+      "learning_rate": 1.4728572935385699e-05,
+      "loss": 0.6971,
+      "step": 6681
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3787971651776996,
+      "learning_rate": 1.4722985941120644e-05,
+      "loss": 0.6993,
+      "step": 6682
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.33023162379681675,
+      "learning_rate": 1.4717399389432379e-05,
+      "loss": 0.6397,
+      "step": 6683
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.40618874243729725,
+      "learning_rate": 1.4711813280789443e-05,
+      "loss": 0.6533,
+      "step": 6684
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.37081115769225925,
+      "learning_rate": 1.4706227615660348e-05,
+      "loss": 0.6359,
+      "step": 6685
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3628108616974634,
+      "learning_rate": 1.4700642394513553e-05,
+      "loss": 0.6743,
+      "step": 6686
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.375956139658121,
+      "learning_rate": 1.4695057617817475e-05,
+      "loss": 0.6705,
+      "step": 6687
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3712364225658256,
+      "learning_rate": 1.4689473286040512e-05,
+      "loss": 0.6661,
+      "step": 6688
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.36545546837927434,
+      "learning_rate": 1.4683889399651005e-05,
+      "loss": 0.6707,
+      "step": 6689
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35390382503569456,
+      "learning_rate": 1.467830595911728e-05,
+      "loss": 0.6479,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.32630614968273647,
+      "learning_rate": 1.46727229649076e-05,
+      "loss": 0.6349,
+      "step": 6691
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.34389020727434355,
+      "learning_rate": 1.4667140417490217e-05,
+      "loss": 0.6359,
+      "step": 6692
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3824611173688045,
+      "learning_rate": 1.4661558317333322e-05,
+      "loss": 0.6635,
+      "step": 6693
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.38400694795680196,
+      "learning_rate": 1.465597666490509e-05,
+      "loss": 0.648,
+      "step": 6694
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4203468587522839,
+      "learning_rate": 1.4650395460673638e-05,
+      "loss": 0.6708,
+      "step": 6695
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.38770450119370103,
+      "learning_rate": 1.4644814705107058e-05,
+      "loss": 0.6982,
+      "step": 6696
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3977830882322857,
+      "learning_rate": 1.4639234398673404e-05,
+      "loss": 0.6827,
+      "step": 6697
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.36095720817584465,
+      "learning_rate": 1.4633654541840679e-05,
+      "loss": 0.6307,
+      "step": 6698
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39581455721705305,
+      "learning_rate": 1.4628075135076874e-05,
+      "loss": 0.6572,
+      "step": 6699
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.37603493564577156,
+      "learning_rate": 1.4622496178849916e-05,
+      "loss": 0.6493,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.40422446794598915,
+      "learning_rate": 1.4616917673627712e-05,
+      "loss": 0.6305,
+      "step": 6701
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.396161228865743,
+      "learning_rate": 1.4611339619878125e-05,
+      "loss": 0.6517,
+      "step": 6702
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.3417240195672246,
+      "learning_rate": 1.4605762018068967e-05,
+      "loss": 6.9828,
+      "step": 6703
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4165963612238559,
+      "learning_rate": 1.4600184868668041e-05,
+      "loss": 0.679,
+      "step": 6704
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4003275280701402,
+      "learning_rate": 1.4594608172143081e-05,
+      "loss": 0.6767,
+      "step": 6705
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39461857705886566,
+      "learning_rate": 1.4589031928961813e-05,
+      "loss": 0.6671,
+      "step": 6706
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3578334063049591,
+      "learning_rate": 1.4583456139591899e-05,
+      "loss": 0.6115,
+      "step": 6707
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3796782396473668,
+      "learning_rate": 1.4577880804500982e-05,
+      "loss": 0.673,
+      "step": 6708
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4354886252913732,
+      "learning_rate": 1.457230592415665e-05,
+      "loss": 0.6582,
+      "step": 6709
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39434489982930576,
+      "learning_rate": 1.4566731499026463e-05,
+      "loss": 0.6905,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4638552282335633,
+      "learning_rate": 1.456115752957795e-05,
+      "loss": 6.9124,
+      "step": 6711
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3789113060595669,
+      "learning_rate": 1.4555584016278575e-05,
+      "loss": 0.6385,
+      "step": 6712
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4020631171582721,
+      "learning_rate": 1.4550010959595801e-05,
+      "loss": 0.6834,
+      "step": 6713
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.40036517767070395,
+      "learning_rate": 1.4544438359997024e-05,
+      "loss": 0.6603,
+      "step": 6714
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.40141936387096927,
+      "learning_rate": 1.4538866217949617e-05,
+      "loss": 0.6576,
+      "step": 6715
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4283354363818659,
+      "learning_rate": 1.4533294533920903e-05,
+      "loss": 0.7434,
+      "step": 6716
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39724160131634423,
+      "learning_rate": 1.4527723308378172e-05,
+      "loss": 0.6983,
+      "step": 6717
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3730902851567248,
+      "learning_rate": 1.4522152541788684e-05,
+      "loss": 0.643,
+      "step": 6718
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4014723262203152,
+      "learning_rate": 1.4516582234619636e-05,
+      "loss": 0.6913,
+      "step": 6719
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3810994365612978,
+      "learning_rate": 1.4511012387338221e-05,
+      "loss": 0.6564,
+      "step": 6720
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.40336269131899527,
+      "learning_rate": 1.4505443000411568e-05,
+      "loss": 0.6348,
+      "step": 6721
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39077647408555516,
+      "learning_rate": 1.4499874074306774e-05,
+      "loss": 0.6559,
+      "step": 6722
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3099832196479517,
+      "learning_rate": 1.44943056094909e-05,
+      "loss": 0.6321,
+      "step": 6723
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35140971322094305,
+      "learning_rate": 1.4488737606430965e-05,
+      "loss": 0.6707,
+      "step": 6724
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.6689595015696896,
+      "learning_rate": 1.4483170065593954e-05,
+      "loss": 7.0947,
+      "step": 6725
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3728636802855669,
+      "learning_rate": 1.4477602987446797e-05,
+      "loss": 0.6586,
+      "step": 6726
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3964642453664437,
+      "learning_rate": 1.4472036372456413e-05,
+      "loss": 0.6885,
+      "step": 6727
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.383763582892126,
+      "learning_rate": 1.4466470221089666e-05,
+      "loss": 0.6847,
+      "step": 6728
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3598313228952519,
+      "learning_rate": 1.4460904533813379e-05,
+      "loss": 0.6796,
+      "step": 6729
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3888501270825747,
+      "learning_rate": 1.4455339311094334e-05,
+      "loss": 0.6825,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3344248383215474,
+      "learning_rate": 1.4449774553399292e-05,
+      "loss": 0.6333,
+      "step": 6731
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.38570789877673023,
+      "learning_rate": 1.4444210261194953e-05,
+      "loss": 0.6961,
+      "step": 6732
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35949688952076375,
+      "learning_rate": 1.4438646434947981e-05,
+      "loss": 0.6396,
+      "step": 6733
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.38177282343910435,
+      "learning_rate": 1.443308307512502e-05,
+      "loss": 0.6533,
+      "step": 6734
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35993475065814573,
+      "learning_rate": 1.4427520182192662e-05,
+      "loss": 0.6318,
+      "step": 6735
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.32202418719283976,
+      "learning_rate": 1.442195775661746e-05,
+      "loss": 0.6649,
+      "step": 6736
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.31311847464881953,
+      "learning_rate": 1.4416395798865917e-05,
+      "loss": 0.6254,
+      "step": 6737
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35747576655291463,
+      "learning_rate": 1.441083430940452e-05,
+      "loss": 0.6565,
+      "step": 6738
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3346984152071661,
+      "learning_rate": 1.4405273288699691e-05,
+      "loss": 0.6372,
+      "step": 6739
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3770615164390653,
+      "learning_rate": 1.4399712737217843e-05,
+      "loss": 0.7058,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3783058910484402,
+      "learning_rate": 1.439415265542532e-05,
+      "loss": 0.6617,
+      "step": 6741
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.36168052163902664,
+      "learning_rate": 1.4388593043788448e-05,
+      "loss": 0.6456,
+      "step": 6742
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39155663427581366,
+      "learning_rate": 1.4383033902773502e-05,
+      "loss": 0.6524,
+      "step": 6743
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3762983989853885,
+      "learning_rate": 1.4377475232846714e-05,
+      "loss": 0.6612,
+      "step": 6744
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4027619630347293,
+      "learning_rate": 1.4371917034474292e-05,
+      "loss": 0.64,
+      "step": 6745
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.37130196825147554,
+      "learning_rate": 1.4366359308122381e-05,
+      "loss": 0.6606,
+      "step": 6746
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3606052631576071,
+      "learning_rate": 1.4360802054257121e-05,
+      "loss": 0.6285,
+      "step": 6747
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.40504501789731634,
+      "learning_rate": 1.435524527334458e-05,
+      "loss": 0.6824,
+      "step": 6748
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3964461500362126,
+      "learning_rate": 1.43496889658508e-05,
+      "loss": 0.6676,
+      "step": 6749
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3408526271031842,
+      "learning_rate": 1.4344133132241785e-05,
+      "loss": 0.6441,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3880990552893353,
+      "learning_rate": 1.4338577772983493e-05,
+      "loss": 0.661,
+      "step": 6751
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3761039832052203,
+      "learning_rate": 1.4333022888541848e-05,
+      "loss": 0.67,
+      "step": 6752
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3738717220127726,
+      "learning_rate": 1.432746847938272e-05,
+      "loss": 0.6196,
+      "step": 6753
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4057928898616426,
+      "learning_rate": 1.4321914545971965e-05,
+      "loss": 0.6771,
+      "step": 6754
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.41875923902854184,
+      "learning_rate": 1.4316361088775384e-05,
+      "loss": 0.6431,
+      "step": 6755
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3801283051695368,
+      "learning_rate": 1.4310808108258731e-05,
+      "loss": 0.6833,
+      "step": 6756
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39401464190712415,
+      "learning_rate": 1.430525560488773e-05,
+      "loss": 0.6544,
+      "step": 6757
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35555913140749884,
+      "learning_rate": 1.4299703579128069e-05,
+      "loss": 0.6275,
+      "step": 6758
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.37950546855964196,
+      "learning_rate": 1.4294152031445383e-05,
+      "loss": 0.6464,
+      "step": 6759
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.361327990453845,
+      "learning_rate": 1.4288600962305265e-05,
+      "loss": 0.6523,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39356857457139605,
+      "learning_rate": 1.4283050372173292e-05,
+      "loss": 0.6638,
+      "step": 6761
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.40175634357179874,
+      "learning_rate": 1.4277500261514987e-05,
+      "loss": 0.6405,
+      "step": 6762
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.36263460919266804,
+      "learning_rate": 1.4271950630795824e-05,
+      "loss": 0.6514,
+      "step": 6763
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3468925145575112,
+      "learning_rate": 1.4266401480481236e-05,
+      "loss": 0.635,
+      "step": 6764
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3702832704458925,
+      "learning_rate": 1.4260852811036637e-05,
+      "loss": 0.7146,
+      "step": 6765
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4047825329874478,
+      "learning_rate": 1.4255304622927382e-05,
+      "loss": 0.6855,
+      "step": 6766
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3629362502464663,
+      "learning_rate": 1.4249756916618782e-05,
+      "loss": 0.6404,
+      "step": 6767
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4318947984921032,
+      "learning_rate": 1.424420969257613e-05,
+      "loss": 0.689,
+      "step": 6768
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4407147245991677,
+      "learning_rate": 1.4238662951264664e-05,
+      "loss": 0.6659,
+      "step": 6769
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.36266301390470324,
+      "learning_rate": 1.4233116693149578e-05,
+      "loss": 0.6801,
+      "step": 6770
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3555815925428872,
+      "learning_rate": 1.4227570918696026e-05,
+      "loss": 0.6552,
+      "step": 6771
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3882880588145424,
+      "learning_rate": 1.422202562836913e-05,
+      "loss": 0.6557,
+      "step": 6772
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.36467031679925377,
+      "learning_rate": 1.421648082263397e-05,
+      "loss": 0.6513,
+      "step": 6773
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3799117669873661,
+      "learning_rate": 1.4210936501955572e-05,
+      "loss": 0.6363,
+      "step": 6774
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.36308644242828436,
+      "learning_rate": 1.420539266679894e-05,
+      "loss": 0.6818,
+      "step": 6775
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.32836359678549076,
+      "learning_rate": 1.4199849317629034e-05,
+      "loss": 0.6334,
+      "step": 6776
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.40907419955953206,
+      "learning_rate": 1.4194306454910757e-05,
+      "loss": 0.6655,
+      "step": 6777
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3874807657245159,
+      "learning_rate": 1.4188764079108985e-05,
+      "loss": 0.7151,
+      "step": 6778
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3474989213017735,
+      "learning_rate": 1.4183222190688555e-05,
+      "loss": 0.6717,
+      "step": 6779
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3382752387512271,
+      "learning_rate": 1.4177680790114248e-05,
+      "loss": 0.6554,
+      "step": 6780
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.39890436860968875,
+      "learning_rate": 1.417213987785082e-05,
+      "loss": 0.6552,
+      "step": 6781
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3613348926010949,
+      "learning_rate": 1.416659945436299e-05,
+      "loss": 0.6373,
+      "step": 6782
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3799809329393534,
+      "learning_rate": 1.416105952011542e-05,
+      "loss": 0.6559,
+      "step": 6783
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.39461495097009935,
+      "learning_rate": 1.415552007557273e-05,
+      "loss": 0.6599,
+      "step": 6784
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.42768783431580576,
+      "learning_rate": 1.4149981121199518e-05,
+      "loss": 0.6561,
+      "step": 6785
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.42398130463560324,
+      "learning_rate": 1.4144442657460324e-05,
+      "loss": 0.6978,
+      "step": 6786
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3586154825491027,
+      "learning_rate": 1.413890468481965e-05,
+      "loss": 0.6813,
+      "step": 6787
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3762834549559129,
+      "learning_rate": 1.4133367203741957e-05,
+      "loss": 0.6458,
+      "step": 6788
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3918111798633796,
+      "learning_rate": 1.4127830214691677e-05,
+      "loss": 0.6744,
+      "step": 6789
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.43425453408923165,
+      "learning_rate": 1.4122293718133189e-05,
+      "loss": 0.6569,
+      "step": 6790
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.36340942426904543,
+      "learning_rate": 1.4116757714530822e-05,
+      "loss": 0.6602,
+      "step": 6791
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.33016706097632054,
+      "learning_rate": 1.4111222204348884e-05,
+      "loss": 0.6327,
+      "step": 6792
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3649109722653065,
+      "learning_rate": 1.4105687188051632e-05,
+      "loss": 0.6336,
+      "step": 6793
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4111453443516978,
+      "learning_rate": 1.4100152666103269e-05,
+      "loss": 0.6752,
+      "step": 6794
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.38959463202927225,
+      "learning_rate": 1.4094618638967972e-05,
+      "loss": 0.6894,
+      "step": 6795
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3861743292047527,
+      "learning_rate": 1.4089085107109887e-05,
+      "loss": 0.6784,
+      "step": 6796
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.1149593559551167,
+      "learning_rate": 1.4083552070993096e-05,
+      "loss": 6.9114,
+      "step": 6797
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.41221854610463643,
+      "learning_rate": 1.4078019531081645e-05,
+      "loss": 0.6367,
+      "step": 6798
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.38027818033596295,
+      "learning_rate": 1.4072487487839546e-05,
+      "loss": 0.6484,
+      "step": 6799
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.37120154604948297,
+      "learning_rate": 1.4066955941730763e-05,
+      "loss": 0.6475,
+      "step": 6800
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.39768203402087726,
+      "learning_rate": 1.4061424893219217e-05,
+      "loss": 0.6688,
+      "step": 6801
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.35624422461961397,
+      "learning_rate": 1.4055894342768788e-05,
+      "loss": 0.6572,
+      "step": 6802
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3836585913738409,
+      "learning_rate": 1.405036429084333e-05,
+      "loss": 0.6516,
+      "step": 6803
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.40339511363231706,
+      "learning_rate": 1.4044834737906634e-05,
+      "loss": 0.6383,
+      "step": 6804
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4356884611868329,
+      "learning_rate": 1.4039305684422454e-05,
+      "loss": 0.7047,
+      "step": 6805
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3393087861859333,
+      "learning_rate": 1.4033777130854506e-05,
+      "loss": 0.6343,
+      "step": 6806
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.35884543395134455,
+      "learning_rate": 1.4028249077666467e-05,
+      "loss": 0.6449,
+      "step": 6807
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.35383497833900557,
+      "learning_rate": 1.4022721525321968e-05,
+      "loss": 0.6412,
+      "step": 6808
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3453178773190052,
+      "learning_rate": 1.4017194474284585e-05,
+      "loss": 0.6622,
+      "step": 6809
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3541410749254515,
+      "learning_rate": 1.4011667925017886e-05,
+      "loss": 0.6488,
+      "step": 6810
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3767100673223085,
+      "learning_rate": 1.4006141877985364e-05,
+      "loss": 0.6766,
+      "step": 6811
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3758500207435623,
+      "learning_rate": 1.4000616333650486e-05,
+      "loss": 0.675,
+      "step": 6812
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3275723608882444,
+      "learning_rate": 1.399509129247667e-05,
+      "loss": 0.6348,
+      "step": 6813
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3923919607364785,
+      "learning_rate": 1.398956675492729e-05,
+      "loss": 0.6712,
+      "step": 6814
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.40544615462690375,
+      "learning_rate": 1.3984042721465697e-05,
+      "loss": 0.6253,
+      "step": 6815
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3417407874402936,
+      "learning_rate": 1.3978519192555165e-05,
+      "loss": 0.5988,
+      "step": 6816
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4215797111024664,
+      "learning_rate": 1.3972996168658962e-05,
+      "loss": 0.6724,
+      "step": 6817
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.3530320165496026,
+      "learning_rate": 1.396747365024029e-05,
+      "loss": 6.9403,
+      "step": 6818
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4388203311223485,
+      "learning_rate": 1.396195163776232e-05,
+      "loss": 0.6731,
+      "step": 6819
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3274372263382807,
+      "learning_rate": 1.3956430131688175e-05,
+      "loss": 0.6297,
+      "step": 6820
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3593256020000969,
+      "learning_rate": 1.395090913248093e-05,
+      "loss": 0.6512,
+      "step": 6821
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4071580651240683,
+      "learning_rate": 1.3945388640603628e-05,
+      "loss": 0.7064,
+      "step": 6822
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.35642651184859825,
+      "learning_rate": 1.3939868656519275e-05,
+      "loss": 0.6463,
+      "step": 6823
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.31217168481460705,
+      "learning_rate": 1.3934349180690822e-05,
+      "loss": 0.6106,
+      "step": 6824
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3622823814740339,
+      "learning_rate": 1.3928830213581172e-05,
+      "loss": 0.6593,
+      "step": 6825
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3587759923658144,
+      "learning_rate": 1.3923311755653206e-05,
+      "loss": 0.6541,
+      "step": 6826
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.38183499638833474,
+      "learning_rate": 1.3917793807369742e-05,
+      "loss": 0.6653,
+      "step": 6827
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.39366731931255283,
+      "learning_rate": 1.3912276369193562e-05,
+      "loss": 0.7009,
+      "step": 6828
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.37932201836501817,
+      "learning_rate": 1.3906759441587408e-05,
+      "loss": 0.6182,
+      "step": 6829
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.34409595585820296,
+      "learning_rate": 1.3901243025013988e-05,
+      "loss": 0.662,
+      "step": 6830
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.40947794239049007,
+      "learning_rate": 1.389572711993595e-05,
+      "loss": 0.699,
+      "step": 6831
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4140298576099916,
+      "learning_rate": 1.38902117268159e-05,
+      "loss": 0.6553,
+      "step": 6832
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.42137662717508845,
+      "learning_rate": 1.3884696846116424e-05,
+      "loss": 0.6562,
+      "step": 6833
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3435394285202764,
+      "learning_rate": 1.3879182478300029e-05,
+      "loss": 0.6517,
+      "step": 6834
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.35646083123972727,
+      "learning_rate": 1.3873668623829213e-05,
+      "loss": 0.6504,
+      "step": 6835
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3843352511446937,
+      "learning_rate": 1.3868155283166403e-05,
+      "loss": 0.6839,
+      "step": 6836
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.41934370449254155,
+      "learning_rate": 1.3862642456774011e-05,
+      "loss": 0.6415,
+      "step": 6837
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.43153338982584394,
+      "learning_rate": 1.3857130145114381e-05,
+      "loss": 0.6495,
+      "step": 6838
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3524823894625582,
+      "learning_rate": 1.385161834864983e-05,
+      "loss": 0.6567,
+      "step": 6839
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.41362848742806235,
+      "learning_rate": 1.3846107067842625e-05,
+      "loss": 0.6987,
+      "step": 6840
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4128468586295997,
+      "learning_rate": 1.3840596303154982e-05,
+      "loss": 0.6583,
+      "step": 6841
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4102203728104274,
+      "learning_rate": 1.3835086055049094e-05,
+      "loss": 0.6632,
+      "step": 6842
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3664903579754618,
+      "learning_rate": 1.3829576323987083e-05,
+      "loss": 0.6752,
+      "step": 6843
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.37835440377699336,
+      "learning_rate": 1.3824067110431063e-05,
+      "loss": 0.6046,
+      "step": 6844
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.381310515740385,
+      "learning_rate": 1.3818558414843073e-05,
+      "loss": 0.6312,
+      "step": 6845
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3827950794506423,
+      "learning_rate": 1.3813050237685126e-05,
+      "loss": 0.6752,
+      "step": 6846
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.34921933540400485,
+      "learning_rate": 1.3807542579419184e-05,
+      "loss": 0.631,
+      "step": 6847
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.39997820487312735,
+      "learning_rate": 1.3802035440507165e-05,
+      "loss": 0.6958,
+      "step": 6848
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.36201674133403683,
+      "learning_rate": 1.379652882141095e-05,
+      "loss": 0.6406,
+      "step": 6849
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4071967051555465,
+      "learning_rate": 1.3791022722592361e-05,
+      "loss": 0.6884,
+      "step": 6850
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.35488413261455753,
+      "learning_rate": 1.378551714451321e-05,
+      "loss": 0.6552,
+      "step": 6851
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3544154691138685,
+      "learning_rate": 1.3780012087635224e-05,
+      "loss": 0.667,
+      "step": 6852
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3900183667397683,
+      "learning_rate": 1.3774507552420118e-05,
+      "loss": 0.7015,
+      "step": 6853
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4035281456157255,
+      "learning_rate": 1.3769003539329547e-05,
+      "loss": 0.703,
+      "step": 6854
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.35043810014581456,
+      "learning_rate": 1.3763500048825117e-05,
+      "loss": 0.6573,
+      "step": 6855
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3500771582668352,
+      "learning_rate": 1.375799708136841e-05,
+      "loss": 0.6552,
+      "step": 6856
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.37676568802664484,
+      "learning_rate": 1.375249463742094e-05,
+      "loss": 0.6984,
+      "step": 6857
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.38042337565864426,
+      "learning_rate": 1.374699271744421e-05,
+      "loss": 0.6663,
+      "step": 6858
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.36076529344300845,
+      "learning_rate": 1.3741491321899645e-05,
+      "loss": 0.6391,
+      "step": 6859
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3964419638204065,
+      "learning_rate": 1.3735990451248649e-05,
+      "loss": 0.6635,
+      "step": 6860
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3888326752399007,
+      "learning_rate": 1.3730490105952567e-05,
+      "loss": 0.664,
+      "step": 6861
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3324887411889105,
+      "learning_rate": 1.3724990286472707e-05,
+      "loss": 0.6365,
+      "step": 6862
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.37835612698574067,
+      "learning_rate": 1.3719490993270335e-05,
+      "loss": 0.6455,
+      "step": 6863
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.37870940629369393,
+      "learning_rate": 1.371399222680666e-05,
+      "loss": 0.6418,
+      "step": 6864
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.40552131821934223,
+      "learning_rate": 1.3708493987542877e-05,
+      "loss": 0.6858,
+      "step": 6865
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.39405661710071366,
+      "learning_rate": 1.3702996275940097e-05,
+      "loss": 0.713,
+      "step": 6866
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3523133040158945,
+      "learning_rate": 1.3697499092459419e-05,
+      "loss": 0.6492,
+      "step": 6867
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3984822142418875,
+      "learning_rate": 1.369200243756188e-05,
+      "loss": 0.6914,
+      "step": 6868
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3666826664411502,
+      "learning_rate": 1.3686506311708479e-05,
+      "loss": 0.6369,
+      "step": 6869
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.34005022686052366,
+      "learning_rate": 1.368101071536017e-05,
+      "loss": 0.6672,
+      "step": 6870
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.4742601037810457,
+      "learning_rate": 1.3675515648977853e-05,
+      "loss": 6.8786,
+      "step": 6871
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3644767025399682,
+      "learning_rate": 1.3670021113022402e-05,
+      "loss": 0.6635,
+      "step": 6872
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3739955652651167,
+      "learning_rate": 1.3664527107954642e-05,
+      "loss": 0.6525,
+      "step": 6873
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.35583441267942917,
+      "learning_rate": 1.3659033634235343e-05,
+      "loss": 0.6863,
+      "step": 6874
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.40587730228370134,
+      "learning_rate": 1.3653540692325229e-05,
+      "loss": 0.6813,
+      "step": 6875
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3594190559934213,
+      "learning_rate": 1.3648048282685e-05,
+      "loss": 0.6414,
+      "step": 6876
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4163923600738022,
+      "learning_rate": 1.3642556405775287e-05,
+      "loss": 0.6522,
+      "step": 6877
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3893931847184246,
+      "learning_rate": 1.3637065062056683e-05,
+      "loss": 0.6212,
+      "step": 6878
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 5.313829367768047,
+      "learning_rate": 1.3631574251989753e-05,
+      "loss": 6.9675,
+      "step": 6879
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.42942083250787383,
+      "learning_rate": 1.3626083976035003e-05,
+      "loss": 0.6622,
+      "step": 6880
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3944694193657842,
+      "learning_rate": 1.362059423465289e-05,
+      "loss": 0.6483,
+      "step": 6881
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3933826406325723,
+      "learning_rate": 1.3615105028303833e-05,
+      "loss": 0.6783,
+      "step": 6882
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3892988366471257,
+      "learning_rate": 1.360961635744821e-05,
+      "loss": 0.6521,
+      "step": 6883
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.39794114034955175,
+      "learning_rate": 1.3604128222546344e-05,
+      "loss": 0.6633,
+      "step": 6884
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4004547403508019,
+      "learning_rate": 1.3598640624058512e-05,
+      "loss": 0.6409,
+      "step": 6885
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.34577787824277384,
+      "learning_rate": 1.3593153562444963e-05,
+      "loss": 0.6211,
+      "step": 6886
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3989654927230439,
+      "learning_rate": 1.3587667038165892e-05,
+      "loss": 0.6879,
+      "step": 6887
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.37453220871820436,
+      "learning_rate": 1.3582181051681444e-05,
+      "loss": 0.659,
+      "step": 6888
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.40967769486050215,
+      "learning_rate": 1.3576695603451718e-05,
+      "loss": 0.6436,
+      "step": 6889
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.39006762381069826,
+      "learning_rate": 1.3571210693936774e-05,
+      "loss": 0.6854,
+      "step": 6890
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.32792504882152546,
+      "learning_rate": 1.3565726323596622e-05,
+      "loss": 0.6282,
+      "step": 6891
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.36782114220633233,
+      "learning_rate": 1.3560242492891234e-05,
+      "loss": 0.6144,
+      "step": 6892
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4133742234122044,
+      "learning_rate": 1.355475920228053e-05,
+      "loss": 0.6474,
+      "step": 6893
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 7.1926633086621745,
+      "learning_rate": 1.354927645222439e-05,
+      "loss": 7.0703,
+      "step": 6894
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3877518861740398,
+      "learning_rate": 1.3543794243182643e-05,
+      "loss": 0.6819,
+      "step": 6895
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.41748286424808684,
+      "learning_rate": 1.3538312575615078e-05,
+      "loss": 0.6949,
+      "step": 6896
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3146232316306589,
+      "learning_rate": 1.3532831449981435e-05,
+      "loss": 0.6137,
+      "step": 6897
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3487900388593942,
+      "learning_rate": 1.3527350866741402e-05,
+      "loss": 0.5997,
+      "step": 6898
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3897920780363329,
+      "learning_rate": 1.3521870826354633e-05,
+      "loss": 0.658,
+      "step": 6899
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4190262907606781,
+      "learning_rate": 1.351639132928074e-05,
+      "loss": 0.6964,
+      "step": 6900
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3344512270022861,
+      "learning_rate": 1.3510912375979274e-05,
+      "loss": 0.6501,
+      "step": 6901
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.377806246121853,
+      "learning_rate": 1.350543396690975e-05,
+      "loss": 0.6093,
+      "step": 6902
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.40697269817157766,
+      "learning_rate": 1.349995610253164e-05,
+      "loss": 0.664,
+      "step": 6903
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3924100019753735,
+      "learning_rate": 1.3494478783304361e-05,
+      "loss": 0.6037,
+      "step": 6904
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3953511442194592,
+      "learning_rate": 1.348900200968728e-05,
+      "loss": 0.6543,
+      "step": 6905
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.35807969967278347,
+      "learning_rate": 1.3483525782139743e-05,
+      "loss": 0.6287,
+      "step": 6906
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.35637386094740453,
+      "learning_rate": 1.347805010112103e-05,
+      "loss": 0.6537,
+      "step": 6907
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3436902428040244,
+      "learning_rate": 1.3472574967090382e-05,
+      "loss": 0.6348,
+      "step": 6908
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.37942517241628637,
+      "learning_rate": 1.3467100380506982e-05,
+      "loss": 0.6399,
+      "step": 6909
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.38303792223960836,
+      "learning_rate": 1.3461626341829988e-05,
+      "loss": 0.6763,
+      "step": 6910
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3377763389157875,
+      "learning_rate": 1.3456152851518499e-05,
+      "loss": 0.6361,
+      "step": 6911
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4051004728998965,
+      "learning_rate": 1.3450679910031554e-05,
+      "loss": 0.6876,
+      "step": 6912
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3843857348823192,
+      "learning_rate": 1.3445207517828182e-05,
+      "loss": 0.6537,
+      "step": 6913
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3900244941542219,
+      "learning_rate": 1.3439735675367345e-05,
+      "loss": 0.6596,
+      "step": 6914
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.41216390891737603,
+      "learning_rate": 1.3434264383107953e-05,
+      "loss": 0.6597,
+      "step": 6915
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.39500252833019056,
+      "learning_rate": 1.3428793641508873e-05,
+      "loss": 0.6859,
+      "step": 6916
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3723762706188768,
+      "learning_rate": 1.3423323451028938e-05,
+      "loss": 0.664,
+      "step": 6917
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3572453756281266,
+      "learning_rate": 1.3417853812126925e-05,
+      "loss": 0.6749,
+      "step": 6918
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3597493835963981,
+      "learning_rate": 1.3412384725261553e-05,
+      "loss": 0.6449,
+      "step": 6919
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.42386393342178685,
+      "learning_rate": 1.3406916190891522e-05,
+      "loss": 0.7058,
+      "step": 6920
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.36049804917892875,
+      "learning_rate": 1.3401448209475474e-05,
+      "loss": 0.6544,
+      "step": 6921
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.41836803189949734,
+      "learning_rate": 1.3395980781471996e-05,
+      "loss": 0.6802,
+      "step": 6922
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.34749715140327353,
+      "learning_rate": 1.3390513907339627e-05,
+      "loss": 0.6575,
+      "step": 6923
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.36566319598932767,
+      "learning_rate": 1.3385047587536879e-05,
+      "loss": 0.6685,
+      "step": 6924
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3950934129504802,
+      "learning_rate": 1.3379581822522198e-05,
+      "loss": 0.6806,
+      "step": 6925
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.37794981593258437,
+      "learning_rate": 1.3374116612753993e-05,
+      "loss": 0.6083,
+      "step": 6926
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3876565159770545,
+      "learning_rate": 1.3368651958690628e-05,
+      "loss": 0.6758,
+      "step": 6927
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.38584153741171234,
+      "learning_rate": 1.3363187860790418e-05,
+      "loss": 0.6724,
+      "step": 6928
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.37045243875670514,
+      "learning_rate": 1.335772431951162e-05,
+      "loss": 0.6541,
+      "step": 6929
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.33583667093637043,
+      "learning_rate": 1.335226133531247e-05,
+      "loss": 0.6478,
+      "step": 6930
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.36140389851299454,
+      "learning_rate": 1.334679890865113e-05,
+      "loss": 0.6766,
+      "step": 6931
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3578036857632031,
+      "learning_rate": 1.3341337039985729e-05,
+      "loss": 0.6613,
+      "step": 6932
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.37818674856173135,
+      "learning_rate": 1.3335875729774341e-05,
+      "loss": 0.6464,
+      "step": 6933
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.32955903987517754,
+      "learning_rate": 1.3330414978475019e-05,
+      "loss": 0.6758,
+      "step": 6934
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4168434302657103,
+      "learning_rate": 1.3324954786545736e-05,
+      "loss": 0.6921,
+      "step": 6935
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.41422831724439435,
+      "learning_rate": 1.331949515444443e-05,
+      "loss": 0.6506,
+      "step": 6936
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3271848844700606,
+      "learning_rate": 1.3314036082629003e-05,
+      "loss": 0.6315,
+      "step": 6937
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3292485431023992,
+      "learning_rate": 1.3308577571557293e-05,
+      "loss": 0.6649,
+      "step": 6938
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.40657798618694446,
+      "learning_rate": 1.3303119621687098e-05,
+      "loss": 0.6546,
+      "step": 6939
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4214111468209047,
+      "learning_rate": 1.3297662233476167e-05,
+      "loss": 0.6962,
+      "step": 6940
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3983175492800378,
+      "learning_rate": 1.3292205407382222e-05,
+      "loss": 0.6584,
+      "step": 6941
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3956364427979855,
+      "learning_rate": 1.3286749143862903e-05,
+      "loss": 0.6483,
+      "step": 6942
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.36820587190662213,
+      "learning_rate": 1.3281293443375827e-05,
+      "loss": 0.6782,
+      "step": 6943
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4038472344947728,
+      "learning_rate": 1.3275838306378558e-05,
+      "loss": 0.6916,
+      "step": 6944
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4265799187300145,
+      "learning_rate": 1.3270383733328607e-05,
+      "loss": 0.7117,
+      "step": 6945
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3602950774202798,
+      "learning_rate": 1.3264929724683443e-05,
+      "loss": 0.6448,
+      "step": 6946
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3737749651590465,
+      "learning_rate": 1.3259476280900483e-05,
+      "loss": 0.6557,
+      "step": 6947
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3893529442283994,
+      "learning_rate": 1.325402340243712e-05,
+      "loss": 0.6147,
+      "step": 6948
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.34690232174375457,
+      "learning_rate": 1.3248571089750662e-05,
+      "loss": 0.656,
+      "step": 6949
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.40273017613264844,
+      "learning_rate": 1.3243119343298392e-05,
+      "loss": 0.6449,
+      "step": 6950
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4521680490925974,
+      "learning_rate": 1.3237668163537545e-05,
+      "loss": 0.6342,
+      "step": 6951
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3826487502090197,
+      "learning_rate": 1.3232217550925298e-05,
+      "loss": 0.6344,
+      "step": 6952
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3456874504640924,
+      "learning_rate": 1.3226767505918796e-05,
+      "loss": 0.6525,
+      "step": 6953
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3806055510740548,
+      "learning_rate": 1.3221318028975116e-05,
+      "loss": 0.682,
+      "step": 6954
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.36330683147277987,
+      "learning_rate": 1.3215869120551315e-05,
+      "loss": 0.7136,
+      "step": 6955
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3637046527419763,
+      "learning_rate": 1.3210420781104371e-05,
+      "loss": 0.6615,
+      "step": 6956
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.41106465413478366,
+      "learning_rate": 1.3204973011091242e-05,
+      "loss": 0.6725,
+      "step": 6957
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.369135780229136,
+      "learning_rate": 1.319952581096882e-05,
+      "loss": 0.6057,
+      "step": 6958
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.376418882906582,
+      "learning_rate": 1.3194079181193953e-05,
+      "loss": 0.647,
+      "step": 6959
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3716425835915002,
+      "learning_rate": 1.3188633122223449e-05,
+      "loss": 0.6764,
+      "step": 6960
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 9.413988550125293,
+      "learning_rate": 1.318318763451405e-05,
+      "loss": 7.142,
+      "step": 6961
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.36595194160677696,
+      "learning_rate": 1.317774271852248e-05,
+      "loss": 0.6544,
+      "step": 6962
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.39568807984050386,
+      "learning_rate": 1.317229837470538e-05,
+      "loss": 0.6506,
+      "step": 6963
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.38909579383705034,
+      "learning_rate": 1.3166854603519381e-05,
+      "loss": 0.6656,
+      "step": 6964
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.391697240617779,
+      "learning_rate": 1.316141140542103e-05,
+      "loss": 0.6757,
+      "step": 6965
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3733567763460464,
+      "learning_rate": 1.3155968780866843e-05,
+      "loss": 0.6784,
+      "step": 6966
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.34108312701847343,
+      "learning_rate": 1.3150526730313293e-05,
+      "loss": 0.6649,
+      "step": 6967
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.35706410342107997,
+      "learning_rate": 1.3145085254216784e-05,
+      "loss": 0.6442,
+      "step": 6968
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3670043836450428,
+      "learning_rate": 1.3139644353033707e-05,
+      "loss": 0.6635,
+      "step": 6969
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3625266935125662,
+      "learning_rate": 1.3134204027220367e-05,
+      "loss": 0.5996,
+      "step": 6970
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3727181434539058,
+      "learning_rate": 1.312876427723305e-05,
+      "loss": 0.688,
+      "step": 6971
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.42553854102357436,
+      "learning_rate": 1.312332510352797e-05,
+      "loss": 0.6305,
+      "step": 6972
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3607063333056718,
+      "learning_rate": 1.3117886506561309e-05,
+      "loss": 0.6126,
+      "step": 6973
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3701375924217155,
+      "learning_rate": 1.3112448486789197e-05,
+      "loss": 0.6678,
+      "step": 6974
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3637857626451235,
+      "learning_rate": 1.3107011044667705e-05,
+      "loss": 0.631,
+      "step": 6975
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.39422813139312896,
+      "learning_rate": 1.310157418065288e-05,
+      "loss": 0.668,
+      "step": 6976
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3550277866753676,
+      "learning_rate": 1.3096137895200693e-05,
+      "loss": 0.6462,
+      "step": 6977
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3440216880280957,
+      "learning_rate": 1.3090702188767088e-05,
+      "loss": 0.6261,
+      "step": 6978
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3387086115105597,
+      "learning_rate": 1.3085267061807945e-05,
+      "loss": 0.6645,
+      "step": 6979
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4816642629389711,
+      "learning_rate": 1.3079832514779099e-05,
+      "loss": 0.732,
+      "step": 6980
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.34946256529382486,
+      "learning_rate": 1.3074398548136346e-05,
+      "loss": 0.6554,
+      "step": 6981
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4188587107047489,
+      "learning_rate": 1.3068965162335417e-05,
+      "loss": 0.7133,
+      "step": 6982
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.39854688461278853,
+      "learning_rate": 1.3063532357832017e-05,
+      "loss": 0.6882,
+      "step": 6983
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.34072336426252503,
+      "learning_rate": 1.3058100135081775e-05,
+      "loss": 0.6344,
+      "step": 6984
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.37289165921638656,
+      "learning_rate": 1.3052668494540295e-05,
+      "loss": 0.6617,
+      "step": 6985
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.36991932405941613,
+      "learning_rate": 1.3047237436663113e-05,
+      "loss": 0.677,
+      "step": 6986
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4363894028890875,
+      "learning_rate": 1.3041806961905739e-05,
+      "loss": 0.6778,
+      "step": 6987
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3536495481597874,
+      "learning_rate": 1.3036377070723607e-05,
+      "loss": 0.6529,
+      "step": 6988
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3678910005790965,
+      "learning_rate": 1.3030947763572112e-05,
+      "loss": 0.6402,
+      "step": 6989
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4084061930831498,
+      "learning_rate": 1.3025519040906617e-05,
+      "loss": 0.6564,
+      "step": 6990
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 9.503954014690823,
+      "learning_rate": 1.3020090903182423e-05,
+      "loss": 7.3921,
+      "step": 6991
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.36956315254340905,
+      "learning_rate": 1.3014663350854775e-05,
+      "loss": 0.6521,
+      "step": 6992
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3824199266101147,
+      "learning_rate": 1.3009236384378872e-05,
+      "loss": 0.6541,
+      "step": 6993
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.41276812974926796,
+      "learning_rate": 1.3003810004209872e-05,
+      "loss": 0.6347,
+      "step": 6994
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4221337385611647,
+      "learning_rate": 1.299838421080287e-05,
+      "loss": 0.6501,
+      "step": 6995
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3552567655874674,
+      "learning_rate": 1.299295900461294e-05,
+      "loss": 0.6247,
+      "step": 6996
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.392699642605516,
+      "learning_rate": 1.2987534386095066e-05,
+      "loss": 0.6394,
+      "step": 6997
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37336040517433977,
+      "learning_rate": 1.2982110355704223e-05,
+      "loss": 0.6732,
+      "step": 6998
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.39324910643805416,
+      "learning_rate": 1.2976686913895306e-05,
+      "loss": 0.6617,
+      "step": 6999
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.36273612581476333,
+      "learning_rate": 1.2971264061123174e-05,
+      "loss": 0.6566,
+      "step": 7000
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3963829277570599,
+      "learning_rate": 1.296584179784264e-05,
+      "loss": 0.6614,
+      "step": 7001
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3606327027327727,
+      "learning_rate": 1.2960420124508451e-05,
+      "loss": 0.6669,
+      "step": 7002
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.39288166660720225,
+      "learning_rate": 1.2954999041575331e-05,
+      "loss": 0.6183,
+      "step": 7003
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3827544384978876,
+      "learning_rate": 1.294957854949793e-05,
+      "loss": 0.678,
+      "step": 7004
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37998441560561336,
+      "learning_rate": 1.2944158648730866e-05,
+      "loss": 0.6596,
+      "step": 7005
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.41177672256919806,
+      "learning_rate": 1.2938739339728695e-05,
+      "loss": 0.6664,
+      "step": 7006
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 6.355340295523512,
+      "learning_rate": 1.2933320622945924e-05,
+      "loss": 7.1666,
+      "step": 7007
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3848099609700227,
+      "learning_rate": 1.2927902498837021e-05,
+      "loss": 0.7018,
+      "step": 7008
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3612776579256559,
+      "learning_rate": 1.2922484967856388e-05,
+      "loss": 0.6325,
+      "step": 7009
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3636097251939481,
+      "learning_rate": 1.29170680304584e-05,
+      "loss": 0.6316,
+      "step": 7010
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4355596085843291,
+      "learning_rate": 1.2911651687097358e-05,
+      "loss": 0.6679,
+      "step": 7011
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.35402597585389883,
+      "learning_rate": 1.2906235938227534e-05,
+      "loss": 0.6304,
+      "step": 7012
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3936292791677247,
+      "learning_rate": 1.290082078430313e-05,
+      "loss": 0.6866,
+      "step": 7013
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.369943395388279,
+      "learning_rate": 1.2895406225778316e-05,
+      "loss": 0.6181,
+      "step": 7014
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3475061118693811,
+      "learning_rate": 1.28899922631072e-05,
+      "loss": 0.6657,
+      "step": 7015
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37904282278666707,
+      "learning_rate": 1.2884578896743842e-05,
+      "loss": 0.6429,
+      "step": 7016
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.34848263745043645,
+      "learning_rate": 1.2879166127142262e-05,
+      "loss": 0.6679,
+      "step": 7017
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4311690756550966,
+      "learning_rate": 1.287375395475642e-05,
+      "loss": 0.7299,
+      "step": 7018
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3153339912242871,
+      "learning_rate": 1.286834238004023e-05,
+      "loss": 0.6004,
+      "step": 7019
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.35358950504226233,
+      "learning_rate": 1.2862931403447549e-05,
+      "loss": 0.624,
+      "step": 7020
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3669495492869159,
+      "learning_rate": 1.2857521025432194e-05,
+      "loss": 0.6196,
+      "step": 7021
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3891569588925105,
+      "learning_rate": 1.2852111246447928e-05,
+      "loss": 0.6492,
+      "step": 7022
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.38651596234720825,
+      "learning_rate": 1.2846702066948446e-05,
+      "loss": 0.6483,
+      "step": 7023
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37877281598942125,
+      "learning_rate": 1.284129348738743e-05,
+      "loss": 0.6635,
+      "step": 7024
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3787441120591068,
+      "learning_rate": 1.2835885508218487e-05,
+      "loss": 0.6331,
+      "step": 7025
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37356321728351505,
+      "learning_rate": 1.2830478129895174e-05,
+      "loss": 0.6683,
+      "step": 7026
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0617915517585366,
+      "learning_rate": 1.2825071352870998e-05,
+      "loss": 7.0322,
+      "step": 7027
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3518988400938294,
+      "learning_rate": 1.2819665177599425e-05,
+      "loss": 0.6335,
+      "step": 7028
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3628803141338706,
+      "learning_rate": 1.2814259604533864e-05,
+      "loss": 0.6649,
+      "step": 7029
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4173627747282024,
+      "learning_rate": 1.2808854634127663e-05,
+      "loss": 0.6711,
+      "step": 7030
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.39868140214694325,
+      "learning_rate": 1.2803450266834144e-05,
+      "loss": 0.6852,
+      "step": 7031
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.39561146533501934,
+      "learning_rate": 1.279804650310656e-05,
+      "loss": 0.6758,
+      "step": 7032
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.36299931064442253,
+      "learning_rate": 1.2792643343398124e-05,
+      "loss": 0.6214,
+      "step": 7033
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3597840116299138,
+      "learning_rate": 1.2787240788161978e-05,
+      "loss": 0.6452,
+      "step": 7034
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37286396339011363,
+      "learning_rate": 1.2781838837851241e-05,
+      "loss": 0.668,
+      "step": 7035
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.39751514227546436,
+      "learning_rate": 1.2776437492918965e-05,
+      "loss": 0.6836,
+      "step": 7036
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3800048344556746,
+      "learning_rate": 1.2771036753818144e-05,
+      "loss": 0.6774,
+      "step": 7037
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.36162742651535235,
+      "learning_rate": 1.2765636621001743e-05,
+      "loss": 0.6631,
+      "step": 7038
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.36965243517132124,
+      "learning_rate": 1.2760237094922667e-05,
+      "loss": 0.675,
+      "step": 7039
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3766055751901421,
+      "learning_rate": 1.2754838176033761e-05,
+      "loss": 0.6639,
+      "step": 7040
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3807366386232923,
+      "learning_rate": 1.2749439864787827e-05,
+      "loss": 0.6559,
+      "step": 7041
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4166151683965028,
+      "learning_rate": 1.2744042161637615e-05,
+      "loss": 0.6755,
+      "step": 7042
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4545467119541529,
+      "learning_rate": 1.2738645067035822e-05,
+      "loss": 0.6299,
+      "step": 7043
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.36494013528696445,
+      "learning_rate": 1.2733248581435096e-05,
+      "loss": 0.6613,
+      "step": 7044
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.36853629108152575,
+      "learning_rate": 1.2727852705288037e-05,
+      "loss": 0.6809,
+      "step": 7045
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4140929432003456,
+      "learning_rate": 1.2722457439047192e-05,
+      "loss": 0.6539,
+      "step": 7046
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.361727423263023,
+      "learning_rate": 1.271706278316505e-05,
+      "loss": 0.6388,
+      "step": 7047
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3865458023602652,
+      "learning_rate": 1.2711668738094058e-05,
+      "loss": 0.6699,
+      "step": 7048
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37556717442504634,
+      "learning_rate": 1.270627530428661e-05,
+      "loss": 0.6442,
+      "step": 7049
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3756430810251017,
+      "learning_rate": 1.2700882482195038e-05,
+      "loss": 0.6283,
+      "step": 7050
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37752196262077015,
+      "learning_rate": 1.2695490272271633e-05,
+      "loss": 0.6749,
+      "step": 7051
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4650460471846496,
+      "learning_rate": 1.2690098674968646e-05,
+      "loss": 0.7099,
+      "step": 7052
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4037538466514268,
+      "learning_rate": 1.2684707690738253e-05,
+      "loss": 0.6526,
+      "step": 7053
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.0110691830888159,
+      "learning_rate": 1.2679317320032592e-05,
+      "loss": 6.9942,
+      "step": 7054
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.1003472194426855,
+      "learning_rate": 1.2673927563303745e-05,
+      "loss": 6.9224,
+      "step": 7055
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3403539402800065,
+      "learning_rate": 1.2668538421003752e-05,
+      "loss": 0.6452,
+      "step": 7056
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4356173396088732,
+      "learning_rate": 1.2663149893584577e-05,
+      "loss": 0.6574,
+      "step": 7057
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.34678914247505555,
+      "learning_rate": 1.2657761981498158e-05,
+      "loss": 0.6345,
+      "step": 7058
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.41486375076960713,
+      "learning_rate": 1.2652374685196383e-05,
+      "loss": 0.7421,
+      "step": 7059
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37783238195942304,
+      "learning_rate": 1.264698800513107e-05,
+      "loss": 0.6944,
+      "step": 7060
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37816530325214665,
+      "learning_rate": 1.2641601941753984e-05,
+      "loss": 0.6476,
+      "step": 7061
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3499109784339661,
+      "learning_rate": 1.2636216495516866e-05,
+      "loss": 0.6514,
+      "step": 7062
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37356043146029316,
+      "learning_rate": 1.2630831666871375e-05,
+      "loss": 0.6898,
+      "step": 7063
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8199977591460863,
+      "learning_rate": 1.2625447456269132e-05,
+      "loss": 6.8841,
+      "step": 7064
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3657145965205773,
+      "learning_rate": 1.2620063864161695e-05,
+      "loss": 0.6421,
+      "step": 7065
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.35838270526997207,
+      "learning_rate": 1.2614680891000601e-05,
+      "loss": 0.6359,
+      "step": 7066
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4176544582536231,
+      "learning_rate": 1.26092985372373e-05,
+      "loss": 0.6633,
+      "step": 7067
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37973881513703583,
+      "learning_rate": 1.2603916803323201e-05,
+      "loss": 0.6758,
+      "step": 7068
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3688832686735354,
+      "learning_rate": 1.259853568970967e-05,
+      "loss": 0.6631,
+      "step": 7069
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3891344639447461,
+      "learning_rate": 1.2593155196848011e-05,
+      "loss": 0.6759,
+      "step": 7070
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4251744244483504,
+      "learning_rate": 1.2587775325189484e-05,
+      "loss": 0.6845,
+      "step": 7071
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.32844607088299593,
+      "learning_rate": 1.258239607518528e-05,
+      "loss": 0.6227,
+      "step": 7072
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.39282195537938036,
+      "learning_rate": 1.2577017447286566e-05,
+      "loss": 0.6781,
+      "step": 7073
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3500432041098051,
+      "learning_rate": 1.257163944194443e-05,
+      "loss": 0.644,
+      "step": 7074
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.34199592315036365,
+      "learning_rate": 1.2566262059609928e-05,
+      "loss": 0.6378,
+      "step": 7075
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3405278472630915,
+      "learning_rate": 1.256088530073405e-05,
+      "loss": 0.6398,
+      "step": 7076
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3329541021664432,
+      "learning_rate": 1.2555509165767732e-05,
+      "loss": 0.6207,
+      "step": 7077
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37418206081303734,
+      "learning_rate": 1.2550133655161867e-05,
+      "loss": 0.6525,
+      "step": 7078
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.44487175090950354,
+      "learning_rate": 1.2544758769367303e-05,
+      "loss": 0.6715,
+      "step": 7079
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4037377407671119,
+      "learning_rate": 1.2539384508834816e-05,
+      "loss": 0.7075,
+      "step": 7080
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3780426347276762,
+      "learning_rate": 1.2534010874015134e-05,
+      "loss": 0.6044,
+      "step": 7081
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.34517085706423073,
+      "learning_rate": 1.2528637865358952e-05,
+      "loss": 0.6727,
+      "step": 7082
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4047254493002127,
+      "learning_rate": 1.2523265483316886e-05,
+      "loss": 0.669,
+      "step": 7083
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.32341280121757715,
+      "learning_rate": 1.2517893728339515e-05,
+      "loss": 0.6124,
+      "step": 7084
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3652637397543063,
+      "learning_rate": 1.251252260087735e-05,
+      "loss": 0.6274,
+      "step": 7085
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3495303792527666,
+      "learning_rate": 1.2507152101380884e-05,
+      "loss": 0.6378,
+      "step": 7086
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.40520486192799504,
+      "learning_rate": 1.2501782230300523e-05,
+      "loss": 0.6518,
+      "step": 7087
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.44567349708802834,
+      "learning_rate": 1.2496412988086626e-05,
+      "loss": 0.6756,
+      "step": 7088
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3660206413837599,
+      "learning_rate": 1.2491044375189514e-05,
+      "loss": 0.6431,
+      "step": 7089
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3545868431429442,
+      "learning_rate": 1.2485676392059438e-05,
+      "loss": 0.6232,
+      "step": 7090
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.41907571371139773,
+      "learning_rate": 1.248030903914661e-05,
+      "loss": 0.6869,
+      "step": 7091
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4434943738887856,
+      "learning_rate": 1.2474942316901171e-05,
+      "loss": 0.6813,
+      "step": 7092
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.35488370432379107,
+      "learning_rate": 1.2469576225773245e-05,
+      "loss": 0.6181,
+      "step": 7093
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3664707554523271,
+      "learning_rate": 1.2464210766212866e-05,
+      "loss": 0.6595,
+      "step": 7094
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.35930864795776424,
+      "learning_rate": 1.2458845938670021e-05,
+      "loss": 0.675,
+      "step": 7095
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.36293433595988916,
+      "learning_rate": 1.245348174359467e-05,
+      "loss": 0.6606,
+      "step": 7096
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.38428721589893544,
+      "learning_rate": 1.2448118181436683e-05,
+      "loss": 0.6897,
+      "step": 7097
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3671380565574724,
+      "learning_rate": 1.2442755252645905e-05,
+      "loss": 0.649,
+      "step": 7098
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3751822780414599,
+      "learning_rate": 1.2437392957672115e-05,
+      "loss": 0.6157,
+      "step": 7099
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37481398990153475,
+      "learning_rate": 1.243203129696505e-05,
+      "loss": 0.6529,
+      "step": 7100
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3236241504707491,
+      "learning_rate": 1.2426670270974378e-05,
+      "loss": 0.6192,
+      "step": 7101
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3639780216470166,
+      "learning_rate": 1.2421309880149723e-05,
+      "loss": 0.6493,
+      "step": 7102
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3470879001841112,
+      "learning_rate": 1.241595012494066e-05,
+      "loss": 0.6177,
+      "step": 7103
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.399303440156996,
+      "learning_rate": 1.2410591005796695e-05,
+      "loss": 0.7016,
+      "step": 7104
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3824922631499917,
+      "learning_rate": 1.2405232523167303e-05,
+      "loss": 0.6775,
+      "step": 7105
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3774739921731343,
+      "learning_rate": 1.2399874677501878e-05,
+      "loss": 0.6186,
+      "step": 7106
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.41552949027284813,
+      "learning_rate": 1.2394517469249791e-05,
+      "loss": 0.672,
+      "step": 7107
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.40437709004977757,
+      "learning_rate": 1.2389160898860336e-05,
+      "loss": 0.633,
+      "step": 7108
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.38026199172146685,
+      "learning_rate": 1.2383804966782771e-05,
+      "loss": 0.6499,
+      "step": 7109
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.5279440319102138,
+      "learning_rate": 1.2378449673466281e-05,
+      "loss": 6.9295,
+      "step": 7110
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3547734685657099,
+      "learning_rate": 1.2373095019360014e-05,
+      "loss": 0.6386,
+      "step": 7111
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4268141356346925,
+      "learning_rate": 1.2367741004913056e-05,
+      "loss": 0.677,
+      "step": 7112
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.35995877607755095,
+      "learning_rate": 1.2362387630574438e-05,
+      "loss": 0.6683,
+      "step": 7113
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.36427939200252285,
+      "learning_rate": 1.235703489679315e-05,
+      "loss": 0.6953,
+      "step": 7114
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.37488163345014686,
+      "learning_rate": 1.2351682804018115e-05,
+      "loss": 0.6646,
+      "step": 7115
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4277121456302301,
+      "learning_rate": 1.2346331352698206e-05,
+      "loss": 0.6714,
+      "step": 7116
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.360605610296369,
+      "learning_rate": 1.2340980543282246e-05,
+      "loss": 0.65,
+      "step": 7117
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3828707742960618,
+      "learning_rate": 1.2335630376218992e-05,
+      "loss": 0.643,
+      "step": 7118
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4189783920635014,
+      "learning_rate": 1.2330280851957168e-05,
+      "loss": 0.6266,
+      "step": 7119
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4242510964522705,
+      "learning_rate": 1.232493197094542e-05,
+      "loss": 0.6461,
+      "step": 7120
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.43259979446459823,
+      "learning_rate": 1.2319583733632363e-05,
+      "loss": 0.6725,
+      "step": 7121
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4231476139971143,
+      "learning_rate": 1.231423614046654e-05,
+      "loss": 0.6846,
+      "step": 7122
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4243695636376085,
+      "learning_rate": 1.2308889191896454e-05,
+      "loss": 0.6688,
+      "step": 7123
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3628857887779355,
+      "learning_rate": 1.230354288837054e-05,
+      "loss": 0.6511,
+      "step": 7124
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4182342662636245,
+      "learning_rate": 1.229819723033719e-05,
+      "loss": 0.6582,
+      "step": 7125
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3708467830950129,
+      "learning_rate": 1.229285221824474e-05,
+      "loss": 0.6481,
+      "step": 7126
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3503576088341682,
+      "learning_rate": 1.2287507852541456e-05,
+      "loss": 0.6509,
+      "step": 7127
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4422594460538667,
+      "learning_rate": 1.2282164133675583e-05,
+      "loss": 0.6915,
+      "step": 7128
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.43025199473049563,
+      "learning_rate": 1.227682106209528e-05,
+      "loss": 0.6911,
+      "step": 7129
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.42017664450932485,
+      "learning_rate": 1.2271478638248671e-05,
+      "loss": 0.6263,
+      "step": 7130
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3884304462339108,
+      "learning_rate": 1.2266136862583811e-05,
+      "loss": 0.6408,
+      "step": 7131
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3965547603600306,
+      "learning_rate": 1.2260795735548716e-05,
+      "loss": 0.7062,
+      "step": 7132
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.40429019049633075,
+      "learning_rate": 1.2255455257591335e-05,
+      "loss": 0.6568,
+      "step": 7133
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4341747621372466,
+      "learning_rate": 1.2250115429159562e-05,
+      "loss": 0.6874,
+      "step": 7134
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.40578145213400574,
+      "learning_rate": 1.2244776250701252e-05,
+      "loss": 0.6577,
+      "step": 7135
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.37669303163300694,
+      "learning_rate": 1.2239437722664195e-05,
+      "loss": 0.655,
+      "step": 7136
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4109686166682733,
+      "learning_rate": 1.2234099845496125e-05,
+      "loss": 0.6382,
+      "step": 7137
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3626296046535224,
+      "learning_rate": 1.2228762619644718e-05,
+      "loss": 0.6482,
+      "step": 7138
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3528353421009073,
+      "learning_rate": 1.222342604555761e-05,
+      "loss": 0.6345,
+      "step": 7139
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4142079618313333,
+      "learning_rate": 1.2218090123682367e-05,
+      "loss": 0.6728,
+      "step": 7140
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4092097720160944,
+      "learning_rate": 1.2212754854466495e-05,
+      "loss": 0.6621,
+      "step": 7141
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3450522887251105,
+      "learning_rate": 1.2207420238357478e-05,
+      "loss": 0.6061,
+      "step": 7142
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3395688158966299,
+      "learning_rate": 1.2202086275802718e-05,
+      "loss": 0.6691,
+      "step": 7143
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.37332681647730664,
+      "learning_rate": 1.2196752967249564e-05,
+      "loss": 0.6408,
+      "step": 7144
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3512826020674814,
+      "learning_rate": 1.2191420313145309e-05,
+      "loss": 0.6345,
+      "step": 7145
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4203455695072457,
+      "learning_rate": 1.2186088313937208e-05,
+      "loss": 0.6356,
+      "step": 7146
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.35491861420491216,
+      "learning_rate": 1.2180756970072443e-05,
+      "loss": 0.667,
+      "step": 7147
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.37026168288328515,
+      "learning_rate": 1.2175426281998137e-05,
+      "loss": 0.6542,
+      "step": 7148
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3917139908747708,
+      "learning_rate": 1.2170096250161386e-05,
+      "loss": 0.6762,
+      "step": 7149
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3661142984428016,
+      "learning_rate": 1.2164766875009208e-05,
+      "loss": 0.6388,
+      "step": 7150
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.8847189333421417,
+      "learning_rate": 1.2159438156988573e-05,
+      "loss": 6.8841,
+      "step": 7151
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3460378667204856,
+      "learning_rate": 1.2154110096546386e-05,
+      "loss": 0.6242,
+      "step": 7152
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.37650511957047145,
+      "learning_rate": 1.2148782694129513e-05,
+      "loss": 0.6532,
+      "step": 7153
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3949514171680536,
+      "learning_rate": 1.2143455950184752e-05,
+      "loss": 0.6735,
+      "step": 7154
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3559655769764713,
+      "learning_rate": 1.2138129865158845e-05,
+      "loss": 0.6647,
+      "step": 7155
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.40118835730759494,
+      "learning_rate": 1.2132804439498496e-05,
+      "loss": 0.6587,
+      "step": 7156
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.39402952037314537,
+      "learning_rate": 1.2127479673650342e-05,
+      "loss": 0.6622,
+      "step": 7157
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.32934703600786636,
+      "learning_rate": 1.212215556806096e-05,
+      "loss": 0.6091,
+      "step": 7158
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4060540692195567,
+      "learning_rate": 1.211683212317687e-05,
+      "loss": 0.6274,
+      "step": 7159
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.34680752248877417,
+      "learning_rate": 1.2111509339444556e-05,
+      "loss": 0.6326,
+      "step": 7160
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.406715052972526,
+      "learning_rate": 1.210618721731042e-05,
+      "loss": 0.6718,
+      "step": 7161
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.36894749227807194,
+      "learning_rate": 1.2100865757220835e-05,
+      "loss": 0.6706,
+      "step": 7162
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.44225431134560433,
+      "learning_rate": 1.2095544959622095e-05,
+      "loss": 0.6996,
+      "step": 7163
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.35494936943187155,
+      "learning_rate": 1.2090224824960455e-05,
+      "loss": 0.6483,
+      "step": 7164
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3357039621922624,
+      "learning_rate": 1.2084905353682106e-05,
+      "loss": 0.6371,
+      "step": 7165
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.35467655472729204,
+      "learning_rate": 1.2079586546233188e-05,
+      "loss": 0.6548,
+      "step": 7166
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.36445849921795415,
+      "learning_rate": 1.207426840305978e-05,
+      "loss": 0.6354,
+      "step": 7167
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4033764314891946,
+      "learning_rate": 1.2068950924607905e-05,
+      "loss": 0.6674,
+      "step": 7168
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3848825270184012,
+      "learning_rate": 1.2063634111323538e-05,
+      "loss": 0.6478,
+      "step": 7169
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.38063086642090044,
+      "learning_rate": 1.2058317963652599e-05,
+      "loss": 0.6489,
+      "step": 7170
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3585932185722525,
+      "learning_rate": 1.2053002482040941e-05,
+      "loss": 0.6417,
+      "step": 7171
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4043617439736987,
+      "learning_rate": 1.2047687666934365e-05,
+      "loss": 0.694,
+      "step": 7172
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.37825190588036095,
+      "learning_rate": 1.204237351877862e-05,
+      "loss": 0.6577,
+      "step": 7173
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4264557198226312,
+      "learning_rate": 1.2037060038019402e-05,
+      "loss": 0.6515,
+      "step": 7174
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.42093373035428705,
+      "learning_rate": 1.2031747225102332e-05,
+      "loss": 0.6684,
+      "step": 7175
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.40504591754310254,
+      "learning_rate": 1.2026435080473005e-05,
+      "loss": 0.6509,
+      "step": 7176
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4000100854036127,
+      "learning_rate": 1.202112360457694e-05,
+      "loss": 0.6918,
+      "step": 7177
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3455459680368162,
+      "learning_rate": 1.20158127978596e-05,
+      "loss": 0.6499,
+      "step": 7178
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3722947821580199,
+      "learning_rate": 1.20105026607664e-05,
+      "loss": 0.6839,
+      "step": 7179
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4118868539470757,
+      "learning_rate": 1.2005193193742692e-05,
+      "loss": 0.6608,
+      "step": 7180
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.342770581512971,
+      "learning_rate": 1.1999884397233776e-05,
+      "loss": 0.6572,
+      "step": 7181
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.37184116139528967,
+      "learning_rate": 1.1994576271684887e-05,
+      "loss": 0.6303,
+      "step": 7182
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4042297591623002,
+      "learning_rate": 1.1989268817541225e-05,
+      "loss": 0.6846,
+      "step": 7183
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.37053145250278274,
+      "learning_rate": 1.1983962035247911e-05,
+      "loss": 0.6693,
+      "step": 7184
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.42766795027834204,
+      "learning_rate": 1.1978655925250024e-05,
+      "loss": 0.675,
+      "step": 7185
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.34925261458955537,
+      "learning_rate": 1.1973350487992572e-05,
+      "loss": 0.6339,
+      "step": 7186
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3819757943842405,
+      "learning_rate": 1.1968045723920528e-05,
+      "loss": 0.6825,
+      "step": 7187
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4099210102129651,
+      "learning_rate": 1.1962741633478779e-05,
+      "loss": 0.6869,
+      "step": 7188
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3539324604669893,
+      "learning_rate": 1.1957438217112187e-05,
+      "loss": 0.6268,
+      "step": 7189
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3302769957434214,
+      "learning_rate": 1.1952135475265538e-05,
+      "loss": 0.6315,
+      "step": 7190
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.34940481981465027,
+      "learning_rate": 1.1946833408383572e-05,
+      "loss": 0.643,
+      "step": 7191
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.34059885396681366,
+      "learning_rate": 1.1941532016910955e-05,
+      "loss": 0.6245,
+      "step": 7192
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.382188200383038,
+      "learning_rate": 1.1936231301292322e-05,
+      "loss": 0.6337,
+      "step": 7193
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.39229348894041133,
+      "learning_rate": 1.1930931261972231e-05,
+      "loss": 0.6536,
+      "step": 7194
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3815194597119499,
+      "learning_rate": 1.1925631899395187e-05,
+      "loss": 0.6613,
+      "step": 7195
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.36516881883375796,
+      "learning_rate": 1.1920333214005639e-05,
+      "loss": 0.6634,
+      "step": 7196
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.39176902948558495,
+      "learning_rate": 1.1915035206247994e-05,
+      "loss": 0.6273,
+      "step": 7197
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3560308934064435,
+      "learning_rate": 1.1909737876566582e-05,
+      "loss": 0.6749,
+      "step": 7198
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3659185690269956,
+      "learning_rate": 1.1904441225405684e-05,
+      "loss": 0.6522,
+      "step": 7199
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3475983052764763,
+      "learning_rate": 1.1899145253209521e-05,
+      "loss": 0.6535,
+      "step": 7200
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3511931012401694,
+      "learning_rate": 1.1893849960422265e-05,
+      "loss": 0.6406,
+      "step": 7201
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.39450222267676804,
+      "learning_rate": 1.188855534748802e-05,
+      "loss": 0.6639,
+      "step": 7202
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.40604465685488794,
+      "learning_rate": 1.1883261414850837e-05,
+      "loss": 0.7132,
+      "step": 7203
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.393089077242574,
+      "learning_rate": 1.1877968162954727e-05,
+      "loss": 0.64,
+      "step": 7204
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3868596078447355,
+      "learning_rate": 1.1872675592243615e-05,
+      "loss": 0.7187,
+      "step": 7205
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3707675422665654,
+      "learning_rate": 1.1867383703161383e-05,
+      "loss": 0.6454,
+      "step": 7206
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.40420640356880255,
+      "learning_rate": 1.1862092496151862e-05,
+      "loss": 0.6815,
+      "step": 7207
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.33384593216891306,
+      "learning_rate": 1.1856801971658815e-05,
+      "loss": 0.6324,
+      "step": 7208
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.35877040514657765,
+      "learning_rate": 1.1851512130125951e-05,
+      "loss": 0.6562,
+      "step": 7209
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4286455295740244,
+      "learning_rate": 1.1846222971996916e-05,
+      "loss": 0.7237,
+      "step": 7210
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3562291377706774,
+      "learning_rate": 1.1840934497715322e-05,
+      "loss": 0.6357,
+      "step": 7211
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3784409301856489,
+      "learning_rate": 1.1835646707724698e-05,
+      "loss": 0.6522,
+      "step": 7212
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3633621859180319,
+      "learning_rate": 1.1830359602468522e-05,
+      "loss": 0.6614,
+      "step": 7213
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3763191100011546,
+      "learning_rate": 1.1825073182390221e-05,
+      "loss": 0.6414,
+      "step": 7214
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.40344966253871567,
+      "learning_rate": 1.181978744793316e-05,
+      "loss": 0.627,
+      "step": 7215
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.38113128717857736,
+      "learning_rate": 1.1814502399540646e-05,
+      "loss": 0.6566,
+      "step": 7216
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.35202880597094643,
+      "learning_rate": 1.1809218037655923e-05,
+      "loss": 0.6539,
+      "step": 7217
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.39411056325115096,
+      "learning_rate": 1.1803934362722199e-05,
+      "loss": 0.6639,
+      "step": 7218
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.38742335870919836,
+      "learning_rate": 1.1798651375182604e-05,
+      "loss": 0.6409,
+      "step": 7219
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3910913621667271,
+      "learning_rate": 1.1793369075480205e-05,
+      "loss": 0.6302,
+      "step": 7220
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3861518874696426,
+      "learning_rate": 1.178808746405804e-05,
+      "loss": 0.6589,
+      "step": 7221
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.38647836745570446,
+      "learning_rate": 1.1782806541359055e-05,
+      "loss": 0.6836,
+      "step": 7222
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3920470116022108,
+      "learning_rate": 1.1777526307826166e-05,
+      "loss": 0.6949,
+      "step": 7223
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.36493080883202467,
+      "learning_rate": 1.1772246763902206e-05,
+      "loss": 0.678,
+      "step": 7224
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.39266661719813123,
+      "learning_rate": 1.1766967910029985e-05,
+      "loss": 0.6798,
+      "step": 7225
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3698219079932879,
+      "learning_rate": 1.1761689746652216e-05,
+      "loss": 0.6553,
+      "step": 7226
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3862623647029884,
+      "learning_rate": 1.1756412274211581e-05,
+      "loss": 0.6828,
+      "step": 7227
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4237113128289609,
+      "learning_rate": 1.1751135493150696e-05,
+      "loss": 0.6413,
+      "step": 7228
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.39271564674114096,
+      "learning_rate": 1.1745859403912108e-05,
+      "loss": 0.6522,
+      "step": 7229
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3914305882368539,
+      "learning_rate": 1.174058400693833e-05,
+      "loss": 0.6507,
+      "step": 7230
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.35993704276791866,
+      "learning_rate": 1.1735309302671786e-05,
+      "loss": 0.6342,
+      "step": 7231
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3441264291682253,
+      "learning_rate": 1.173003529155488e-05,
+      "loss": 0.6502,
+      "step": 7232
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.40326948556816977,
+      "learning_rate": 1.1724761974029918e-05,
+      "loss": 0.7017,
+      "step": 7233
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.386727032879893,
+      "learning_rate": 1.1719489350539185e-05,
+      "loss": 0.6286,
+      "step": 7234
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.38412765666107446,
+      "learning_rate": 1.1714217421524878e-05,
+      "loss": 0.643,
+      "step": 7235
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4214796269579726,
+      "learning_rate": 1.1708946187429145e-05,
+      "loss": 0.6822,
+      "step": 7236
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.38468815537186757,
+      "learning_rate": 1.1703675648694083e-05,
+      "loss": 0.6428,
+      "step": 7237
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3908658654917581,
+      "learning_rate": 1.1698405805761723e-05,
+      "loss": 0.7069,
+      "step": 7238
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.40040905869417454,
+      "learning_rate": 1.1693136659074048e-05,
+      "loss": 0.668,
+      "step": 7239
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3453398659657584,
+      "learning_rate": 1.1687868209072964e-05,
+      "loss": 0.665,
+      "step": 7240
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3864296305123651,
+      "learning_rate": 1.1682600456200337e-05,
+      "loss": 0.6528,
+      "step": 7241
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4127401994870905,
+      "learning_rate": 1.1677333400897968e-05,
+      "loss": 0.6866,
+      "step": 7242
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3700754993725158,
+      "learning_rate": 1.1672067043607593e-05,
+      "loss": 0.6382,
+      "step": 7243
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3572083325745243,
+      "learning_rate": 1.1666801384770889e-05,
+      "loss": 0.6654,
+      "step": 7244
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.34542374862086833,
+      "learning_rate": 1.1661536424829493e-05,
+      "loss": 0.6541,
+      "step": 7245
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.39097361147356374,
+      "learning_rate": 1.1656272164224978e-05,
+      "loss": 0.6491,
+      "step": 7246
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3932104629989375,
+      "learning_rate": 1.165100860339883e-05,
+      "loss": 0.6379,
+      "step": 7247
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.36244752135486136,
+      "learning_rate": 1.164574574279251e-05,
+      "loss": 0.6482,
+      "step": 7248
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3979708150667716,
+      "learning_rate": 1.1640483582847408e-05,
+      "loss": 0.6636,
+      "step": 7249
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.36349560006834997,
+      "learning_rate": 1.1635222124004848e-05,
+      "loss": 0.6449,
+      "step": 7250
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3522402840722064,
+      "learning_rate": 1.1629961366706106e-05,
+      "loss": 0.6127,
+      "step": 7251
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.358755627513573,
+      "learning_rate": 1.1624701311392395e-05,
+      "loss": 0.6349,
+      "step": 7252
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4229911444014417,
+      "learning_rate": 1.1619441958504871e-05,
+      "loss": 0.7035,
+      "step": 7253
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.38359448946542823,
+      "learning_rate": 1.1614183308484637e-05,
+      "loss": 0.6711,
+      "step": 7254
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 7.431483249555298,
+      "learning_rate": 1.1608925361772712e-05,
+      "loss": 6.8991,
+      "step": 7255
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.37121955651251537,
+      "learning_rate": 1.1603668118810084e-05,
+      "loss": 0.644,
+      "step": 7256
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3724115935212546,
+      "learning_rate": 1.1598411580037677e-05,
+      "loss": 0.6472,
+      "step": 7257
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3909185697739522,
+      "learning_rate": 1.159315574589633e-05,
+      "loss": 0.6845,
+      "step": 7258
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.31990956799397335,
+      "learning_rate": 1.1587900616826864e-05,
+      "loss": 0.6553,
+      "step": 7259
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.37489828777063694,
+      "learning_rate": 1.1582646193270024e-05,
+      "loss": 0.6624,
+      "step": 7260
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3861949870935665,
+      "learning_rate": 1.1577392475666473e-05,
+      "loss": 0.6321,
+      "step": 7261
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.400177234541898,
+      "learning_rate": 1.1572139464456846e-05,
+      "loss": 0.6441,
+      "step": 7262
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.37496118505938464,
+      "learning_rate": 1.1566887160081708e-05,
+      "loss": 0.6575,
+      "step": 7263
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4832068320223653,
+      "learning_rate": 1.1561635562981555e-05,
+      "loss": 0.6704,
+      "step": 7264
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.43606724329687696,
+      "learning_rate": 1.1556384673596832e-05,
+      "loss": 0.72,
+      "step": 7265
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3643213455632387,
+      "learning_rate": 1.1551134492367933e-05,
+      "loss": 0.6829,
+      "step": 7266
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.42462607512717604,
+      "learning_rate": 1.1545885019735176e-05,
+      "loss": 0.6451,
+      "step": 7267
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.37413745837120815,
+      "learning_rate": 1.1540636256138837e-05,
+      "loss": 0.6691,
+      "step": 7268
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3624462551955566,
+      "learning_rate": 1.1535388202019125e-05,
+      "loss": 0.6798,
+      "step": 7269
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.39307011502828654,
+      "learning_rate": 1.1530140857816172e-05,
+      "loss": 0.673,
+      "step": 7270
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.362277959361547,
+      "learning_rate": 1.1524894223970077e-05,
+      "loss": 0.5938,
+      "step": 7271
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3588784153770997,
+      "learning_rate": 1.1519648300920872e-05,
+      "loss": 0.6763,
+      "step": 7272
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3422722487572578,
+      "learning_rate": 1.151440308910852e-05,
+      "loss": 0.6465,
+      "step": 7273
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3602382676793653,
+      "learning_rate": 1.150915858897294e-05,
+      "loss": 0.643,
+      "step": 7274
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3665167006251667,
+      "learning_rate": 1.1503914800953969e-05,
+      "loss": 0.6742,
+      "step": 7275
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3442290067480788,
+      "learning_rate": 1.14986717254914e-05,
+      "loss": 0.6369,
+      "step": 7276
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 10.907701464528028,
+      "learning_rate": 1.1493429363024975e-05,
+      "loss": 6.9873,
+      "step": 7277
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.39387785763734057,
+      "learning_rate": 1.148818771399435e-05,
+      "loss": 0.7023,
+      "step": 7278
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.40401045821921944,
+      "learning_rate": 1.148294677883913e-05,
+      "loss": 0.6787,
+      "step": 7279
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4464762614364083,
+      "learning_rate": 1.1477706557998895e-05,
+      "loss": 0.6581,
+      "step": 7280
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3887390840458265,
+      "learning_rate": 1.1472467051913112e-05,
+      "loss": 0.6516,
+      "step": 7281
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3404170740679704,
+      "learning_rate": 1.146722826102122e-05,
+      "loss": 0.6546,
+      "step": 7282
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.38743188737250145,
+      "learning_rate": 1.1461990185762595e-05,
+      "loss": 0.6567,
+      "step": 7283
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3715016530543112,
+      "learning_rate": 1.1456752826576535e-05,
+      "loss": 0.6938,
+      "step": 7284
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3584676978650177,
+      "learning_rate": 1.1451516183902295e-05,
+      "loss": 0.6253,
+      "step": 7285
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.39085144917044795,
+      "learning_rate": 1.144628025817907e-05,
+      "loss": 0.6539,
+      "step": 7286
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.040104654118856,
+      "learning_rate": 1.144104504984599e-05,
+      "loss": 6.8439,
+      "step": 7287
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.32794238544324056,
+      "learning_rate": 1.1435810559342136e-05,
+      "loss": 0.6326,
+      "step": 7288
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4352870770076058,
+      "learning_rate": 1.1430576787106492e-05,
+      "loss": 0.6876,
+      "step": 7289
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4001017252943788,
+      "learning_rate": 1.1425343733578029e-05,
+      "loss": 0.6107,
+      "step": 7290
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.35852649254208063,
+      "learning_rate": 1.1420111399195633e-05,
+      "loss": 0.6569,
+      "step": 7291
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4049000635675444,
+      "learning_rate": 1.1414879784398136e-05,
+      "loss": 0.666,
+      "step": 7292
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.38295151439899155,
+      "learning_rate": 1.140964888962429e-05,
+      "loss": 0.6968,
+      "step": 7293
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.381235706831356,
+      "learning_rate": 1.1404418715312832e-05,
+      "loss": 0.6906,
+      "step": 7294
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.383087002024989,
+      "learning_rate": 1.1399189261902387e-05,
+      "loss": 0.6509,
+      "step": 7295
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3838119377827922,
+      "learning_rate": 1.1393960529831553e-05,
+      "loss": 0.6514,
+      "step": 7296
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.41832976394847665,
+      "learning_rate": 1.138873251953886e-05,
+      "loss": 0.6948,
+      "step": 7297
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.36596820853477097,
+      "learning_rate": 1.1383505231462768e-05,
+      "loss": 0.6536,
+      "step": 7298
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.366914527921701,
+      "learning_rate": 1.1378278666041684e-05,
+      "loss": 0.6388,
+      "step": 7299
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.374994295187349,
+      "learning_rate": 1.1373052823713957e-05,
+      "loss": 0.6235,
+      "step": 7300
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3780136688823589,
+      "learning_rate": 1.1367827704917868e-05,
+      "loss": 0.5779,
+      "step": 7301
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3893566452872682,
+      "learning_rate": 1.1362603310091647e-05,
+      "loss": 0.6053,
+      "step": 7302
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.36138460242390363,
+      "learning_rate": 1.1357379639673463e-05,
+      "loss": 0.6424,
+      "step": 7303
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.330608578192381,
+      "learning_rate": 1.13521566941014e-05,
+      "loss": 0.6519,
+      "step": 7304
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.38948776965731113,
+      "learning_rate": 1.1346934473813516e-05,
+      "loss": 0.6473,
+      "step": 7305
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.39879441001891225,
+      "learning_rate": 1.134171297924779e-05,
+      "loss": 0.6666,
+      "step": 7306
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3277690706704764,
+      "learning_rate": 1.1336492210842127e-05,
+      "loss": 0.6444,
+      "step": 7307
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.37004999293082713,
+      "learning_rate": 1.1331272169034415e-05,
+      "loss": 0.6544,
+      "step": 7308
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3900244962125542,
+      "learning_rate": 1.132605285426243e-05,
+      "loss": 0.6629,
+      "step": 7309
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.34088549778911875,
+      "learning_rate": 1.1320834266963914e-05,
+      "loss": 0.6476,
+      "step": 7310
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.2451564756584967,
+      "learning_rate": 1.1315616407576555e-05,
+      "loss": 0.6563,
+      "step": 7311
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.37023191524475235,
+      "learning_rate": 1.1310399276537956e-05,
+      "loss": 0.6241,
+      "step": 7312
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3906536063641971,
+      "learning_rate": 1.1305182874285669e-05,
+      "loss": 0.5907,
+      "step": 7313
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3692809152932706,
+      "learning_rate": 1.1299967201257198e-05,
+      "loss": 0.6408,
+      "step": 7314
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.9711004365361657,
+      "learning_rate": 1.129475225788997e-05,
+      "loss": 0.6375,
+      "step": 7315
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3672854999405341,
+      "learning_rate": 1.1289538044621355e-05,
+      "loss": 0.6386,
+      "step": 7316
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3776863580157459,
+      "learning_rate": 1.1284324561888672e-05,
+      "loss": 0.6011,
+      "step": 7317
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.38625866014855614,
+      "learning_rate": 1.127911181012916e-05,
+      "loss": 0.6866,
+      "step": 7318
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.35544440283167933,
+      "learning_rate": 1.1273899789780003e-05,
+      "loss": 0.6441,
+      "step": 7319
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.37978237169572804,
+      "learning_rate": 1.126868850127834e-05,
+      "loss": 0.6718,
+      "step": 7320
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4264302006944471,
+      "learning_rate": 1.1263477945061214e-05,
+      "loss": 0.6513,
+      "step": 7321
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3626418225348755,
+      "learning_rate": 1.1258268121565656e-05,
+      "loss": 0.6353,
+      "step": 7322
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4440144464526193,
+      "learning_rate": 1.1253059031228589e-05,
+      "loss": 0.7161,
+      "step": 7323
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.40674659541114727,
+      "learning_rate": 1.1247850674486894e-05,
+      "loss": 0.6908,
+      "step": 7324
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.35928293550315327,
+      "learning_rate": 1.1242643051777393e-05,
+      "loss": 0.6687,
+      "step": 7325
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.42649959173569685,
+      "learning_rate": 1.1237436163536853e-05,
+      "loss": 0.6879,
+      "step": 7326
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.850337015027042,
+      "learning_rate": 1.123223001020195e-05,
+      "loss": 6.8278,
+      "step": 7327
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.375964233969099,
+      "learning_rate": 1.1227024592209326e-05,
+      "loss": 0.647,
+      "step": 7328
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3731406786675524,
+      "learning_rate": 1.1221819909995558e-05,
+      "loss": 0.6598,
+      "step": 7329
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.41721932010325286,
+      "learning_rate": 1.1216615963997148e-05,
+      "loss": 0.6546,
+      "step": 7330
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3849443325741999,
+      "learning_rate": 1.1211412754650559e-05,
+      "loss": 0.6637,
+      "step": 7331
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4295388974746015,
+      "learning_rate": 1.1206210282392162e-05,
+      "loss": 0.7231,
+      "step": 7332
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.40895328964675665,
+      "learning_rate": 1.1201008547658287e-05,
+      "loss": 0.6934,
+      "step": 7333
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3733296946442639,
+      "learning_rate": 1.11958075508852e-05,
+      "loss": 0.6193,
+      "step": 7334
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3911020797083727,
+      "learning_rate": 1.1190607292509101e-05,
+      "loss": 0.617,
+      "step": 7335
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4063230499226848,
+      "learning_rate": 1.1185407772966129e-05,
+      "loss": 0.6428,
+      "step": 7336
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.42070532358700446,
+      "learning_rate": 1.118020899269237e-05,
+      "loss": 0.6666,
+      "step": 7337
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3641310518836489,
+      "learning_rate": 1.1175010952123821e-05,
+      "loss": 0.6464,
+      "step": 7338
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.39358628061561163,
+      "learning_rate": 1.1169813651696448e-05,
+      "loss": 0.6614,
+      "step": 7339
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.38017114903475724,
+      "learning_rate": 1.1164617091846149e-05,
+      "loss": 0.6182,
+      "step": 7340
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.43372686558490836,
+      "learning_rate": 1.1159421273008723e-05,
+      "loss": 0.6827,
+      "step": 7341
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4331899487421791,
+      "learning_rate": 1.1154226195619979e-05,
+      "loss": 0.6631,
+      "step": 7342
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3963274388849969,
+      "learning_rate": 1.114903186011559e-05,
+      "loss": 0.6474,
+      "step": 7343
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.36304090331333133,
+      "learning_rate": 1.114383826693121e-05,
+      "loss": 0.6244,
+      "step": 7344
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.327757506082114,
+      "learning_rate": 1.113864541650243e-05,
+      "loss": 0.6609,
+      "step": 7345
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3545234288805482,
+      "learning_rate": 1.1133453309264745e-05,
+      "loss": 0.6476,
+      "step": 7346
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.43064280605154365,
+      "learning_rate": 1.1128261945653627e-05,
+      "loss": 0.694,
+      "step": 7347
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3901293083272029,
+      "learning_rate": 1.1123071326104462e-05,
+      "loss": 0.6348,
+      "step": 7348
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4061125863396371,
+      "learning_rate": 1.1117881451052583e-05,
+      "loss": 0.6645,
+      "step": 7349
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.34876906921665535,
+      "learning_rate": 1.1112692320933265e-05,
+      "loss": 0.6738,
+      "step": 7350
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.40036215235799877,
+      "learning_rate": 1.1107503936181715e-05,
+      "loss": 0.6332,
+      "step": 7351
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3515865577730456,
+      "learning_rate": 1.110231629723306e-05,
+      "loss": 0.6724,
+      "step": 7352
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.358063774679861,
+      "learning_rate": 1.1097129404522394e-05,
+      "loss": 0.6418,
+      "step": 7353
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3917491709180115,
+      "learning_rate": 1.1091943258484739e-05,
+      "loss": 0.6736,
+      "step": 7354
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.40593314547321613,
+      "learning_rate": 1.1086757859555028e-05,
+      "loss": 0.6574,
+      "step": 7355
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.39455850071368276,
+      "learning_rate": 1.1081573208168181e-05,
+      "loss": 0.6397,
+      "step": 7356
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4036324963978148,
+      "learning_rate": 1.1076389304759028e-05,
+      "loss": 0.6778,
+      "step": 7357
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3984277854713651,
+      "learning_rate": 1.1071206149762315e-05,
+      "loss": 0.6468,
+      "step": 7358
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.34388035517885596,
+      "learning_rate": 1.1066023743612759e-05,
+      "loss": 0.6624,
+      "step": 7359
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3646822671973718,
+      "learning_rate": 1.1060842086745014e-05,
+      "loss": 0.6468,
+      "step": 7360
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3373662524593489,
+      "learning_rate": 1.1055661179593638e-05,
+      "loss": 0.6182,
+      "step": 7361
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.42935773166821206,
+      "learning_rate": 1.1050481022593154e-05,
+      "loss": 0.6882,
+      "step": 7362
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.35271704696788136,
+      "learning_rate": 1.104530161617802e-05,
+      "loss": 0.6317,
+      "step": 7363
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3583478354946688,
+      "learning_rate": 1.1040122960782626e-05,
+      "loss": 0.6369,
+      "step": 7364
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3798964109292233,
+      "learning_rate": 1.1034945056841306e-05,
+      "loss": 0.6623,
+      "step": 7365
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.45207060620918366,
+      "learning_rate": 1.102976790478831e-05,
+      "loss": 0.6522,
+      "step": 7366
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3414816202053782,
+      "learning_rate": 1.1024591505057845e-05,
+      "loss": 0.6245,
+      "step": 7367
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.38093868517952945,
+      "learning_rate": 1.101941585808406e-05,
+      "loss": 0.6182,
+      "step": 7368
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.41345080870516165,
+      "learning_rate": 1.1014240964301001e-05,
+      "loss": 0.6909,
+      "step": 7369
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3687315575247472,
+      "learning_rate": 1.1009066824142713e-05,
+      "loss": 0.6447,
+      "step": 7370
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3380493437565399,
+      "learning_rate": 1.1003893438043138e-05,
+      "loss": 0.6657,
+      "step": 7371
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3656318232172675,
+      "learning_rate": 1.0998720806436149e-05,
+      "loss": 0.6527,
+      "step": 7372
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3689586061984745,
+      "learning_rate": 1.0993548929755576e-05,
+      "loss": 0.6744,
+      "step": 7373
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3686092477739446,
+      "learning_rate": 1.0988377808435184e-05,
+      "loss": 0.6768,
+      "step": 7374
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4023022095982876,
+      "learning_rate": 1.0983207442908654e-05,
+      "loss": 0.6618,
+      "step": 7375
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.36141652845167554,
+      "learning_rate": 1.0978037833609627e-05,
+      "loss": 0.643,
+      "step": 7376
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4330166388249478,
+      "learning_rate": 1.0972868980971669e-05,
+      "loss": 0.6358,
+      "step": 7377
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.34927521061516276,
+      "learning_rate": 1.096770088542829e-05,
+      "loss": 0.6255,
+      "step": 7378
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4033574768865578,
+      "learning_rate": 1.0962533547412937e-05,
+      "loss": 0.6409,
+      "step": 7379
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3797153822464957,
+      "learning_rate": 1.0957366967358971e-05,
+      "loss": 0.6828,
+      "step": 7380
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3623081741554146,
+      "learning_rate": 1.0952201145699719e-05,
+      "loss": 0.6317,
+      "step": 7381
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3573161028645841,
+      "learning_rate": 1.0947036082868427e-05,
+      "loss": 0.6438,
+      "step": 7382
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3817306815334202,
+      "learning_rate": 1.0941871779298289e-05,
+      "loss": 0.6832,
+      "step": 7383
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.33673125601307224,
+      "learning_rate": 1.0936708235422425e-05,
+      "loss": 0.6368,
+      "step": 7384
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3951720184273603,
+      "learning_rate": 1.0931545451673902e-05,
+      "loss": 0.6678,
+      "step": 7385
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3333221007305672,
+      "learning_rate": 1.0926383428485704e-05,
+      "loss": 0.65,
+      "step": 7386
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.39149402773889286,
+      "learning_rate": 1.0921222166290768e-05,
+      "loss": 0.6459,
+      "step": 7387
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3819372646376782,
+      "learning_rate": 1.0916061665521971e-05,
+      "loss": 0.6214,
+      "step": 7388
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4124165982477935,
+      "learning_rate": 1.0910901926612105e-05,
+      "loss": 0.6479,
+      "step": 7389
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.41098238886040017,
+      "learning_rate": 1.090574294999391e-05,
+      "loss": 0.6861,
+      "step": 7390
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3482791236391959,
+      "learning_rate": 1.0900584736100086e-05,
+      "loss": 0.625,
+      "step": 7391
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.37803742081863917,
+      "learning_rate": 1.0895427285363224e-05,
+      "loss": 0.6619,
+      "step": 7392
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3018888107579745,
+      "learning_rate": 1.089027059821588e-05,
+      "loss": 0.5878,
+      "step": 7393
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.37110112694636377,
+      "learning_rate": 1.0885114675090546e-05,
+      "loss": 0.6961,
+      "step": 7394
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4091175258745423,
+      "learning_rate": 1.0879959516419629e-05,
+      "loss": 0.6595,
+      "step": 7395
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.32668413717837885,
+      "learning_rate": 1.0874805122635492e-05,
+      "loss": 0.6139,
+      "step": 7396
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.39775453142010797,
+      "learning_rate": 1.086965149417043e-05,
+      "loss": 0.6692,
+      "step": 7397
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3714230630939929,
+      "learning_rate": 1.0864498631456672e-05,
+      "loss": 0.6619,
+      "step": 7398
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3212370502201813,
+      "learning_rate": 1.0859346534926387e-05,
+      "loss": 0.6296,
+      "step": 7399
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4087949783218244,
+      "learning_rate": 1.0854195205011662e-05,
+      "loss": 0.6446,
+      "step": 7400
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.36468834127600536,
+      "learning_rate": 1.084904464214454e-05,
+      "loss": 0.6154,
+      "step": 7401
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.36535721128189874,
+      "learning_rate": 1.0843894846756999e-05,
+      "loss": 0.6665,
+      "step": 7402
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.39880892625567604,
+      "learning_rate": 1.083874581928093e-05,
+      "loss": 0.6744,
+      "step": 7403
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3438038175534189,
+      "learning_rate": 1.0833597560148181e-05,
+      "loss": 0.6669,
+      "step": 7404
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.36234928707556197,
+      "learning_rate": 1.0828450069790546e-05,
+      "loss": 0.6566,
+      "step": 7405
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.42686695226882737,
+      "learning_rate": 1.0823303348639725e-05,
+      "loss": 0.6972,
+      "step": 7406
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3838881638858008,
+      "learning_rate": 1.0818157397127365e-05,
+      "loss": 0.6628,
+      "step": 7407
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3766848973173203,
+      "learning_rate": 1.0813012215685064e-05,
+      "loss": 0.6544,
+      "step": 7408
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.367406289584331,
+      "learning_rate": 1.0807867804744326e-05,
+      "loss": 0.6558,
+      "step": 7409
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3707024611129527,
+      "learning_rate": 1.0802724164736614e-05,
+      "loss": 0.666,
+      "step": 7410
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.37914581280627035,
+      "learning_rate": 1.0797581296093317e-05,
+      "loss": 0.6514,
+      "step": 7411
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.36145547522050625,
+      "learning_rate": 1.0792439199245766e-05,
+      "loss": 0.6378,
+      "step": 7412
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.39210528982276405,
+      "learning_rate": 1.0787297874625216e-05,
+      "loss": 0.6301,
+      "step": 7413
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.31622591971048075,
+      "learning_rate": 1.0782157322662879e-05,
+      "loss": 0.6431,
+      "step": 7414
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3852191274729132,
+      "learning_rate": 1.0777017543789863e-05,
+      "loss": 0.6481,
+      "step": 7415
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4278315294275441,
+      "learning_rate": 1.0771878538437251e-05,
+      "loss": 0.6934,
+      "step": 7416
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.828043487764878,
+      "learning_rate": 1.0766740307036039e-05,
+      "loss": 6.6748,
+      "step": 7417
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3369271417228628,
+      "learning_rate": 1.0761602850017167e-05,
+      "loss": 0.6258,
+      "step": 7418
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.36401311999065206,
+      "learning_rate": 1.0756466167811514e-05,
+      "loss": 0.6633,
+      "step": 7419
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.35717617436971205,
+      "learning_rate": 1.0751330260849873e-05,
+      "loss": 0.6503,
+      "step": 7420
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.31584611500152066,
+      "learning_rate": 1.0746195129562993e-05,
+      "loss": 0.624,
+      "step": 7421
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.2179378266334655,
+      "learning_rate": 1.074106077438156e-05,
+      "loss": 6.8791,
+      "step": 7422
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.37600287353700756,
+      "learning_rate": 1.0735927195736169e-05,
+      "loss": 0.6623,
+      "step": 7423
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.7300685887578684,
+      "learning_rate": 1.0730794394057372e-05,
+      "loss": 6.9333,
+      "step": 7424
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.35381613347244206,
+      "learning_rate": 1.0725662369775666e-05,
+      "loss": 0.6453,
+      "step": 7425
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.41160545376736624,
+      "learning_rate": 1.072053112332145e-05,
+      "loss": 0.6627,
+      "step": 7426
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.40656891875298,
+      "learning_rate": 1.071540065512508e-05,
+      "loss": 0.6224,
+      "step": 7427
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.38960805215437533,
+      "learning_rate": 1.0710270965616856e-05,
+      "loss": 0.6515,
+      "step": 7428
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3247536935014724,
+      "learning_rate": 1.0705142055226979e-05,
+      "loss": 0.6294,
+      "step": 7429
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3943346964734162,
+      "learning_rate": 1.0700013924385609e-05,
+      "loss": 0.6609,
+      "step": 7430
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.35950605717428386,
+      "learning_rate": 1.069488657352284e-05,
+      "loss": 0.6433,
+      "step": 7431
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.31832524953744795,
+      "learning_rate": 1.0689760003068698e-05,
+      "loss": 0.6749,
+      "step": 7432
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3483866254507419,
+      "learning_rate": 1.0684634213453149e-05,
+      "loss": 0.6219,
+      "step": 7433
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3476435603250195,
+      "learning_rate": 1.0679509205106068e-05,
+      "loss": 0.6225,
+      "step": 7434
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4181030288041998,
+      "learning_rate": 1.0674384978457297e-05,
+      "loss": 0.6518,
+      "step": 7435
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3310574388433627,
+      "learning_rate": 1.0669261533936601e-05,
+      "loss": 0.6232,
+      "step": 7436
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.42812617829578453,
+      "learning_rate": 1.0664138871973664e-05,
+      "loss": 0.7174,
+      "step": 7437
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4069145062794461,
+      "learning_rate": 1.0659016992998116e-05,
+      "loss": 0.6337,
+      "step": 7438
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.37240031587815664,
+      "learning_rate": 1.065389589743955e-05,
+      "loss": 0.6693,
+      "step": 7439
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4396238408306206,
+      "learning_rate": 1.0648775585727437e-05,
+      "loss": 0.6703,
+      "step": 7440
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.31971254860537546,
+      "learning_rate": 1.0643656058291226e-05,
+      "loss": 0.6291,
+      "step": 7441
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.981675137837048,
+      "learning_rate": 1.0638537315560286e-05,
+      "loss": 6.8125,
+      "step": 7442
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.371808085788058,
+      "learning_rate": 1.063341935796391e-05,
+      "loss": 0.6763,
+      "step": 7443
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3510090436471596,
+      "learning_rate": 1.0628302185931338e-05,
+      "loss": 0.6374,
+      "step": 7444
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40887535284908627,
+      "learning_rate": 1.0623185799891746e-05,
+      "loss": 0.685,
+      "step": 7445
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.37106583658867043,
+      "learning_rate": 1.0618070200274237e-05,
+      "loss": 0.6023,
+      "step": 7446
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.365508239833903,
+      "learning_rate": 1.0612955387507848e-05,
+      "loss": 0.6609,
+      "step": 7447
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 3.3821239631029103,
+      "learning_rate": 1.0607841362021562e-05,
+      "loss": 6.8602,
+      "step": 7448
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3953989824380062,
+      "learning_rate": 1.0602728124244273e-05,
+      "loss": 0.6447,
+      "step": 7449
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.35508598200951935,
+      "learning_rate": 1.0597615674604821e-05,
+      "loss": 0.6502,
+      "step": 7450
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3654990671458969,
+      "learning_rate": 1.0592504013532002e-05,
+      "loss": 0.6524,
+      "step": 7451
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3853983388548925,
+      "learning_rate": 1.0587393141454489e-05,
+      "loss": 0.6467,
+      "step": 7452
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3627133670147055,
+      "learning_rate": 1.0582283058800967e-05,
+      "loss": 0.6306,
+      "step": 7453
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4157788754250255,
+      "learning_rate": 1.0577173765999977e-05,
+      "loss": 0.6621,
+      "step": 7454
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.38077524964736387,
+      "learning_rate": 1.0572065263480046e-05,
+      "loss": 0.6379,
+      "step": 7455
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3656701817663183,
+      "learning_rate": 1.0566957551669622e-05,
+      "loss": 0.6743,
+      "step": 7456
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.33996267033570854,
+      "learning_rate": 1.0561850630997073e-05,
+      "loss": 0.6493,
+      "step": 7457
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3508668963676564,
+      "learning_rate": 1.0556744501890708e-05,
+      "loss": 0.6334,
+      "step": 7458
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4065767887807208,
+      "learning_rate": 1.055163916477878e-05,
+      "loss": 0.6376,
+      "step": 7459
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40314807169269584,
+      "learning_rate": 1.0546534620089464e-05,
+      "loss": 0.6673,
+      "step": 7460
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3869217174645483,
+      "learning_rate": 1.0541430868250871e-05,
+      "loss": 0.6529,
+      "step": 7461
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4105861783693164,
+      "learning_rate": 1.0536327909691057e-05,
+      "loss": 0.6643,
+      "step": 7462
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.45971891789067393,
+      "learning_rate": 1.0531225744837984e-05,
+      "loss": 0.6746,
+      "step": 7463
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.34095563396461015,
+      "learning_rate": 1.052612437411957e-05,
+      "loss": 0.6096,
+      "step": 7464
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.44416207871750807,
+      "learning_rate": 1.0521023797963674e-05,
+      "loss": 0.6726,
+      "step": 7465
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3861091180846501,
+      "learning_rate": 1.0515924016798048e-05,
+      "loss": 0.6074,
+      "step": 7466
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.36969800531215885,
+      "learning_rate": 1.0510825031050435e-05,
+      "loss": 0.6713,
+      "step": 7467
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.39140859426608404,
+      "learning_rate": 1.050572684114846e-05,
+      "loss": 0.669,
+      "step": 7468
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.41662798924209193,
+      "learning_rate": 1.050062944751971e-05,
+      "loss": 0.6556,
+      "step": 7469
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4005416922272144,
+      "learning_rate": 1.0495532850591696e-05,
+      "loss": 0.6307,
+      "step": 7470
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.34281583968695517,
+      "learning_rate": 1.0490437050791868e-05,
+      "loss": 0.6346,
+      "step": 7471
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4012502467025569,
+      "learning_rate": 1.0485342048547596e-05,
+      "loss": 0.7013,
+      "step": 7472
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.44220866904268297,
+      "learning_rate": 1.0480247844286192e-05,
+      "loss": 0.6426,
+      "step": 7473
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.42229733455791685,
+      "learning_rate": 1.0475154438434908e-05,
+      "loss": 0.6579,
+      "step": 7474
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3480418962528168,
+      "learning_rate": 1.0470061831420915e-05,
+      "loss": 0.6375,
+      "step": 7475
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.32111234815124573,
+      "learning_rate": 1.0464970023671335e-05,
+      "loss": 0.6301,
+      "step": 7476
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40232667519443044,
+      "learning_rate": 1.0459879015613196e-05,
+      "loss": 0.6267,
+      "step": 7477
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4870674690552783,
+      "learning_rate": 1.0454788807673484e-05,
+      "loss": 0.6449,
+      "step": 7478
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.36068399086233816,
+      "learning_rate": 1.0449699400279114e-05,
+      "loss": 0.624,
+      "step": 7479
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3988005965581348,
+      "learning_rate": 1.0444610793856906e-05,
+      "loss": 0.6306,
+      "step": 7480
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.43230013126575556,
+      "learning_rate": 1.0439522988833656e-05,
+      "loss": 0.6929,
+      "step": 7481
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.41284700149299586,
+      "learning_rate": 1.043443598563608e-05,
+      "loss": 0.6691,
+      "step": 7482
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3836157717199145,
+      "learning_rate": 1.0429349784690795e-05,
+      "loss": 0.6738,
+      "step": 7483
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3751191970191105,
+      "learning_rate": 1.0424264386424383e-05,
+      "loss": 0.6313,
+      "step": 7484
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3346891707074948,
+      "learning_rate": 1.0419179791263362e-05,
+      "loss": 0.6602,
+      "step": 7485
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3819511667539261,
+      "learning_rate": 1.0414095999634153e-05,
+      "loss": 0.6677,
+      "step": 7486
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3488531147264126,
+      "learning_rate": 1.0409013011963136e-05,
+      "loss": 0.6355,
+      "step": 7487
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.39651543449626087,
+      "learning_rate": 1.0403930828676616e-05,
+      "loss": 0.6978,
+      "step": 7488
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40306307077418546,
+      "learning_rate": 1.0398849450200827e-05,
+      "loss": 0.6409,
+      "step": 7489
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3879764970576329,
+      "learning_rate": 1.0393768876961948e-05,
+      "loss": 0.6739,
+      "step": 7490
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.36625743600215893,
+      "learning_rate": 1.0388689109386063e-05,
+      "loss": 0.655,
+      "step": 7491
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3790403222533165,
+      "learning_rate": 1.0383610147899214e-05,
+      "loss": 0.6533,
+      "step": 7492
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3746633791863898,
+      "learning_rate": 1.0378531992927378e-05,
+      "loss": 0.6393,
+      "step": 7493
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.42588683730992644,
+      "learning_rate": 1.0373454644896428e-05,
+      "loss": 0.6576,
+      "step": 7494
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4212825019931983,
+      "learning_rate": 1.0368378104232217e-05,
+      "loss": 0.6852,
+      "step": 7495
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.38553179052987396,
+      "learning_rate": 1.0363302371360513e-05,
+      "loss": 0.6495,
+      "step": 7496
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3615060692394303,
+      "learning_rate": 1.035822744670699e-05,
+      "loss": 0.627,
+      "step": 7497
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3813965040839219,
+      "learning_rate": 1.035315333069729e-05,
+      "loss": 0.6713,
+      "step": 7498
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.38247096791723195,
+      "learning_rate": 1.0348080023756974e-05,
+      "loss": 0.6361,
+      "step": 7499
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40885227388225104,
+      "learning_rate": 1.0343007526311518e-05,
+      "loss": 0.6701,
+      "step": 7500
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3789830820495343,
+      "learning_rate": 1.0337935838786372e-05,
+      "loss": 0.6575,
+      "step": 7501
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.385607772754444,
+      "learning_rate": 1.033286496160687e-05,
+      "loss": 0.642,
+      "step": 7502
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4347267806690868,
+      "learning_rate": 1.032779489519831e-05,
+      "loss": 0.6862,
+      "step": 7503
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3913381250169739,
+      "learning_rate": 1.0322725639985908e-05,
+      "loss": 0.6717,
+      "step": 7504
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4189153379876509,
+      "learning_rate": 1.0317657196394833e-05,
+      "loss": 0.6535,
+      "step": 7505
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3707205893517312,
+      "learning_rate": 1.0312589564850142e-05,
+      "loss": 0.6513,
+      "step": 7506
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4172156076448639,
+      "learning_rate": 1.0307522745776868e-05,
+      "loss": 0.6578,
+      "step": 7507
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40194260671928994,
+      "learning_rate": 1.0302456739599953e-05,
+      "loss": 0.6465,
+      "step": 7508
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.42073428023626935,
+      "learning_rate": 1.0297391546744281e-05,
+      "loss": 0.7061,
+      "step": 7509
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3738168403197374,
+      "learning_rate": 1.0292327167634673e-05,
+      "loss": 0.6273,
+      "step": 7510
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3511263854018193,
+      "learning_rate": 1.0287263602695852e-05,
+      "loss": 0.6211,
+      "step": 7511
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3901513519892764,
+      "learning_rate": 1.02822008523525e-05,
+      "loss": 0.6654,
+      "step": 7512
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3557103362655701,
+      "learning_rate": 1.0277138917029235e-05,
+      "loss": 0.6866,
+      "step": 7513
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3991514476022457,
+      "learning_rate": 1.0272077797150572e-05,
+      "loss": 0.667,
+      "step": 7514
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.42811861529575596,
+      "learning_rate": 1.0267017493141005e-05,
+      "loss": 0.6728,
+      "step": 7515
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4335574590537662,
+      "learning_rate": 1.0261958005424932e-05,
+      "loss": 0.6712,
+      "step": 7516
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4006424462305724,
+      "learning_rate": 1.0256899334426675e-05,
+      "loss": 0.6524,
+      "step": 7517
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.38729987561926543,
+      "learning_rate": 1.0251841480570502e-05,
+      "loss": 0.6475,
+      "step": 7518
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.36661332682425807,
+      "learning_rate": 1.024678444428062e-05,
+      "loss": 0.6483,
+      "step": 7519
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3722475924904536,
+      "learning_rate": 1.0241728225981139e-05,
+      "loss": 0.6293,
+      "step": 7520
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.39869630025940117,
+      "learning_rate": 1.0236672826096127e-05,
+      "loss": 0.6704,
+      "step": 7521
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3890599390640179,
+      "learning_rate": 1.0231618245049576e-05,
+      "loss": 0.6522,
+      "step": 7522
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.39350070052673475,
+      "learning_rate": 1.0226564483265405e-05,
+      "loss": 0.6691,
+      "step": 7523
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.37306897989413706,
+      "learning_rate": 1.0221511541167473e-05,
+      "loss": 0.6881,
+      "step": 7524
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4081025397980068,
+      "learning_rate": 1.0216459419179552e-05,
+      "loss": 0.6538,
+      "step": 7525
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4340126275759106,
+      "learning_rate": 1.0211408117725362e-05,
+      "loss": 0.6655,
+      "step": 7526
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3792143237563521,
+      "learning_rate": 1.0206357637228555e-05,
+      "loss": 0.6527,
+      "step": 7527
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4354911868191083,
+      "learning_rate": 1.0201307978112704e-05,
+      "loss": 0.6531,
+      "step": 7528
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3474376500923559,
+      "learning_rate": 1.0196259140801317e-05,
+      "loss": 0.6158,
+      "step": 7529
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3846246766378743,
+      "learning_rate": 1.0191211125717846e-05,
+      "loss": 0.6827,
+      "step": 7530
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3249858417670381,
+      "learning_rate": 1.0186163933285644e-05,
+      "loss": 0.6313,
+      "step": 7531
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.35835366296612103,
+      "learning_rate": 1.018111756392802e-05,
+      "loss": 0.6511,
+      "step": 7532
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4009454576197942,
+      "learning_rate": 1.0176072018068216e-05,
+      "loss": 0.6799,
+      "step": 7533
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4001307246355026,
+      "learning_rate": 1.0171027296129381e-05,
+      "loss": 0.6891,
+      "step": 7534
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3765325608654193,
+      "learning_rate": 1.0165983398534606e-05,
+      "loss": 0.6362,
+      "step": 7535
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.330995643020246,
+      "learning_rate": 1.0160940325706946e-05,
+      "loss": 0.6422,
+      "step": 7536
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.34039086037247374,
+      "learning_rate": 1.0155898078069326e-05,
+      "loss": 0.5931,
+      "step": 7537
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.350565549175436,
+      "learning_rate": 1.015085665604465e-05,
+      "loss": 0.6637,
+      "step": 7538
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.35573168843187664,
+      "learning_rate": 1.0145816060055736e-05,
+      "loss": 0.6455,
+      "step": 7539
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.37683855026893276,
+      "learning_rate": 1.014077629052532e-05,
+      "loss": 0.6163,
+      "step": 7540
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3909830116464924,
+      "learning_rate": 1.013573734787609e-05,
+      "loss": 0.671,
+      "step": 7541
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.34833549478944326,
+      "learning_rate": 1.013069923253066e-05,
+      "loss": 0.6185,
+      "step": 7542
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3783181660887292,
+      "learning_rate": 1.0125661944911562e-05,
+      "loss": 0.6568,
+      "step": 7543
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 7.65829273597354,
+      "learning_rate": 1.0120625485441278e-05,
+      "loss": 6.8091,
+      "step": 7544
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.44675100922244637,
+      "learning_rate": 1.0115589854542201e-05,
+      "loss": 0.6546,
+      "step": 7545
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.34434697495015654,
+      "learning_rate": 1.011055505263666e-05,
+      "loss": 0.664,
+      "step": 7546
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.36796119482259315,
+      "learning_rate": 1.0105521080146934e-05,
+      "loss": 0.6116,
+      "step": 7547
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4130402483423393,
+      "learning_rate": 1.0100487937495197e-05,
+      "loss": 0.6515,
+      "step": 7548
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.36680024868043243,
+      "learning_rate": 1.0095455625103571e-05,
+      "loss": 0.6287,
+      "step": 7549
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3684604559629592,
+      "learning_rate": 1.009042414339414e-05,
+      "loss": 0.6634,
+      "step": 7550
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4311612918226518,
+      "learning_rate": 1.0085393492788857e-05,
+      "loss": 0.6346,
+      "step": 7551
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5203123057252329,
+      "learning_rate": 1.008036367370965e-05,
+      "loss": 0.7271,
+      "step": 7552
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39036332567711973,
+      "learning_rate": 1.007533468657837e-05,
+      "loss": 0.6303,
+      "step": 7553
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.43214914514293284,
+      "learning_rate": 1.0070306531816775e-05,
+      "loss": 0.6748,
+      "step": 7554
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.359150812754643,
+      "learning_rate": 1.006527920984658e-05,
+      "loss": 0.646,
+      "step": 7555
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.43682988985449456,
+      "learning_rate": 1.0060252721089419e-05,
+      "loss": 0.6842,
+      "step": 7556
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3501656369577353,
+      "learning_rate": 1.0055227065966857e-05,
+      "loss": 0.6209,
+      "step": 7557
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.35009510584264425,
+      "learning_rate": 1.00502022449004e-05,
+      "loss": 0.6298,
+      "step": 7558
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4325593301392653,
+      "learning_rate": 1.0045178258311457e-05,
+      "loss": 0.6996,
+      "step": 7559
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3945815196786347,
+      "learning_rate": 1.0040155106621388e-05,
+      "loss": 0.661,
+      "step": 7560
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5391605344684963,
+      "learning_rate": 1.0035132790251485e-05,
+      "loss": 0.6636,
+      "step": 7561
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.34863197488528463,
+      "learning_rate": 1.0030111309622967e-05,
+      "loss": 0.6235,
+      "step": 7562
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3255309340989786,
+      "learning_rate": 1.0025090665156958e-05,
+      "loss": 0.6314,
+      "step": 7563
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.37249373129638247,
+      "learning_rate": 1.0020070857274562e-05,
+      "loss": 0.6442,
+      "step": 7564
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.1775870021224515,
+      "learning_rate": 1.0015051886396765e-05,
+      "loss": 0.6637,
+      "step": 7565
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4314120148389101,
+      "learning_rate": 1.0010033752944506e-05,
+      "loss": 0.6599,
+      "step": 7566
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4204131424446657,
+      "learning_rate": 1.0005016457338659e-05,
+      "loss": 0.6166,
+      "step": 7567
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4121512626863889,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.6537,
+      "step": 7568
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.35222968993033227,
+      "learning_rate": 9.994984381349272e-06,
+      "loss": 0.6573,
+      "step": 7569
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.43614336398475884,
+      "learning_rate": 9.989969601807117e-06,
+      "loss": 0.6685,
+      "step": 7570
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4120967701709341,
+      "learning_rate": 9.984955661794123e-06,
+      "loss": 0.6974,
+      "step": 7571
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4134579225031956,
+      "learning_rate": 9.979942561730803e-06,
+      "loss": 0.6367,
+      "step": 7572
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.34240369847610647,
+      "learning_rate": 9.974930302037604e-06,
+      "loss": 0.6642,
+      "step": 7573
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.36293933533126915,
+      "learning_rate": 9.969918883134886e-06,
+      "loss": 0.6563,
+      "step": 7574
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3242021905719639,
+      "learning_rate": 9.96490830544296e-06,
+      "loss": 0.5898,
+      "step": 7575
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3674167626356592,
+      "learning_rate": 9.959898569382063e-06,
+      "loss": 0.6454,
+      "step": 7576
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.36077202475290987,
+      "learning_rate": 9.95488967537233e-06,
+      "loss": 0.6685,
+      "step": 7577
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39750479081386997,
+      "learning_rate": 9.949881623833886e-06,
+      "loss": 0.6516,
+      "step": 7578
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.35264701941158616,
+      "learning_rate": 9.944874415186722e-06,
+      "loss": 0.646,
+      "step": 7579
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3996510257512833,
+      "learning_rate": 9.9398680498508e-06,
+      "loss": 0.641,
+      "step": 7580
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3793423036255947,
+      "learning_rate": 9.934862528246004e-06,
+      "loss": 0.635,
+      "step": 7581
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3726141512581794,
+      "learning_rate": 9.929857850792122e-06,
+      "loss": 0.6649,
+      "step": 7582
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3965487247320373,
+      "learning_rate": 9.924854017908893e-06,
+      "loss": 0.677,
+      "step": 7583
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.38328195105339496,
+      "learning_rate": 9.919851030016006e-06,
+      "loss": 0.6268,
+      "step": 7584
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.40403173261892733,
+      "learning_rate": 9.914848887533031e-06,
+      "loss": 0.694,
+      "step": 7585
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3819596927414457,
+      "learning_rate": 9.9098475908795e-06,
+      "loss": 0.6437,
+      "step": 7586
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3524771952197245,
+      "learning_rate": 9.904847140474874e-06,
+      "loss": 0.6206,
+      "step": 7587
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4035816974886187,
+      "learning_rate": 9.899847536738519e-06,
+      "loss": 0.7116,
+      "step": 7588
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.37655179448195525,
+      "learning_rate": 9.894848780089752e-06,
+      "loss": 0.659,
+      "step": 7589
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.38057532625714774,
+      "learning_rate": 9.889850870947814e-06,
+      "loss": 0.6493,
+      "step": 7590
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3545512901461599,
+      "learning_rate": 9.884853809731875e-06,
+      "loss": 0.5954,
+      "step": 7591
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.38673926769351635,
+      "learning_rate": 9.879857596861032e-06,
+      "loss": 0.6714,
+      "step": 7592
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39451654443355294,
+      "learning_rate": 9.874862232754314e-06,
+      "loss": 0.683,
+      "step": 7593
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.36578567990435934,
+      "learning_rate": 9.869867717830668e-06,
+      "loss": 0.6572,
+      "step": 7594
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3438061729196846,
+      "learning_rate": 9.864874052508982e-06,
+      "loss": 0.6381,
+      "step": 7595
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.30862732465416176,
+      "learning_rate": 9.859881237208075e-06,
+      "loss": 0.6508,
+      "step": 7596
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39656556886901273,
+      "learning_rate": 9.854889272346669e-06,
+      "loss": 0.7003,
+      "step": 7597
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.33427856561755553,
+      "learning_rate": 9.84989815834346e-06,
+      "loss": 0.6365,
+      "step": 7598
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 12.30706013814069,
+      "learning_rate": 9.844907895617028e-06,
+      "loss": 7.2776,
+      "step": 7599
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3550272874154421,
+      "learning_rate": 9.839918484585909e-06,
+      "loss": 0.6694,
+      "step": 7600
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.2618094311300045,
+      "learning_rate": 9.834929925668561e-06,
+      "loss": 6.813,
+      "step": 7601
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3711559914986435,
+      "learning_rate": 9.829942219283358e-06,
+      "loss": 0.6797,
+      "step": 7602
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3216302393964919,
+      "learning_rate": 9.82495536584862e-06,
+      "loss": 0.628,
+      "step": 7603
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3743943053532932,
+      "learning_rate": 9.819969365782582e-06,
+      "loss": 0.6731,
+      "step": 7604
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.41418289739441216,
+      "learning_rate": 9.814984219503423e-06,
+      "loss": 0.6683,
+      "step": 7605
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.37051334154114957,
+      "learning_rate": 9.809999927429237e-06,
+      "loss": 0.6428,
+      "step": 7606
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.342768160452448,
+      "learning_rate": 9.805016489978057e-06,
+      "loss": 0.6572,
+      "step": 7607
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4296059323210193,
+      "learning_rate": 9.800033907567825e-06,
+      "loss": 0.6691,
+      "step": 7608
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3867826443855781,
+      "learning_rate": 9.795052180616431e-06,
+      "loss": 0.6537,
+      "step": 7609
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3787936315424498,
+      "learning_rate": 9.790071309541692e-06,
+      "loss": 0.6642,
+      "step": 7610
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3998999935987391,
+      "learning_rate": 9.785091294761329e-06,
+      "loss": 0.6407,
+      "step": 7611
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.41669678830358337,
+      "learning_rate": 9.780112136693036e-06,
+      "loss": 0.6613,
+      "step": 7612
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.36380947748376136,
+      "learning_rate": 9.77513383575439e-06,
+      "loss": 0.6624,
+      "step": 7613
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4413858069534179,
+      "learning_rate": 9.770156392362917e-06,
+      "loss": 0.6719,
+      "step": 7614
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.40794003057689293,
+      "learning_rate": 9.765179806936083e-06,
+      "loss": 0.6537,
+      "step": 7615
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.45444452095258914,
+      "learning_rate": 9.76020407989125e-06,
+      "loss": 0.6819,
+      "step": 7616
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.34719483048833855,
+      "learning_rate": 9.755229211645734e-06,
+      "loss": 0.624,
+      "step": 7617
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3723473389815005,
+      "learning_rate": 9.750255202616772e-06,
+      "loss": 0.6522,
+      "step": 7618
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3634772361945422,
+      "learning_rate": 9.745282053221527e-06,
+      "loss": 0.6265,
+      "step": 7619
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3579103031426103,
+      "learning_rate": 9.740309763877092e-06,
+      "loss": 0.6356,
+      "step": 7620
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4135033161444818,
+      "learning_rate": 9.735338335000494e-06,
+      "loss": 0.6492,
+      "step": 7621
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4143998015106742,
+      "learning_rate": 9.730367767008667e-06,
+      "loss": 0.6351,
+      "step": 7622
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.41703331553881573,
+      "learning_rate": 9.725398060318491e-06,
+      "loss": 0.6645,
+      "step": 7623
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4233738892564662,
+      "learning_rate": 9.72042921534678e-06,
+      "loss": 0.6698,
+      "step": 7624
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.35523219827720864,
+      "learning_rate": 9.71546123251024e-06,
+      "loss": 0.6364,
+      "step": 7625
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3511141834597604,
+      "learning_rate": 9.710494112225555e-06,
+      "loss": 0.6275,
+      "step": 7626
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3696493686474533,
+      "learning_rate": 9.705527854909306e-06,
+      "loss": 0.6571,
+      "step": 7627
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4104992399813378,
+      "learning_rate": 9.700562460978001e-06,
+      "loss": 0.6348,
+      "step": 7628
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.34900998803803296,
+      "learning_rate": 9.69559793084808e-06,
+      "loss": 0.6297,
+      "step": 7629
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4100891202773554,
+      "learning_rate": 9.690634264935925e-06,
+      "loss": 0.6587,
+      "step": 7630
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39031627117568884,
+      "learning_rate": 9.685671463657816e-06,
+      "loss": 0.629,
+      "step": 7631
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.40251245146790193,
+      "learning_rate": 9.680709527429983e-06,
+      "loss": 0.6683,
+      "step": 7632
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4222723617432094,
+      "learning_rate": 9.675748456668582e-06,
+      "loss": 0.6927,
+      "step": 7633
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3683036539887188,
+      "learning_rate": 9.670788251789686e-06,
+      "loss": 0.6105,
+      "step": 7634
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4498166735689307,
+      "learning_rate": 9.665828913209317e-06,
+      "loss": 0.6471,
+      "step": 7635
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.40194886499501614,
+      "learning_rate": 9.660870441343387e-06,
+      "loss": 0.6203,
+      "step": 7636
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3741249485703623,
+      "learning_rate": 9.655912836607765e-06,
+      "loss": 0.6215,
+      "step": 7637
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4280094808424291,
+      "learning_rate": 9.65095609941825e-06,
+      "loss": 0.6871,
+      "step": 7638
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3727043127696918,
+      "learning_rate": 9.646000230190533e-06,
+      "loss": 0.6381,
+      "step": 7639
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.36534356590184786,
+      "learning_rate": 9.64104522934028e-06,
+      "loss": 0.6554,
+      "step": 7640
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3540525428584335,
+      "learning_rate": 9.63609109728306e-06,
+      "loss": 0.6594,
+      "step": 7641
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3322977837589779,
+      "learning_rate": 9.631137834434357e-06,
+      "loss": 0.6279,
+      "step": 7642
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.43152879723327275,
+      "learning_rate": 9.6261854412096e-06,
+      "loss": 0.6256,
+      "step": 7643
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 3.202106753371545,
+      "learning_rate": 9.621233918024148e-06,
+      "loss": 7.1039,
+      "step": 7644
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5265389780135219,
+      "learning_rate": 9.616283265293269e-06,
+      "loss": 0.6868,
+      "step": 7645
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39567089065560845,
+      "learning_rate": 9.61133348343217e-06,
+      "loss": 0.6576,
+      "step": 7646
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3748118279327276,
+      "learning_rate": 9.606384572855985e-06,
+      "loss": 0.6853,
+      "step": 7647
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3629266714487879,
+      "learning_rate": 9.601436533979776e-06,
+      "loss": 0.6712,
+      "step": 7648
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39308420606066724,
+      "learning_rate": 9.596489367218524e-06,
+      "loss": 0.6428,
+      "step": 7649
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39450381082735697,
+      "learning_rate": 9.591543072987152e-06,
+      "loss": 0.6839,
+      "step": 7650
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.40989662638144375,
+      "learning_rate": 9.586597651700486e-06,
+      "loss": 0.6614,
+      "step": 7651
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3435858672871077,
+      "learning_rate": 9.581653103773298e-06,
+      "loss": 0.6327,
+      "step": 7652
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3678085444623384,
+      "learning_rate": 9.576709429620283e-06,
+      "loss": 0.6552,
+      "step": 7653
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4193738822471785,
+      "learning_rate": 9.571766629656059e-06,
+      "loss": 0.6633,
+      "step": 7654
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.365943717970494,
+      "learning_rate": 9.566824704295182e-06,
+      "loss": 0.6465,
+      "step": 7655
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.35013481993683093,
+      "learning_rate": 9.561883653952108e-06,
+      "loss": 0.6469,
+      "step": 7656
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.35873384238169576,
+      "learning_rate": 9.556943479041248e-06,
+      "loss": 0.6508,
+      "step": 7657
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39977061976807043,
+      "learning_rate": 9.552004179976932e-06,
+      "loss": 0.6328,
+      "step": 7658
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4012064295664943,
+      "learning_rate": 9.547065757173402e-06,
+      "loss": 0.6456,
+      "step": 7659
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3626225617224645,
+      "learning_rate": 9.542128211044835e-06,
+      "loss": 0.6464,
+      "step": 7660
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.35046404891611727,
+      "learning_rate": 9.537191542005357e-06,
+      "loss": 0.634,
+      "step": 7661
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3946963155380747,
+      "learning_rate": 9.532255750468984e-06,
+      "loss": 0.6268,
+      "step": 7662
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4050578951980934,
+      "learning_rate": 9.527320836849679e-06,
+      "loss": 0.6781,
+      "step": 7663
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.41636273171500593,
+      "learning_rate": 9.522386801561334e-06,
+      "loss": 0.6923,
+      "step": 7664
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4086150091731894,
+      "learning_rate": 9.517453645017749e-06,
+      "loss": 0.6692,
+      "step": 7665
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.347937706346889,
+      "learning_rate": 9.512521367632665e-06,
+      "loss": 0.6264,
+      "step": 7666
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.41712229285358615,
+      "learning_rate": 9.507589969819748e-06,
+      "loss": 0.6553,
+      "step": 7667
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.35403044777827486,
+      "learning_rate": 9.50265945199259e-06,
+      "loss": 0.6397,
+      "step": 7668
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.35261967354018153,
+      "learning_rate": 9.497729814564714e-06,
+      "loss": 0.641,
+      "step": 7669
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.43578381535170907,
+      "learning_rate": 9.492801057949545e-06,
+      "loss": 0.6582,
+      "step": 7670
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3681760883143675,
+      "learning_rate": 9.487873182560463e-06,
+      "loss": 0.6271,
+      "step": 7671
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.35736420580901557,
+      "learning_rate": 9.48294618881077e-06,
+      "loss": 0.6576,
+      "step": 7672
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.34870431299896804,
+      "learning_rate": 9.478020077113664e-06,
+      "loss": 0.6499,
+      "step": 7673
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4359576364082482,
+      "learning_rate": 9.473094847882314e-06,
+      "loss": 0.6522,
+      "step": 7674
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3965495057660577,
+      "learning_rate": 9.468170501529792e-06,
+      "loss": 0.6869,
+      "step": 7675
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.31881962123856145,
+      "learning_rate": 9.463247038469083e-06,
+      "loss": 0.5856,
+      "step": 7676
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3816812401886499,
+      "learning_rate": 9.458324459113123e-06,
+      "loss": 0.6499,
+      "step": 7677
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1206529968758003,
+      "learning_rate": 9.453402763874763e-06,
+      "loss": 7.0566,
+      "step": 7678
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4056703716289023,
+      "learning_rate": 9.448481953166773e-06,
+      "loss": 0.6495,
+      "step": 7679
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.38713350884915443,
+      "learning_rate": 9.443562027401856e-06,
+      "loss": 0.6626,
+      "step": 7680
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40038771140218604,
+      "learning_rate": 9.438642986992641e-06,
+      "loss": 0.6765,
+      "step": 7681
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4068620959488255,
+      "learning_rate": 9.433724832351688e-06,
+      "loss": 0.6398,
+      "step": 7682
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40840149817651406,
+      "learning_rate": 9.428807563891469e-06,
+      "loss": 0.672,
+      "step": 7683
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4290723817660468,
+      "learning_rate": 9.4238911820244e-06,
+      "loss": 0.6835,
+      "step": 7684
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3327556208698933,
+      "learning_rate": 9.4189756871628e-06,
+      "loss": 0.6314,
+      "step": 7685
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.36260434903226535,
+      "learning_rate": 9.41406107971893e-06,
+      "loss": 0.6299,
+      "step": 7686
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.38059283715403436,
+      "learning_rate": 9.409147360104971e-06,
+      "loss": 0.6286,
+      "step": 7687
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3382444855777491,
+      "learning_rate": 9.404234528733035e-06,
+      "loss": 0.6248,
+      "step": 7688
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3687264988988909,
+      "learning_rate": 9.399322586015159e-06,
+      "loss": 0.6516,
+      "step": 7689
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3997879076064258,
+      "learning_rate": 9.394411532363285e-06,
+      "loss": 0.7057,
+      "step": 7690
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.35977730058254814,
+      "learning_rate": 9.389501368189311e-06,
+      "loss": 0.6125,
+      "step": 7691
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3706806810410361,
+      "learning_rate": 9.384592093905049e-06,
+      "loss": 0.6376,
+      "step": 7692
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3463513657015901,
+      "learning_rate": 9.379683709922221e-06,
+      "loss": 0.6767,
+      "step": 7693
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.34156791664791797,
+      "learning_rate": 9.374776216652486e-06,
+      "loss": 0.6471,
+      "step": 7694
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3790790496617597,
+      "learning_rate": 9.369869614507455e-06,
+      "loss": 0.6324,
+      "step": 7695
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3898254517573087,
+      "learning_rate": 9.364963903898612e-06,
+      "loss": 0.6838,
+      "step": 7696
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3684642625265979,
+      "learning_rate": 9.360059085237401e-06,
+      "loss": 0.6696,
+      "step": 7697
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.379415543763342,
+      "learning_rate": 9.355155158935194e-06,
+      "loss": 0.6833,
+      "step": 7698
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40146148827021627,
+      "learning_rate": 9.350252125403261e-06,
+      "loss": 0.6546,
+      "step": 7699
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40418607262091666,
+      "learning_rate": 9.345349985052821e-06,
+      "loss": 0.6385,
+      "step": 7700
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40836813756243145,
+      "learning_rate": 9.340448738295009e-06,
+      "loss": 0.694,
+      "step": 7701
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3174300801920346,
+      "learning_rate": 9.335548385540887e-06,
+      "loss": 0.6258,
+      "step": 7702
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.43531720055475875,
+      "learning_rate": 9.330648927201451e-06,
+      "loss": 0.6731,
+      "step": 7703
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.38726667141772503,
+      "learning_rate": 9.325750363687599e-06,
+      "loss": 0.6528,
+      "step": 7704
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40007401932779635,
+      "learning_rate": 9.32085269541017e-06,
+      "loss": 0.6194,
+      "step": 7705
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4065576582595193,
+      "learning_rate": 9.31595592277993e-06,
+      "loss": 0.6629,
+      "step": 7706
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3713418799594974,
+      "learning_rate": 9.311060046207567e-06,
+      "loss": 0.6672,
+      "step": 7707
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3842044310049704,
+      "learning_rate": 9.306165066103678e-06,
+      "loss": 0.6714,
+      "step": 7708
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.37188629229232717,
+      "learning_rate": 9.301270982878826e-06,
+      "loss": 0.6479,
+      "step": 7709
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.39590028306725056,
+      "learning_rate": 9.296377796943446e-06,
+      "loss": 0.6709,
+      "step": 7710
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4109484699793117,
+      "learning_rate": 9.291485508707932e-06,
+      "loss": 0.672,
+      "step": 7711
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.37723923212353844,
+      "learning_rate": 9.286594118582606e-06,
+      "loss": 0.6751,
+      "step": 7712
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.36933152282231624,
+      "learning_rate": 9.281703626977685e-06,
+      "loss": 0.6253,
+      "step": 7713
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3885543110394385,
+      "learning_rate": 9.276814034303334e-06,
+      "loss": 0.6562,
+      "step": 7714
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3924654429613331,
+      "learning_rate": 9.271925340969644e-06,
+      "loss": 0.6652,
+      "step": 7715
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3814317425647739,
+      "learning_rate": 9.267037547386615e-06,
+      "loss": 0.6386,
+      "step": 7716
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3836322417764117,
+      "learning_rate": 9.262150653964187e-06,
+      "loss": 0.6844,
+      "step": 7717
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4382800706541674,
+      "learning_rate": 9.257264661112222e-06,
+      "loss": 0.6536,
+      "step": 7718
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3628336474795638,
+      "learning_rate": 9.25237956924049e-06,
+      "loss": 0.6515,
+      "step": 7719
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.39180250434710856,
+      "learning_rate": 9.247495378758703e-06,
+      "loss": 0.6561,
+      "step": 7720
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3752942815507944,
+      "learning_rate": 9.242612090076499e-06,
+      "loss": 0.7077,
+      "step": 7721
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3921860780603663,
+      "learning_rate": 9.237729703603416e-06,
+      "loss": 0.6769,
+      "step": 7722
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3991098226246614,
+      "learning_rate": 9.232848219748956e-06,
+      "loss": 0.6539,
+      "step": 7723
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3844118944676503,
+      "learning_rate": 9.227967638922505e-06,
+      "loss": 0.687,
+      "step": 7724
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.48807677210147704,
+      "learning_rate": 9.223087961533402e-06,
+      "loss": 0.6788,
+      "step": 7725
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40230086219698324,
+      "learning_rate": 9.218209187990903e-06,
+      "loss": 0.6421,
+      "step": 7726
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.36771138367753153,
+      "learning_rate": 9.213331318704171e-06,
+      "loss": 0.6405,
+      "step": 7727
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.37912336887292,
+      "learning_rate": 9.208454354082312e-06,
+      "loss": 0.6362,
+      "step": 7728
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3875092673439245,
+      "learning_rate": 9.203578294534355e-06,
+      "loss": 0.6496,
+      "step": 7729
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.356554537859966,
+      "learning_rate": 9.198703140469247e-06,
+      "loss": 0.668,
+      "step": 7730
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3617442678985432,
+      "learning_rate": 9.193828892295864e-06,
+      "loss": 0.655,
+      "step": 7731
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 3.818416291973941,
+      "learning_rate": 9.188955550423007e-06,
+      "loss": 6.9988,
+      "step": 7732
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4204121523018086,
+      "learning_rate": 9.184083115259384e-06,
+      "loss": 0.6586,
+      "step": 7733
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40502774677675263,
+      "learning_rate": 9.179211587213648e-06,
+      "loss": 0.6347,
+      "step": 7734
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.37858144345954486,
+      "learning_rate": 9.174340966694375e-06,
+      "loss": 0.664,
+      "step": 7735
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.33544464923992134,
+      "learning_rate": 9.169471254110038e-06,
+      "loss": 0.6132,
+      "step": 7736
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3908754034980543,
+      "learning_rate": 9.16460244986908e-06,
+      "loss": 0.6542,
+      "step": 7737
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.39031014468898173,
+      "learning_rate": 9.159734554379823e-06,
+      "loss": 0.6801,
+      "step": 7738
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4052096977571526,
+      "learning_rate": 9.154867568050534e-06,
+      "loss": 0.6893,
+      "step": 7739
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40624973567403355,
+      "learning_rate": 9.150001491289406e-06,
+      "loss": 0.6772,
+      "step": 7740
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3840968935048381,
+      "learning_rate": 9.14513632450456e-06,
+      "loss": 0.6829,
+      "step": 7741
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3445384427415515,
+      "learning_rate": 9.14027206810401e-06,
+      "loss": 0.6093,
+      "step": 7742
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4066489566184601,
+      "learning_rate": 9.135408722495727e-06,
+      "loss": 0.6797,
+      "step": 7743
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.37007038327814923,
+      "learning_rate": 9.130546288087594e-06,
+      "loss": 0.6477,
+      "step": 7744
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.38393172245554685,
+      "learning_rate": 9.125684765287416e-06,
+      "loss": 0.6445,
+      "step": 7745
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3471758271393195,
+      "learning_rate": 9.120824154502933e-06,
+      "loss": 0.6499,
+      "step": 7746
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3983453903184549,
+      "learning_rate": 9.115964456141781e-06,
+      "loss": 0.6823,
+      "step": 7747
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4514547986756669,
+      "learning_rate": 9.111105670611545e-06,
+      "loss": 0.7251,
+      "step": 7748
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.259800178263635,
+      "learning_rate": 9.106247798319735e-06,
+      "loss": 6.8845,
+      "step": 7749
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3835771252512383,
+      "learning_rate": 9.101390839673749e-06,
+      "loss": 0.6544,
+      "step": 7750
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.39133770815116364,
+      "learning_rate": 9.09653479508096e-06,
+      "loss": 0.6751,
+      "step": 7751
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40272403192278244,
+      "learning_rate": 9.091679664948637e-06,
+      "loss": 0.6449,
+      "step": 7752
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3733715270975737,
+      "learning_rate": 9.086825449683958e-06,
+      "loss": 0.6603,
+      "step": 7753
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.34287807551058513,
+      "learning_rate": 9.08197214969405e-06,
+      "loss": 0.5997,
+      "step": 7754
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.35108009929551853,
+      "learning_rate": 9.077119765385958e-06,
+      "loss": 0.6513,
+      "step": 7755
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4336162347307827,
+      "learning_rate": 9.072268297166626e-06,
+      "loss": 0.6547,
+      "step": 7756
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.39152110181263233,
+      "learning_rate": 9.067417745442969e-06,
+      "loss": 0.6483,
+      "step": 7757
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40050487707907545,
+      "learning_rate": 9.062568110621774e-06,
+      "loss": 0.6625,
+      "step": 7758
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4022922908906547,
+      "learning_rate": 9.057719393109785e-06,
+      "loss": 0.6677,
+      "step": 7759
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3528214694671644,
+      "learning_rate": 9.05287159331366e-06,
+      "loss": 0.6466,
+      "step": 7760
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.37758011140828013,
+      "learning_rate": 9.048024711639968e-06,
+      "loss": 0.6797,
+      "step": 7761
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.351983874620918,
+      "learning_rate": 9.043178748495217e-06,
+      "loss": 0.6419,
+      "step": 7762
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3901616637770336,
+      "learning_rate": 9.038333704285834e-06,
+      "loss": 0.678,
+      "step": 7763
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3388443454898604,
+      "learning_rate": 9.033489579418162e-06,
+      "loss": 0.6194,
+      "step": 7764
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.36892899418847963,
+      "learning_rate": 9.028646374298475e-06,
+      "loss": 0.6693,
+      "step": 7765
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3728096110021488,
+      "learning_rate": 9.023804089332974e-06,
+      "loss": 0.6471,
+      "step": 7766
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3812456897705734,
+      "learning_rate": 9.018962724927763e-06,
+      "loss": 0.6425,
+      "step": 7767
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3833096379921045,
+      "learning_rate": 9.014122281488883e-06,
+      "loss": 0.6445,
+      "step": 7768
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.37744667067873927,
+      "learning_rate": 9.009282759422308e-06,
+      "loss": 0.6345,
+      "step": 7769
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3471342523780923,
+      "learning_rate": 9.004444159133903e-06,
+      "loss": 0.6436,
+      "step": 7770
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3599478253328132,
+      "learning_rate": 8.99960648102949e-06,
+      "loss": 0.6542,
+      "step": 7771
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.38644704844994043,
+      "learning_rate": 8.994769725514806e-06,
+      "loss": 0.6543,
+      "step": 7772
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.37634776939676584,
+      "learning_rate": 8.989933892995486e-06,
+      "loss": 0.6687,
+      "step": 7773
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3784186673520666,
+      "learning_rate": 8.985098983877114e-06,
+      "loss": 0.6468,
+      "step": 7774
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.44688642267992756,
+      "learning_rate": 8.980264998565196e-06,
+      "loss": 0.7126,
+      "step": 7775
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.37315913563359077,
+      "learning_rate": 8.975431937465136e-06,
+      "loss": 0.6724,
+      "step": 7776
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38673658755329526,
+      "learning_rate": 8.970599800982284e-06,
+      "loss": 0.6505,
+      "step": 7777
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3973164989203668,
+      "learning_rate": 8.965768589521908e-06,
+      "loss": 0.6778,
+      "step": 7778
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.37209325206010324,
+      "learning_rate": 8.960938303489193e-06,
+      "loss": 0.6443,
+      "step": 7779
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3804554250202677,
+      "learning_rate": 8.956108943289259e-06,
+      "loss": 0.6291,
+      "step": 7780
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3560279218838168,
+      "learning_rate": 8.951280509327122e-06,
+      "loss": 0.6473,
+      "step": 7781
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3749859605679323,
+      "learning_rate": 8.946453002007747e-06,
+      "loss": 0.6094,
+      "step": 7782
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.40474697728235337,
+      "learning_rate": 8.941626421736016e-06,
+      "loss": 0.6445,
+      "step": 7783
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 8.34071502670043,
+      "learning_rate": 8.936800768916706e-06,
+      "loss": 6.872,
+      "step": 7784
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.319862703590106,
+      "learning_rate": 8.931976043954564e-06,
+      "loss": 0.6269,
+      "step": 7785
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4178563547591432,
+      "learning_rate": 8.92715224725423e-06,
+      "loss": 0.6509,
+      "step": 7786
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.43446201310456367,
+      "learning_rate": 8.922329379220257e-06,
+      "loss": 0.6583,
+      "step": 7787
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4429743575617748,
+      "learning_rate": 8.917507440257143e-06,
+      "loss": 0.6228,
+      "step": 7788
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.40196487327572267,
+      "learning_rate": 8.9126864307693e-06,
+      "loss": 0.6812,
+      "step": 7789
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.41610643742464104,
+      "learning_rate": 8.90786635116105e-06,
+      "loss": 0.6582,
+      "step": 7790
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.42178595432196386,
+      "learning_rate": 8.903047201836655e-06,
+      "loss": 0.6876,
+      "step": 7791
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4596619254603386,
+      "learning_rate": 8.898228983200286e-06,
+      "loss": 0.6958,
+      "step": 7792
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38766664571841186,
+      "learning_rate": 8.893411695656047e-06,
+      "loss": 0.6705,
+      "step": 7793
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38585565563816493,
+      "learning_rate": 8.888595339607961e-06,
+      "loss": 0.68,
+      "step": 7794
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3364859284141269,
+      "learning_rate": 8.883779915459957e-06,
+      "loss": 0.6344,
+      "step": 7795
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.420880480340064,
+      "learning_rate": 8.87896542361591e-06,
+      "loss": 0.6742,
+      "step": 7796
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.36970280670742584,
+      "learning_rate": 8.874151864479601e-06,
+      "loss": 0.6156,
+      "step": 7797
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 7.00157837340572,
+      "learning_rate": 8.869339238454737e-06,
+      "loss": 6.8474,
+      "step": 7798
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.42036636393344895,
+      "learning_rate": 8.864527545944949e-06,
+      "loss": 0.6238,
+      "step": 7799
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.764589421089816,
+      "learning_rate": 8.859716787353796e-06,
+      "loss": 6.839,
+      "step": 7800
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.35041118891029566,
+      "learning_rate": 8.854906963084733e-06,
+      "loss": 0.6724,
+      "step": 7801
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4037806518169979,
+      "learning_rate": 8.850098073541165e-06,
+      "loss": 0.6428,
+      "step": 7802
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.39017823105344973,
+      "learning_rate": 8.845290119126411e-06,
+      "loss": 0.6533,
+      "step": 7803
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3817439371173368,
+      "learning_rate": 8.840483100243698e-06,
+      "loss": 0.6005,
+      "step": 7804
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3252663755332725,
+      "learning_rate": 8.83567701729618e-06,
+      "loss": 0.6673,
+      "step": 7805
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4306003486774012,
+      "learning_rate": 8.830871870686965e-06,
+      "loss": 0.6885,
+      "step": 7806
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3908294410977007,
+      "learning_rate": 8.826067660819028e-06,
+      "loss": 0.6585,
+      "step": 7807
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.379328364601951,
+      "learning_rate": 8.821264388095304e-06,
+      "loss": 0.6681,
+      "step": 7808
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.36041402465091654,
+      "learning_rate": 8.816462052918641e-06,
+      "loss": 0.6039,
+      "step": 7809
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3402001925934523,
+      "learning_rate": 8.811660655691792e-06,
+      "loss": 0.6586,
+      "step": 7810
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3447982785343603,
+      "learning_rate": 8.806860196817452e-06,
+      "loss": 0.6619,
+      "step": 7811
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.34826749237503524,
+      "learning_rate": 8.80206067669823e-06,
+      "loss": 0.6225,
+      "step": 7812
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38276432453353854,
+      "learning_rate": 8.797262095736653e-06,
+      "loss": 0.6608,
+      "step": 7813
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4059347880669571,
+      "learning_rate": 8.792464454335184e-06,
+      "loss": 0.6657,
+      "step": 7814
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3921271818982482,
+      "learning_rate": 8.787667752896179e-06,
+      "loss": 0.6511,
+      "step": 7815
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3570545371161239,
+      "learning_rate": 8.782871991821938e-06,
+      "loss": 0.6333,
+      "step": 7816
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3486748389128987,
+      "learning_rate": 8.778077171514684e-06,
+      "loss": 0.645,
+      "step": 7817
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.34560043062949364,
+      "learning_rate": 8.773283292376535e-06,
+      "loss": 0.6246,
+      "step": 7818
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.387614749430544,
+      "learning_rate": 8.76849035480955e-06,
+      "loss": 0.604,
+      "step": 7819
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3662590296938834,
+      "learning_rate": 8.763698359215733e-06,
+      "loss": 0.6653,
+      "step": 7820
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 16.426158451546574,
+      "learning_rate": 8.758907305996957e-06,
+      "loss": 7.1022,
+      "step": 7821
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3794438066328254,
+      "learning_rate": 8.754117195555047e-06,
+      "loss": 0.6644,
+      "step": 7822
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.36838123901017605,
+      "learning_rate": 8.749328028291757e-06,
+      "loss": 0.6312,
+      "step": 7823
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3993784772695812,
+      "learning_rate": 8.744539804608727e-06,
+      "loss": 0.6438,
+      "step": 7824
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3831496496366789,
+      "learning_rate": 8.739752524907552e-06,
+      "loss": 0.6491,
+      "step": 7825
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.33522962098600617,
+      "learning_rate": 8.734966189589733e-06,
+      "loss": 0.6151,
+      "step": 7826
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.44390216333726257,
+      "learning_rate": 8.730180799056698e-06,
+      "loss": 0.6631,
+      "step": 7827
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.41004106005974084,
+      "learning_rate": 8.725396353709785e-06,
+      "loss": 0.6825,
+      "step": 7828
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38066091589675843,
+      "learning_rate": 8.720612853950273e-06,
+      "loss": 0.6577,
+      "step": 7829
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3850276929491199,
+      "learning_rate": 8.715830300179331e-06,
+      "loss": 0.6254,
+      "step": 7830
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3685717541907824,
+      "learning_rate": 8.711048692798075e-06,
+      "loss": 0.6849,
+      "step": 7831
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.36850090594662815,
+      "learning_rate": 8.706268032207536e-06,
+      "loss": 0.6268,
+      "step": 7832
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.46251177719718767,
+      "learning_rate": 8.701488318808646e-06,
+      "loss": 0.6615,
+      "step": 7833
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.33756394790319993,
+      "learning_rate": 8.696709553002297e-06,
+      "loss": 0.6225,
+      "step": 7834
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3811744315682081,
+      "learning_rate": 8.691931735189262e-06,
+      "loss": 0.6676,
+      "step": 7835
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3943468987839813,
+      "learning_rate": 8.687154865770256e-06,
+      "loss": 0.6678,
+      "step": 7836
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.35726896241912925,
+      "learning_rate": 8.682378945145915e-06,
+      "loss": 0.6474,
+      "step": 7837
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 19.995272474300695,
+      "learning_rate": 8.677603973716777e-06,
+      "loss": 6.9018,
+      "step": 7838
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.40361488705041904,
+      "learning_rate": 8.672829951883312e-06,
+      "loss": 0.619,
+      "step": 7839
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3640611091911738,
+      "learning_rate": 8.668056880045935e-06,
+      "loss": 0.6434,
+      "step": 7840
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.36232407607283673,
+      "learning_rate": 8.663284758604934e-06,
+      "loss": 0.6259,
+      "step": 7841
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.407795274312361,
+      "learning_rate": 8.658513587960549e-06,
+      "loss": 0.6974,
+      "step": 7842
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.40411269593803084,
+      "learning_rate": 8.65374336851294e-06,
+      "loss": 0.6693,
+      "step": 7843
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3862619242080286,
+      "learning_rate": 8.648974100662165e-06,
+      "loss": 0.633,
+      "step": 7844
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.41567377837391656,
+      "learning_rate": 8.644205784808221e-06,
+      "loss": 0.6625,
+      "step": 7845
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.41980156621424386,
+      "learning_rate": 8.639438421351025e-06,
+      "loss": 0.6691,
+      "step": 7846
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.8299805025336573,
+      "learning_rate": 8.634672010690409e-06,
+      "loss": 6.7108,
+      "step": 7847
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.34860775846061964,
+      "learning_rate": 8.629906553226132e-06,
+      "loss": 0.6138,
+      "step": 7848
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.41037907733430895,
+      "learning_rate": 8.625142049357853e-06,
+      "loss": 0.6108,
+      "step": 7849
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4001502670662147,
+      "learning_rate": 8.620378499485176e-06,
+      "loss": 0.6704,
+      "step": 7850
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38478270124950215,
+      "learning_rate": 8.615615904007607e-06,
+      "loss": 0.646,
+      "step": 7851
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.37460664225944496,
+      "learning_rate": 8.610854263324593e-06,
+      "loss": 0.6664,
+      "step": 7852
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.37042315135206966,
+      "learning_rate": 8.606093577835465e-06,
+      "loss": 0.6631,
+      "step": 7853
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.34704153953916567,
+      "learning_rate": 8.601333847939519e-06,
+      "loss": 0.5964,
+      "step": 7854
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3931781826744117,
+      "learning_rate": 8.596575074035931e-06,
+      "loss": 0.6363,
+      "step": 7855
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3764873376090973,
+      "learning_rate": 8.59181725652382e-06,
+      "loss": 0.6353,
+      "step": 7856
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3507188548754617,
+      "learning_rate": 8.587060395802225e-06,
+      "loss": 0.6169,
+      "step": 7857
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.39101646606563284,
+      "learning_rate": 8.582304492270084e-06,
+      "loss": 0.6343,
+      "step": 7858
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3854410508506883,
+      "learning_rate": 8.577549546326276e-06,
+      "loss": 0.6076,
+      "step": 7859
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38090882181700897,
+      "learning_rate": 8.572795558369594e-06,
+      "loss": 0.6221,
+      "step": 7860
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.36518431606877344,
+      "learning_rate": 8.568042528798745e-06,
+      "loss": 0.6627,
+      "step": 7861
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 6.064059450740184,
+      "learning_rate": 8.563290458012366e-06,
+      "loss": 6.8506,
+      "step": 7862
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4475871116980005,
+      "learning_rate": 8.558539346409009e-06,
+      "loss": 0.712,
+      "step": 7863
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4302662319391368,
+      "learning_rate": 8.553789194387134e-06,
+      "loss": 0.6368,
+      "step": 7864
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3579946606279885,
+      "learning_rate": 8.549040002345134e-06,
+      "loss": 0.6206,
+      "step": 7865
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3467172056055059,
+      "learning_rate": 8.54429177068133e-06,
+      "loss": 0.6321,
+      "step": 7866
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38873031670185165,
+      "learning_rate": 8.539544499793924e-06,
+      "loss": 0.6413,
+      "step": 7867
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3957794529515381,
+      "learning_rate": 8.534798190081093e-06,
+      "loss": 0.6584,
+      "step": 7868
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.30783749329635457,
+      "learning_rate": 8.530052841940885e-06,
+      "loss": 0.597,
+      "step": 7869
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3636301039766874,
+      "learning_rate": 8.525308455771293e-06,
+      "loss": 0.6711,
+      "step": 7870
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.36366081702841224,
+      "learning_rate": 8.520565031970232e-06,
+      "loss": 0.6456,
+      "step": 7871
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.37840325662317376,
+      "learning_rate": 8.515822570935508e-06,
+      "loss": 0.644,
+      "step": 7872
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.36462723976585754,
+      "learning_rate": 8.511081073064878e-06,
+      "loss": 0.6562,
+      "step": 7873
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38077130750254,
+      "learning_rate": 8.506340538756002e-06,
+      "loss": 0.6312,
+      "step": 7874
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3907882850582173,
+      "learning_rate": 8.501600968406465e-06,
+      "loss": 0.6104,
+      "step": 7875
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.38149978179208,
+      "learning_rate": 8.496862362413768e-06,
+      "loss": 0.682,
+      "step": 7876
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4449399162808238,
+      "learning_rate": 8.492124721175339e-06,
+      "loss": 0.6753,
+      "step": 7877
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.42044879102975585,
+      "learning_rate": 8.487388045088502e-06,
+      "loss": 0.6256,
+      "step": 7878
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.37702103403376774,
+      "learning_rate": 8.48265233455053e-06,
+      "loss": 0.643,
+      "step": 7879
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3783849276884644,
+      "learning_rate": 8.477917589958604e-06,
+      "loss": 0.6515,
+      "step": 7880
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.37048020662278486,
+      "learning_rate": 8.473183811709797e-06,
+      "loss": 0.6661,
+      "step": 7881
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.41121790044207457,
+      "learning_rate": 8.468451000201161e-06,
+      "loss": 0.6383,
+      "step": 7882
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3543508586132453,
+      "learning_rate": 8.463719155829604e-06,
+      "loss": 0.6148,
+      "step": 7883
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3745545098345709,
+      "learning_rate": 8.45898827899199e-06,
+      "loss": 0.64,
+      "step": 7884
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3940255901884953,
+      "learning_rate": 8.454258370085091e-06,
+      "loss": 0.6714,
+      "step": 7885
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.35840179702558905,
+      "learning_rate": 8.449529429505605e-06,
+      "loss": 0.6349,
+      "step": 7886
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4129309634388072,
+      "learning_rate": 8.44480145765013e-06,
+      "loss": 0.6814,
+      "step": 7887
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4074134101867618,
+      "learning_rate": 8.440074454915202e-06,
+      "loss": 0.6683,
+      "step": 7888
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38549686789640475,
+      "learning_rate": 8.435348421697267e-06,
+      "loss": 0.6191,
+      "step": 7889
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3383232029494424,
+      "learning_rate": 8.430623358392693e-06,
+      "loss": 0.6278,
+      "step": 7890
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38115864067737465,
+      "learning_rate": 8.425899265397775e-06,
+      "loss": 0.6433,
+      "step": 7891
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3777584543812526,
+      "learning_rate": 8.421176143108698e-06,
+      "loss": 0.6336,
+      "step": 7892
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.36991310975660063,
+      "learning_rate": 8.416453991921597e-06,
+      "loss": 0.6518,
+      "step": 7893
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.37288218729819084,
+      "learning_rate": 8.411732812232516e-06,
+      "loss": 0.6262,
+      "step": 7894
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4387384906976988,
+      "learning_rate": 8.407012604437394e-06,
+      "loss": 0.7063,
+      "step": 7895
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.42053263354642006,
+      "learning_rate": 8.40229336893213e-06,
+      "loss": 0.6401,
+      "step": 7896
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.40165319217512424,
+      "learning_rate": 8.397575106112526e-06,
+      "loss": 0.6497,
+      "step": 7897
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4175139704639349,
+      "learning_rate": 8.392857816374276e-06,
+      "loss": 0.6659,
+      "step": 7898
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.39506487996493933,
+      "learning_rate": 8.388141500113023e-06,
+      "loss": 0.6673,
+      "step": 7899
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.39254061216600866,
+      "learning_rate": 8.383426157724326e-06,
+      "loss": 0.6826,
+      "step": 7900
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.36486956324260494,
+      "learning_rate": 8.378711789603642e-06,
+      "loss": 0.6352,
+      "step": 7901
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.43194386578253335,
+      "learning_rate": 8.373998396146363e-06,
+      "loss": 0.62,
+      "step": 7902
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3223550094446385,
+      "learning_rate": 8.3692859777478e-06,
+      "loss": 0.64,
+      "step": 7903
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.391834348732597,
+      "learning_rate": 8.364574534803174e-06,
+      "loss": 0.6814,
+      "step": 7904
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.31810359006455474,
+      "learning_rate": 8.359864067707639e-06,
+      "loss": 0.6234,
+      "step": 7905
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.345338541676928,
+      "learning_rate": 8.355154576856237e-06,
+      "loss": 0.6067,
+      "step": 7906
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.388290325235762,
+      "learning_rate": 8.35044606264396e-06,
+      "loss": 0.6633,
+      "step": 7907
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4475462709015156,
+      "learning_rate": 8.3457385254657e-06,
+      "loss": 0.6453,
+      "step": 7908
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3504413332953837,
+      "learning_rate": 8.341031965716278e-06,
+      "loss": 0.6472,
+      "step": 7909
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3835864741621025,
+      "learning_rate": 8.336326383790423e-06,
+      "loss": 0.661,
+      "step": 7910
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3945378255484848,
+      "learning_rate": 8.331621780082795e-06,
+      "loss": 0.6133,
+      "step": 7911
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.35611091307635945,
+      "learning_rate": 8.326918154987948e-06,
+      "loss": 0.6111,
+      "step": 7912
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.35170014680491885,
+      "learning_rate": 8.32221550890038e-06,
+      "loss": 0.6121,
+      "step": 7913
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 15.330022585211932,
+      "learning_rate": 8.317513842214502e-06,
+      "loss": 6.7923,
+      "step": 7914
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3761267668944057,
+      "learning_rate": 8.31281315532462e-06,
+      "loss": 0.6506,
+      "step": 7915
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4011588472547432,
+      "learning_rate": 8.308113448624976e-06,
+      "loss": 0.6302,
+      "step": 7916
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.31162634767155994,
+      "learning_rate": 8.303414722509754e-06,
+      "loss": 0.6466,
+      "step": 7917
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3979997894138535,
+      "learning_rate": 8.298716977373004e-06,
+      "loss": 0.6866,
+      "step": 7918
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4222864722628574,
+      "learning_rate": 8.294020213608731e-06,
+      "loss": 0.655,
+      "step": 7919
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3751051273187393,
+      "learning_rate": 8.289324431610852e-06,
+      "loss": 0.6659,
+      "step": 7920
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4114624251902133,
+      "learning_rate": 8.284629631773183e-06,
+      "loss": 0.6298,
+      "step": 7921
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.42727420350302153,
+      "learning_rate": 8.279935814489476e-06,
+      "loss": 0.6895,
+      "step": 7922
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.34877971413859427,
+      "learning_rate": 8.275242980153402e-06,
+      "loss": 0.6367,
+      "step": 7923
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3845386481089303,
+      "learning_rate": 8.270551129158538e-06,
+      "loss": 0.6615,
+      "step": 7924
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.42980325547131637,
+      "learning_rate": 8.26586026189839e-06,
+      "loss": 0.6664,
+      "step": 7925
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.39648466535667093,
+      "learning_rate": 8.261170378766365e-06,
+      "loss": 0.6532,
+      "step": 7926
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.37850020335818013,
+      "learning_rate": 8.256481480155804e-06,
+      "loss": 0.6474,
+      "step": 7927
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4091844171693135,
+      "learning_rate": 8.251793566459967e-06,
+      "loss": 0.6699,
+      "step": 7928
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3524674669716733,
+      "learning_rate": 8.247106638071999e-06,
+      "loss": 0.6363,
+      "step": 7929
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38717888237460757,
+      "learning_rate": 8.242420695385012e-06,
+      "loss": 0.6713,
+      "step": 7930
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4018190305563009,
+      "learning_rate": 8.237735738792008e-06,
+      "loss": 0.6423,
+      "step": 7931
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38777558959569075,
+      "learning_rate": 8.233051768685893e-06,
+      "loss": 0.653,
+      "step": 7932
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.369613156853213,
+      "learning_rate": 8.228368785459517e-06,
+      "loss": 0.6443,
+      "step": 7933
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3601328248817167,
+      "learning_rate": 8.223686789505643e-06,
+      "loss": 0.5875,
+      "step": 7934
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38931781631660134,
+      "learning_rate": 8.219005781216926e-06,
+      "loss": 0.6477,
+      "step": 7935
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4137106223691977,
+      "learning_rate": 8.214325760985969e-06,
+      "loss": 0.6566,
+      "step": 7936
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3845821031319478,
+      "learning_rate": 8.209646729205276e-06,
+      "loss": 0.6441,
+      "step": 7937
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3890165974304402,
+      "learning_rate": 8.204968686267272e-06,
+      "loss": 0.6656,
+      "step": 7938
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38953546512551457,
+      "learning_rate": 8.200291632564307e-06,
+      "loss": 0.6751,
+      "step": 7939
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3676811145990009,
+      "learning_rate": 8.195615568488627e-06,
+      "loss": 0.6392,
+      "step": 7940
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.36765585292892805,
+      "learning_rate": 8.19094049443241e-06,
+      "loss": 0.6468,
+      "step": 7941
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.373214948705158,
+      "learning_rate": 8.186266410787754e-06,
+      "loss": 0.6876,
+      "step": 7942
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.39502217684984187,
+      "learning_rate": 8.181593317946665e-06,
+      "loss": 0.6911,
+      "step": 7943
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4389901448999746,
+      "learning_rate": 8.176921216301075e-06,
+      "loss": 0.6821,
+      "step": 7944
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.328779694850041,
+      "learning_rate": 8.172250106242827e-06,
+      "loss": 0.6473,
+      "step": 7945
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3342424089246226,
+      "learning_rate": 8.167579988163675e-06,
+      "loss": 0.6397,
+      "step": 7946
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3941377543007801,
+      "learning_rate": 8.162910862455296e-06,
+      "loss": 0.6866,
+      "step": 7947
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3707144592829431,
+      "learning_rate": 8.158242729509295e-06,
+      "loss": 0.693,
+      "step": 7948
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.36091635973093533,
+      "learning_rate": 8.153575589717168e-06,
+      "loss": 0.6542,
+      "step": 7949
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3630688380900174,
+      "learning_rate": 8.148909443470343e-06,
+      "loss": 0.6461,
+      "step": 7950
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.37430318157552933,
+      "learning_rate": 8.144244291160186e-06,
+      "loss": 0.6431,
+      "step": 7951
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 12.18848200837596,
+      "learning_rate": 8.139580133177933e-06,
+      "loss": 6.7774,
+      "step": 7952
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.41289751229980043,
+      "learning_rate": 8.134916969914773e-06,
+      "loss": 0.6399,
+      "step": 7953
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 8.465234869258158,
+      "learning_rate": 8.1302548017618e-06,
+      "loss": 6.7669,
+      "step": 7954
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3493732765620324,
+      "learning_rate": 8.12559362911002e-06,
+      "loss": 0.6398,
+      "step": 7955
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4046481652004361,
+      "learning_rate": 8.120933452350358e-06,
+      "loss": 0.6441,
+      "step": 7956
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.30435503331918357,
+      "learning_rate": 8.116274271873664e-06,
+      "loss": 0.6186,
+      "step": 7957
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3429660643126186,
+      "learning_rate": 8.11161608807069e-06,
+      "loss": 0.598,
+      "step": 7958
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3658701790610781,
+      "learning_rate": 8.106958901332129e-06,
+      "loss": 0.6367,
+      "step": 7959
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3414917896351224,
+      "learning_rate": 8.102302712048553e-06,
+      "loss": 0.6503,
+      "step": 7960
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3691417196386966,
+      "learning_rate": 8.097647520610478e-06,
+      "loss": 0.6665,
+      "step": 7961
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3957691960302937,
+      "learning_rate": 8.09299332740834e-06,
+      "loss": 0.6894,
+      "step": 7962
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3613646441027552,
+      "learning_rate": 8.088340132832467e-06,
+      "loss": 0.6419,
+      "step": 7963
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38068112234054025,
+      "learning_rate": 8.08368793727311e-06,
+      "loss": 0.6225,
+      "step": 7964
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.35988646468843416,
+      "learning_rate": 8.079036741120471e-06,
+      "loss": 0.6108,
+      "step": 7965
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.36251741746235167,
+      "learning_rate": 8.074386544764617e-06,
+      "loss": 0.6321,
+      "step": 7966
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.39218820663540255,
+      "learning_rate": 8.069737348595559e-06,
+      "loss": 0.6517,
+      "step": 7967
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.40757724585321575,
+      "learning_rate": 8.065089153003229e-06,
+      "loss": 0.6813,
+      "step": 7968
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.35732618805659405,
+      "learning_rate": 8.060441958377451e-06,
+      "loss": 0.6443,
+      "step": 7969
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4655392918552191,
+      "learning_rate": 8.055795765107988e-06,
+      "loss": 0.6499,
+      "step": 7970
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.33608784292758037,
+      "learning_rate": 8.051150573584508e-06,
+      "loss": 0.6417,
+      "step": 7971
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3097407320697564,
+      "learning_rate": 8.046506384196599e-06,
+      "loss": 0.6301,
+      "step": 7972
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4030420067974269,
+      "learning_rate": 8.041863197333764e-06,
+      "loss": 0.6291,
+      "step": 7973
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3673030712206469,
+      "learning_rate": 8.037221013385425e-06,
+      "loss": 0.6289,
+      "step": 7974
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3912888542295964,
+      "learning_rate": 8.03257983274091e-06,
+      "loss": 0.6846,
+      "step": 7975
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38493735682391095,
+      "learning_rate": 8.027939655789467e-06,
+      "loss": 0.6044,
+      "step": 7976
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3678771785739619,
+      "learning_rate": 8.023300482920278e-06,
+      "loss": 0.646,
+      "step": 7977
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38624474366661493,
+      "learning_rate": 8.018662314522396e-06,
+      "loss": 0.6294,
+      "step": 7978
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.36314921237850584,
+      "learning_rate": 8.01402515098485e-06,
+      "loss": 0.6981,
+      "step": 7979
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4000292052297536,
+      "learning_rate": 8.009388992696534e-06,
+      "loss": 0.6472,
+      "step": 7980
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4730752250641285,
+      "learning_rate": 8.00475384004628e-06,
+      "loss": 0.71,
+      "step": 7981
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4077733996693555,
+      "learning_rate": 8.000119693422843e-06,
+      "loss": 0.6592,
+      "step": 7982
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.35756063773941915,
+      "learning_rate": 7.995486553214868e-06,
+      "loss": 0.6294,
+      "step": 7983
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4258559232917495,
+      "learning_rate": 7.990854419810939e-06,
+      "loss": 0.6525,
+      "step": 7984
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.41461702077225054,
+      "learning_rate": 7.986223293599544e-06,
+      "loss": 0.6609,
+      "step": 7985
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.38800165049613716,
+      "learning_rate": 7.981593174969096e-06,
+      "loss": 0.6423,
+      "step": 7986
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3691573451210225,
+      "learning_rate": 7.97696406430791e-06,
+      "loss": 0.6489,
+      "step": 7987
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.460450618720769,
+      "learning_rate": 7.972335962004237e-06,
+      "loss": 0.674,
+      "step": 7988
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 25.12746671294497,
+      "learning_rate": 7.967708868446213e-06,
+      "loss": 7.9607,
+      "step": 7989
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.34179328817190013,
+      "learning_rate": 7.963082784021916e-06,
+      "loss": 0.6142,
+      "step": 7990
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3554904722516442,
+      "learning_rate": 7.958457709119337e-06,
+      "loss": 0.6433,
+      "step": 7991
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.36221258936809997,
+      "learning_rate": 7.953833644126349e-06,
+      "loss": 0.6228,
+      "step": 7992
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.39262708009244257,
+      "learning_rate": 7.949210589430805e-06,
+      "loss": 0.6777,
+      "step": 7993
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3987344392860419,
+      "learning_rate": 7.944588545420404e-06,
+      "loss": 0.6543,
+      "step": 7994
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7683480973018736,
+      "learning_rate": 7.939967512482804e-06,
+      "loss": 0.6383,
+      "step": 7995
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5418892683496502,
+      "learning_rate": 7.935347491005569e-06,
+      "loss": 0.6498,
+      "step": 7996
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.34951142313345607,
+      "learning_rate": 7.930728481376166e-06,
+      "loss": 0.6122,
+      "step": 7997
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4179664472210601,
+      "learning_rate": 7.926110483981988e-06,
+      "loss": 0.6784,
+      "step": 7998
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3468672232691001,
+      "learning_rate": 7.92149349921034e-06,
+      "loss": 0.637,
+      "step": 7999
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3586680692379352,
+      "learning_rate": 7.916877527448447e-06,
+      "loss": 0.6075,
+      "step": 8000
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.41043617183333697,
+      "learning_rate": 7.912262569083444e-06,
+      "loss": 0.6608,
+      "step": 8001
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.44573512217836564,
+      "learning_rate": 7.907648624502386e-06,
+      "loss": 0.6596,
+      "step": 8002
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.42797232528318885,
+      "learning_rate": 7.903035694092227e-06,
+      "loss": 0.6557,
+      "step": 8003
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6034721527239488,
+      "learning_rate": 7.898423778239857e-06,
+      "loss": 0.6482,
+      "step": 8004
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4175373384056808,
+      "learning_rate": 7.893812877332075e-06,
+      "loss": 0.6767,
+      "step": 8005
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.36066663923991926,
+      "learning_rate": 7.889202991755573e-06,
+      "loss": 0.6544,
+      "step": 8006
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.39623156818418825,
+      "learning_rate": 7.884594121896995e-06,
+      "loss": 0.6258,
+      "step": 8007
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.37259355613353273,
+      "learning_rate": 7.879986268142888e-06,
+      "loss": 0.6099,
+      "step": 8008
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.862721490389433,
+      "learning_rate": 7.875379430879684e-06,
+      "loss": 6.7493,
+      "step": 8009
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3917465896534183,
+      "learning_rate": 7.870773610493764e-06,
+      "loss": 0.6601,
+      "step": 8010
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3976747957255453,
+      "learning_rate": 7.866168807371424e-06,
+      "loss": 0.6734,
+      "step": 8011
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.36734503581406497,
+      "learning_rate": 7.861565021898836e-06,
+      "loss": 0.6667,
+      "step": 8012
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.35299745532112453,
+      "learning_rate": 7.856962254462144e-06,
+      "loss": 0.6372,
+      "step": 8013
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.36124887152341306,
+      "learning_rate": 7.852360505447353e-06,
+      "loss": 0.64,
+      "step": 8014
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38896730497946225,
+      "learning_rate": 7.84775977524042e-06,
+      "loss": 0.6772,
+      "step": 8015
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3795283722673078,
+      "learning_rate": 7.843160064227204e-06,
+      "loss": 0.6109,
+      "step": 8016
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3550870755027191,
+      "learning_rate": 7.838561372793466e-06,
+      "loss": 0.6014,
+      "step": 8017
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3446821108212194,
+      "learning_rate": 7.833963701324897e-06,
+      "loss": 0.5901,
+      "step": 8018
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3712535395563367,
+      "learning_rate": 7.829367050207097e-06,
+      "loss": 0.6401,
+      "step": 8019
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3736830149367879,
+      "learning_rate": 7.824771419825588e-06,
+      "loss": 0.6441,
+      "step": 8020
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3821439337009565,
+      "learning_rate": 7.820176810565793e-06,
+      "loss": 0.6589,
+      "step": 8021
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.49285116385638195,
+      "learning_rate": 7.815583222813067e-06,
+      "loss": 0.6614,
+      "step": 8022
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.409000188354108,
+      "learning_rate": 7.810990656952657e-06,
+      "loss": 0.5945,
+      "step": 8023
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.399150313774141,
+      "learning_rate": 7.806399113369738e-06,
+      "loss": 0.6402,
+      "step": 8024
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.41676486209132313,
+      "learning_rate": 7.801808592449405e-06,
+      "loss": 0.6149,
+      "step": 8025
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3530186160993551,
+      "learning_rate": 7.79721909457664e-06,
+      "loss": 0.6626,
+      "step": 8026
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3752637545590638,
+      "learning_rate": 7.792630620136388e-06,
+      "loss": 0.661,
+      "step": 8027
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.31793739867524573,
+      "learning_rate": 7.788043169513454e-06,
+      "loss": 0.5696,
+      "step": 8028
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38729817535806976,
+      "learning_rate": 7.783456743092595e-06,
+      "loss": 0.6165,
+      "step": 8029
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4281203399834847,
+      "learning_rate": 7.77887134125846e-06,
+      "loss": 0.5921,
+      "step": 8030
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3925809015716216,
+      "learning_rate": 7.774286964395636e-06,
+      "loss": 0.6378,
+      "step": 8031
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3352882193914323,
+      "learning_rate": 7.769703612888593e-06,
+      "loss": 0.6062,
+      "step": 8032
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3854142961857897,
+      "learning_rate": 7.765121287121737e-06,
+      "loss": 0.6503,
+      "step": 8033
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38100236444944224,
+      "learning_rate": 7.760539987479383e-06,
+      "loss": 0.6722,
+      "step": 8034
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3478239278293308,
+      "learning_rate": 7.755959714345757e-06,
+      "loss": 0.6331,
+      "step": 8035
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.0753018707824469,
+      "learning_rate": 7.75138046810501e-06,
+      "loss": 6.7641,
+      "step": 8036
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3945473743672958,
+      "learning_rate": 7.746802249141186e-06,
+      "loss": 0.6504,
+      "step": 8037
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4002214172436174,
+      "learning_rate": 7.742225057838258e-06,
+      "loss": 0.668,
+      "step": 8038
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3594114820806455,
+      "learning_rate": 7.737648894580117e-06,
+      "loss": 0.6124,
+      "step": 8039
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6309201620824065,
+      "learning_rate": 7.733073759750537e-06,
+      "loss": 6.9009,
+      "step": 8040
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.3123987000188984,
+      "learning_rate": 7.728499653733259e-06,
+      "loss": 6.766,
+      "step": 8041
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3697515085758755,
+      "learning_rate": 7.723926576911898e-06,
+      "loss": 0.645,
+      "step": 8042
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3953785439745693,
+      "learning_rate": 7.719354529669985e-06,
+      "loss": 0.6373,
+      "step": 8043
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.39521457433343615,
+      "learning_rate": 7.714783512390973e-06,
+      "loss": 0.6903,
+      "step": 8044
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3633201672034797,
+      "learning_rate": 7.71021352545824e-06,
+      "loss": 0.6456,
+      "step": 8045
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3952678669680449,
+      "learning_rate": 7.705644569255048e-06,
+      "loss": 0.6306,
+      "step": 8046
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.42773409770685766,
+      "learning_rate": 7.7010766441646e-06,
+      "loss": 0.6304,
+      "step": 8047
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3806670344040606,
+      "learning_rate": 7.69650975057e-06,
+      "loss": 0.6447,
+      "step": 8048
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4061984043081321,
+      "learning_rate": 7.69194388885427e-06,
+      "loss": 0.6619,
+      "step": 8049
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.44282171724661284,
+      "learning_rate": 7.687379059400347e-06,
+      "loss": 0.6085,
+      "step": 8050
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38227971739244665,
+      "learning_rate": 7.682815262591064e-06,
+      "loss": 0.6551,
+      "step": 8051
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.36140098001558385,
+      "learning_rate": 7.678252498809191e-06,
+      "loss": 0.6421,
+      "step": 8052
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.4254525148474657,
+      "learning_rate": 7.673690768437406e-06,
+      "loss": 6.6577,
+      "step": 8053
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.40699434858417927,
+      "learning_rate": 7.669130071858277e-06,
+      "loss": 0.6595,
+      "step": 8054
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4003918742609683,
+      "learning_rate": 7.664570409454323e-06,
+      "loss": 0.674,
+      "step": 8055
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4174684686021708,
+      "learning_rate": 7.66001178160796e-06,
+      "loss": 0.6449,
+      "step": 8056
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.36443912505959175,
+      "learning_rate": 7.655454188701496e-06,
+      "loss": 0.6546,
+      "step": 8057
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.44437807097522297,
+      "learning_rate": 7.650897631117185e-06,
+      "loss": 0.62,
+      "step": 8058
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3841548534430395,
+      "learning_rate": 7.646342109237179e-06,
+      "loss": 0.6764,
+      "step": 8059
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.42848693733769255,
+      "learning_rate": 7.641787623443535e-06,
+      "loss": 0.6521,
+      "step": 8060
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3835045859840867,
+      "learning_rate": 7.637234174118237e-06,
+      "loss": 0.6514,
+      "step": 8061
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3606605982933518,
+      "learning_rate": 7.632681761643182e-06,
+      "loss": 0.5709,
+      "step": 8062
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.45445474890836024,
+      "learning_rate": 7.628130386400168e-06,
+      "loss": 0.6808,
+      "step": 8063
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.39423449238210606,
+      "learning_rate": 7.62358004877092e-06,
+      "loss": 0.6426,
+      "step": 8064
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.37643279366926385,
+      "learning_rate": 7.619030749137071e-06,
+      "loss": 0.6687,
+      "step": 8065
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3756053228820128,
+      "learning_rate": 7.614482487880155e-06,
+      "loss": 0.6144,
+      "step": 8066
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3564948746084957,
+      "learning_rate": 7.609935265381634e-06,
+      "loss": 0.6693,
+      "step": 8067
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.42789774633910277,
+      "learning_rate": 7.605389082022878e-06,
+      "loss": 0.6846,
+      "step": 8068
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3885484525748804,
+      "learning_rate": 7.600843938185172e-06,
+      "loss": 0.6192,
+      "step": 8069
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3642533699600225,
+      "learning_rate": 7.5962998342497185e-06,
+      "loss": 0.645,
+      "step": 8070
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4293659063634673,
+      "learning_rate": 7.591756770597609e-06,
+      "loss": 0.6659,
+      "step": 8071
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.35255615054319756,
+      "learning_rate": 7.587214747609875e-06,
+      "loss": 0.6323,
+      "step": 8072
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.310009482405574,
+      "learning_rate": 7.582673765667457e-06,
+      "loss": 0.6176,
+      "step": 8073
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3633617687286801,
+      "learning_rate": 7.578133825151186e-06,
+      "loss": 0.6325,
+      "step": 8074
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.32820921058970925,
+      "learning_rate": 7.573594926441821e-06,
+      "loss": 0.6346,
+      "step": 8075
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3970474241393539,
+      "learning_rate": 7.569057069920058e-06,
+      "loss": 0.6205,
+      "step": 8076
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4123586285199135,
+      "learning_rate": 7.564520255966459e-06,
+      "loss": 0.6504,
+      "step": 8077
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4055342439072991,
+      "learning_rate": 7.559984484961527e-06,
+      "loss": 0.6605,
+      "step": 8078
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3560431502420545,
+      "learning_rate": 7.555449757285678e-06,
+      "loss": 0.6343,
+      "step": 8079
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3691615997341273,
+      "learning_rate": 7.550916073319223e-06,
+      "loss": 0.6433,
+      "step": 8080
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.37030447630786223,
+      "learning_rate": 7.546383433442403e-06,
+      "loss": 0.6864,
+      "step": 8081
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.42206452744728157,
+      "learning_rate": 7.541851838035363e-06,
+      "loss": 0.7089,
+      "step": 8082
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4097219766692276,
+      "learning_rate": 7.537321287478165e-06,
+      "loss": 0.626,
+      "step": 8083
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.30321948519110975,
+      "learning_rate": 7.532791782150788e-06,
+      "loss": 0.6218,
+      "step": 8084
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4082855979846953,
+      "learning_rate": 7.528263322433098e-06,
+      "loss": 0.6559,
+      "step": 8085
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.181339293720998,
+      "learning_rate": 7.523735908704901e-06,
+      "loss": 6.6392,
+      "step": 8086
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.39160277125279114,
+      "learning_rate": 7.519209541345907e-06,
+      "loss": 0.613,
+      "step": 8087
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4268429819963815,
+      "learning_rate": 7.5146842207357415e-06,
+      "loss": 0.6469,
+      "step": 8088
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3653853285360772,
+      "learning_rate": 7.510159947253919e-06,
+      "loss": 0.6257,
+      "step": 8089
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.34627374382121295,
+      "learning_rate": 7.505636721279912e-06,
+      "loss": 0.6142,
+      "step": 8090
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4291721674468427,
+      "learning_rate": 7.5011145431930574e-06,
+      "loss": 0.6802,
+      "step": 8091
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3569458075636719,
+      "learning_rate": 7.49659341337263e-06,
+      "loss": 0.6422,
+      "step": 8092
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4045647788444454,
+      "learning_rate": 7.49207333219782e-06,
+      "loss": 0.66,
+      "step": 8093
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4362083891880832,
+      "learning_rate": 7.487554300047706e-06,
+      "loss": 0.6827,
+      "step": 8094
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.34049713308081336,
+      "learning_rate": 7.483036317301295e-06,
+      "loss": 0.6391,
+      "step": 8095
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4016487055642118,
+      "learning_rate": 7.478519384337528e-06,
+      "loss": 0.6324,
+      "step": 8096
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.34137324545535874,
+      "learning_rate": 7.474003501535208e-06,
+      "loss": 0.6329,
+      "step": 8097
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3872450595038709,
+      "learning_rate": 7.469488669273088e-06,
+      "loss": 0.6757,
+      "step": 8098
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3903010266687534,
+      "learning_rate": 7.464974887929828e-06,
+      "loss": 0.6408,
+      "step": 8099
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.419038830456713,
+      "learning_rate": 7.460462157883976e-06,
+      "loss": 0.683,
+      "step": 8100
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4000959930300951,
+      "learning_rate": 7.455950479514023e-06,
+      "loss": 0.6533,
+      "step": 8101
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.31670421532227383,
+      "learning_rate": 7.45143985319835e-06,
+      "loss": 0.6617,
+      "step": 8102
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.39331725716946236,
+      "learning_rate": 7.446930279315263e-06,
+      "loss": 0.6404,
+      "step": 8103
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3893090295319451,
+      "learning_rate": 7.442421758242981e-06,
+      "loss": 0.6514,
+      "step": 8104
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.350998636968553,
+      "learning_rate": 7.437914290359613e-06,
+      "loss": 0.6047,
+      "step": 8105
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4421515071012707,
+      "learning_rate": 7.433407876043201e-06,
+      "loss": 0.6963,
+      "step": 8106
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.37417612520756743,
+      "learning_rate": 7.428902515671703e-06,
+      "loss": 0.6561,
+      "step": 8107
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38418587842082197,
+      "learning_rate": 7.424398209622961e-06,
+      "loss": 0.675,
+      "step": 8108
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4118465971855138,
+      "learning_rate": 7.4198949582747474e-06,
+      "loss": 0.6976,
+      "step": 8109
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.43738759416971873,
+      "learning_rate": 7.415392762004763e-06,
+      "loss": 0.6731,
+      "step": 8110
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.43128117079688694,
+      "learning_rate": 7.410891621190584e-06,
+      "loss": 0.6675,
+      "step": 8111
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3442994404528607,
+      "learning_rate": 7.406391536209721e-06,
+      "loss": 0.6656,
+      "step": 8112
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4282502271146253,
+      "learning_rate": 7.401892507439597e-06,
+      "loss": 0.6744,
+      "step": 8113
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3389881787462292,
+      "learning_rate": 7.397394535257525e-06,
+      "loss": 0.6129,
+      "step": 8114
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3559581305032728,
+      "learning_rate": 7.392897620040755e-06,
+      "loss": 0.6798,
+      "step": 8115
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3847473753044412,
+      "learning_rate": 7.388401762166437e-06,
+      "loss": 0.6176,
+      "step": 8116
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3647430704902622,
+      "learning_rate": 7.383906962011631e-06,
+      "loss": 0.6135,
+      "step": 8117
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3384239609427257,
+      "learning_rate": 7.37941321995332e-06,
+      "loss": 0.6349,
+      "step": 8118
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3340865067419013,
+      "learning_rate": 7.374920536368371e-06,
+      "loss": 0.6317,
+      "step": 8119
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3802688541406304,
+      "learning_rate": 7.370428911633592e-06,
+      "loss": 0.6426,
+      "step": 8120
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.35871400052109875,
+      "learning_rate": 7.365938346125685e-06,
+      "loss": 0.6271,
+      "step": 8121
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3520982399335283,
+      "learning_rate": 7.361448840221279e-06,
+      "loss": 0.6345,
+      "step": 8122
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3990223137951404,
+      "learning_rate": 7.356960394296882e-06,
+      "loss": 0.6233,
+      "step": 8123
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3529819774218246,
+      "learning_rate": 7.352473008728962e-06,
+      "loss": 0.6567,
+      "step": 8124
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.35084624408587817,
+      "learning_rate": 7.347986683893848e-06,
+      "loss": 0.6201,
+      "step": 8125
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.379708508663485,
+      "learning_rate": 7.343501420167811e-06,
+      "loss": 0.6516,
+      "step": 8126
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3970049499478319,
+      "learning_rate": 7.3390172179270314e-06,
+      "loss": 0.6649,
+      "step": 8127
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.33729126259476516,
+      "learning_rate": 7.334534077547582e-06,
+      "loss": 0.6265,
+      "step": 8128
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.38995500510308334,
+      "learning_rate": 7.330051999405463e-06,
+      "loss": 0.6796,
+      "step": 8129
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4499052537739729,
+      "learning_rate": 7.32557098387658e-06,
+      "loss": 0.6778,
+      "step": 8130
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3752406280938449,
+      "learning_rate": 7.3210910313367535e-06,
+      "loss": 0.6846,
+      "step": 8131
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.40153511972338984,
+      "learning_rate": 7.31661214216171e-06,
+      "loss": 0.6565,
+      "step": 8132
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.385400599232496,
+      "learning_rate": 7.312134316727093e-06,
+      "loss": 0.6676,
+      "step": 8133
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.38559564524600737,
+      "learning_rate": 7.3076575554084425e-06,
+      "loss": 0.6486,
+      "step": 8134
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.35717623353876826,
+      "learning_rate": 7.303181858581225e-06,
+      "loss": 0.6332,
+      "step": 8135
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3565126646983936,
+      "learning_rate": 7.298707226620818e-06,
+      "loss": 0.633,
+      "step": 8136
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3944037122445903,
+      "learning_rate": 7.294233659902481e-06,
+      "loss": 0.6316,
+      "step": 8137
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3899795009286474,
+      "learning_rate": 7.289761158801438e-06,
+      "loss": 0.675,
+      "step": 8138
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.415224937183314,
+      "learning_rate": 7.2852897236927675e-06,
+      "loss": 0.6789,
+      "step": 8139
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3712407722371979,
+      "learning_rate": 7.280819354951496e-06,
+      "loss": 0.6431,
+      "step": 8140
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3792267953131774,
+      "learning_rate": 7.276350052952548e-06,
+      "loss": 0.6274,
+      "step": 8141
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.362101020914675,
+      "learning_rate": 7.27188181807075e-06,
+      "loss": 0.6392,
+      "step": 8142
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.36720225701946135,
+      "learning_rate": 7.267414650680851e-06,
+      "loss": 0.6737,
+      "step": 8143
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3893782359536392,
+      "learning_rate": 7.2629485511575075e-06,
+      "loss": 0.6286,
+      "step": 8144
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3606995903736646,
+      "learning_rate": 7.2584835198752875e-06,
+      "loss": 0.6044,
+      "step": 8145
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.343421763495758,
+      "learning_rate": 7.254019557208667e-06,
+      "loss": 0.6443,
+      "step": 8146
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3784153816139314,
+      "learning_rate": 7.249556663532038e-06,
+      "loss": 0.6396,
+      "step": 8147
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.36472263976532304,
+      "learning_rate": 7.245094839219688e-06,
+      "loss": 0.6655,
+      "step": 8148
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9323941204216846,
+      "learning_rate": 7.24063408464583e-06,
+      "loss": 6.5966,
+      "step": 8149
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3953891957787082,
+      "learning_rate": 7.236174400184588e-06,
+      "loss": 0.6532,
+      "step": 8150
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3590163600944982,
+      "learning_rate": 7.231715786209972e-06,
+      "loss": 0.6502,
+      "step": 8151
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.42259503765887757,
+      "learning_rate": 7.227258243095938e-06,
+      "loss": 0.6532,
+      "step": 8152
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3831752397619625,
+      "learning_rate": 7.222801771216337e-06,
+      "loss": 0.6405,
+      "step": 8153
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4119917302015242,
+      "learning_rate": 7.218346370944913e-06,
+      "loss": 0.6537,
+      "step": 8154
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.37899403001210935,
+      "learning_rate": 7.213892042655344e-06,
+      "loss": 0.6054,
+      "step": 8155
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.38639299884298905,
+      "learning_rate": 7.209438786721212e-06,
+      "loss": 0.6392,
+      "step": 8156
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.47118869005958647,
+      "learning_rate": 7.2049866035159975e-06,
+      "loss": 0.6493,
+      "step": 8157
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.39610945056371394,
+      "learning_rate": 7.200535493413104e-06,
+      "loss": 0.6719,
+      "step": 8158
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3530667420978717,
+      "learning_rate": 7.196085456785839e-06,
+      "loss": 0.6181,
+      "step": 8159
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.38790250490602196,
+      "learning_rate": 7.1916364940074254e-06,
+      "loss": 0.6324,
+      "step": 8160
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.41174633670930355,
+      "learning_rate": 7.187188605450996e-06,
+      "loss": 0.6502,
+      "step": 8161
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.34559514735191477,
+      "learning_rate": 7.182741791489578e-06,
+      "loss": 0.6304,
+      "step": 8162
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.35978264179861463,
+      "learning_rate": 7.1782960524961276e-06,
+      "loss": 0.6294,
+      "step": 8163
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.33721974388837755,
+      "learning_rate": 7.173851388843507e-06,
+      "loss": 0.5894,
+      "step": 8164
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.34219954870369323,
+      "learning_rate": 7.169407800904468e-06,
+      "loss": 0.5863,
+      "step": 8165
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3213732640066766,
+      "learning_rate": 7.164965289051709e-06,
+      "loss": 0.6149,
+      "step": 8166
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.8217559662439085,
+      "learning_rate": 7.160523853657817e-06,
+      "loss": 6.6991,
+      "step": 8167
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.406716573224584,
+      "learning_rate": 7.156083495095276e-06,
+      "loss": 0.635,
+      "step": 8168
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3806598210810136,
+      "learning_rate": 7.151644213736499e-06,
+      "loss": 0.6846,
+      "step": 8169
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4042630230656977,
+      "learning_rate": 7.147206009953815e-06,
+      "loss": 0.673,
+      "step": 8170
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.35498703085368827,
+      "learning_rate": 7.142768884119431e-06,
+      "loss": 0.6591,
+      "step": 8171
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3521261606897371,
+      "learning_rate": 7.138332836605495e-06,
+      "loss": 0.6005,
+      "step": 8172
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.33720904414586617,
+      "learning_rate": 7.13389786778405e-06,
+      "loss": 0.6554,
+      "step": 8173
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.41284997507647714,
+      "learning_rate": 7.129463978027054e-06,
+      "loss": 0.6267,
+      "step": 8174
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.36592517852761797,
+      "learning_rate": 7.125031167706375e-06,
+      "loss": 0.6291,
+      "step": 8175
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.32331395085400744,
+      "learning_rate": 7.120599437193778e-06,
+      "loss": 0.6312,
+      "step": 8176
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.41751684706013914,
+      "learning_rate": 7.116168786860951e-06,
+      "loss": 0.6771,
+      "step": 8177
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.377860503972265,
+      "learning_rate": 7.1117392170794876e-06,
+      "loss": 0.6689,
+      "step": 8178
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.33442465694264173,
+      "learning_rate": 7.107310728220893e-06,
+      "loss": 0.6013,
+      "step": 8179
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4367530852089331,
+      "learning_rate": 7.102883320656575e-06,
+      "loss": 0.6521,
+      "step": 8180
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.39034363425659446,
+      "learning_rate": 7.098456994757865e-06,
+      "loss": 0.6446,
+      "step": 8181
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.37121506078406025,
+      "learning_rate": 7.094031750895977e-06,
+      "loss": 0.6588,
+      "step": 8182
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3349995331435016,
+      "learning_rate": 7.0896075894420624e-06,
+      "loss": 0.6421,
+      "step": 8183
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4221432106079605,
+      "learning_rate": 7.085184510767173e-06,
+      "loss": 0.6687,
+      "step": 8184
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3622075236736698,
+      "learning_rate": 7.08076251524225e-06,
+      "loss": 0.6296,
+      "step": 8185
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3595934547316445,
+      "learning_rate": 7.076341603238177e-06,
+      "loss": 0.6259,
+      "step": 8186
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.35288941607572777,
+      "learning_rate": 7.071921775125732e-06,
+      "loss": 0.593,
+      "step": 8187
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3561196378550871,
+      "learning_rate": 7.067503031275589e-06,
+      "loss": 0.6164,
+      "step": 8188
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3752081185160079,
+      "learning_rate": 7.063085372058347e-06,
+      "loss": 0.6191,
+      "step": 8189
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3961816491293714,
+      "learning_rate": 7.058668797844519e-06,
+      "loss": 0.6317,
+      "step": 8190
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.36886388887191457,
+      "learning_rate": 7.054253309004504e-06,
+      "loss": 0.6261,
+      "step": 8191
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4017411142752068,
+      "learning_rate": 7.049838905908628e-06,
+      "loss": 0.6311,
+      "step": 8192
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4164014866292532,
+      "learning_rate": 7.045425588927122e-06,
+      "loss": 0.7108,
+      "step": 8193
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.42747570646264554,
+      "learning_rate": 7.041013358430129e-06,
+      "loss": 0.6625,
+      "step": 8194
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3848011501226488,
+      "learning_rate": 7.0366022147877e-06,
+      "loss": 0.6304,
+      "step": 8195
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3447214938496069,
+      "learning_rate": 7.0321921583697815e-06,
+      "loss": 0.6329,
+      "step": 8196
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.37272087070602816,
+      "learning_rate": 7.027783189546247e-06,
+      "loss": 0.6225,
+      "step": 8197
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.33563515122733995,
+      "learning_rate": 7.023375308686875e-06,
+      "loss": 0.6064,
+      "step": 8198
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.382076142701129,
+      "learning_rate": 7.018968516161331e-06,
+      "loss": 0.6043,
+      "step": 8199
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3958566570895365,
+      "learning_rate": 7.014562812339227e-06,
+      "loss": 0.6826,
+      "step": 8200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.35448130596330235,
+      "learning_rate": 7.0101581975900644e-06,
+      "loss": 0.6171,
+      "step": 8201
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3765343781031801,
+      "learning_rate": 7.005754672283238e-06,
+      "loss": 0.6775,
+      "step": 8202
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3995708973051078,
+      "learning_rate": 7.001352236788077e-06,
+      "loss": 0.6281,
+      "step": 8203
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.40179826106316857,
+      "learning_rate": 6.996950891473811e-06,
+      "loss": 0.6731,
+      "step": 8204
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3992671621659078,
+      "learning_rate": 6.992550636709561e-06,
+      "loss": 0.6322,
+      "step": 8205
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.40596923846268723,
+      "learning_rate": 6.9881514728643815e-06,
+      "loss": 0.6717,
+      "step": 8206
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.40986984643949714,
+      "learning_rate": 6.983753400307225e-06,
+      "loss": 0.6441,
+      "step": 8207
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.37632997303324506,
+      "learning_rate": 6.979356419406951e-06,
+      "loss": 0.6083,
+      "step": 8208
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3771801771688144,
+      "learning_rate": 6.974960530532329e-06,
+      "loss": 0.6676,
+      "step": 8209
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.38342266775967576,
+      "learning_rate": 6.970565734052042e-06,
+      "loss": 0.6386,
+      "step": 8210
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.36133742106090533,
+      "learning_rate": 6.966172030334666e-06,
+      "loss": 0.6618,
+      "step": 8211
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3837503516655582,
+      "learning_rate": 6.9617794197487e-06,
+      "loss": 0.6521,
+      "step": 8212
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.33839537985847035,
+      "learning_rate": 6.957387902662549e-06,
+      "loss": 0.6628,
+      "step": 8213
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3253806193810102,
+      "learning_rate": 6.952997479444523e-06,
+      "loss": 0.6026,
+      "step": 8214
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.44488870469496383,
+      "learning_rate": 6.9486081504628476e-06,
+      "loss": 0.6718,
+      "step": 8215
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.44352847035990633,
+      "learning_rate": 6.944219916085637e-06,
+      "loss": 0.7238,
+      "step": 8216
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3790381374790851,
+      "learning_rate": 6.939832776680937e-06,
+      "loss": 0.5836,
+      "step": 8217
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4039376128394566,
+      "learning_rate": 6.935446732616695e-06,
+      "loss": 0.7008,
+      "step": 8218
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.38791241266722376,
+      "learning_rate": 6.9310617842607505e-06,
+      "loss": 0.6848,
+      "step": 8219
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.36017719923463776,
+      "learning_rate": 6.926677931980866e-06,
+      "loss": 0.6386,
+      "step": 8220
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4061324398956461,
+      "learning_rate": 6.922295176144724e-06,
+      "loss": 0.6475,
+      "step": 8221
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3746182705691556,
+      "learning_rate": 6.91791351711989e-06,
+      "loss": 0.6337,
+      "step": 8222
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39676263386338667,
+      "learning_rate": 6.913532955273848e-06,
+      "loss": 0.6515,
+      "step": 8223
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.40237899636671765,
+      "learning_rate": 6.9091534909739984e-06,
+      "loss": 0.6382,
+      "step": 8224
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4112595043002176,
+      "learning_rate": 6.904775124587629e-06,
+      "loss": 0.663,
+      "step": 8225
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3742323222732074,
+      "learning_rate": 6.900397856481955e-06,
+      "loss": 0.6405,
+      "step": 8226
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39526719144122163,
+      "learning_rate": 6.896021687024092e-06,
+      "loss": 0.6478,
+      "step": 8227
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.37222494173790494,
+      "learning_rate": 6.891646616581067e-06,
+      "loss": 0.651,
+      "step": 8228
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.36648167489950234,
+      "learning_rate": 6.8872726455198155e-06,
+      "loss": 0.6246,
+      "step": 8229
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.36108611415635633,
+      "learning_rate": 6.882899774207163e-06,
+      "loss": 0.5776,
+      "step": 8230
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3789630517365604,
+      "learning_rate": 6.878528003009866e-06,
+      "loss": 0.6885,
+      "step": 8231
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4245204366118476,
+      "learning_rate": 6.8741573322945845e-06,
+      "loss": 0.6226,
+      "step": 8232
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3773608479842764,
+      "learning_rate": 6.869787762427869e-06,
+      "loss": 0.5842,
+      "step": 8233
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3707105950760314,
+      "learning_rate": 6.86541929377619e-06,
+      "loss": 0.6575,
+      "step": 8234
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3436877964119401,
+      "learning_rate": 6.861051926705946e-06,
+      "loss": 0.6638,
+      "step": 8235
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3883653353968144,
+      "learning_rate": 6.856685661583402e-06,
+      "loss": 0.6307,
+      "step": 8236
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3476904038824729,
+      "learning_rate": 6.852320498774758e-06,
+      "loss": 0.6119,
+      "step": 8237
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3766760149273687,
+      "learning_rate": 6.847956438646124e-06,
+      "loss": 0.6809,
+      "step": 8238
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3505260613306351,
+      "learning_rate": 6.843593481563491e-06,
+      "loss": 0.6372,
+      "step": 8239
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.35967648503135635,
+      "learning_rate": 6.839231627892784e-06,
+      "loss": 0.6242,
+      "step": 8240
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4293721866762456,
+      "learning_rate": 6.834870877999829e-06,
+      "loss": 0.6696,
+      "step": 8241
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4615841228532243,
+      "learning_rate": 6.83051123225035e-06,
+      "loss": 0.6132,
+      "step": 8242
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4058013938533374,
+      "learning_rate": 6.82615269100999e-06,
+      "loss": 0.6071,
+      "step": 8243
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.35374593729812837,
+      "learning_rate": 6.821795254644304e-06,
+      "loss": 0.6561,
+      "step": 8244
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.37906504931840934,
+      "learning_rate": 6.817438923518725e-06,
+      "loss": 0.6512,
+      "step": 8245
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3845642107074866,
+      "learning_rate": 6.8130836979986236e-06,
+      "loss": 0.6367,
+      "step": 8246
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.36988705285853574,
+      "learning_rate": 6.808729578449273e-06,
+      "loss": 0.6033,
+      "step": 8247
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39708843512253816,
+      "learning_rate": 6.804376565235828e-06,
+      "loss": 0.6917,
+      "step": 8248
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39680001562675316,
+      "learning_rate": 6.800024658723399e-06,
+      "loss": 0.6545,
+      "step": 8249
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.364036414073239,
+      "learning_rate": 6.795673859276952e-06,
+      "loss": 0.6667,
+      "step": 8250
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3622009530623417,
+      "learning_rate": 6.7913241672613925e-06,
+      "loss": 0.5964,
+      "step": 8251
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.35624054699380264,
+      "learning_rate": 6.7869755830415305e-06,
+      "loss": 0.6194,
+      "step": 8252
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39420534178239847,
+      "learning_rate": 6.782628106982061e-06,
+      "loss": 0.6089,
+      "step": 8253
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41983721549641223,
+      "learning_rate": 6.778281739447612e-06,
+      "loss": 0.6207,
+      "step": 8254
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.43189307946597316,
+      "learning_rate": 6.773936480802703e-06,
+      "loss": 0.6353,
+      "step": 8255
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.36816958178599857,
+      "learning_rate": 6.76959233141177e-06,
+      "loss": 0.6411,
+      "step": 8256
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4002131062174359,
+      "learning_rate": 6.765249291639148e-06,
+      "loss": 0.6638,
+      "step": 8257
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41157868160776273,
+      "learning_rate": 6.760907361849092e-06,
+      "loss": 0.6232,
+      "step": 8258
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.36462467189718906,
+      "learning_rate": 6.7565665424057404e-06,
+      "loss": 0.6104,
+      "step": 8259
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39032352880088456,
+      "learning_rate": 6.752226833673159e-06,
+      "loss": 0.6447,
+      "step": 8260
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.38393742019637794,
+      "learning_rate": 6.747888236015314e-06,
+      "loss": 0.6597,
+      "step": 8261
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3827173582682589,
+      "learning_rate": 6.743550749796077e-06,
+      "loss": 0.6689,
+      "step": 8262
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3897866254316098,
+      "learning_rate": 6.739214375379237e-06,
+      "loss": 0.6468,
+      "step": 8263
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3711010474703614,
+      "learning_rate": 6.734879113128465e-06,
+      "loss": 0.6503,
+      "step": 8264
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.40704888782478876,
+      "learning_rate": 6.730544963407362e-06,
+      "loss": 0.6181,
+      "step": 8265
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.38132393466661746,
+      "learning_rate": 6.726211926579427e-06,
+      "loss": 0.6723,
+      "step": 8266
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3566383993039098,
+      "learning_rate": 6.721880003008074e-06,
+      "loss": 0.6315,
+      "step": 8267
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3529818615158553,
+      "learning_rate": 6.717549193056596e-06,
+      "loss": 0.6319,
+      "step": 8268
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.48102945415072,
+      "learning_rate": 6.713219497088239e-06,
+      "loss": 6.6792,
+      "step": 8269
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39490788280180233,
+      "learning_rate": 6.708890915466108e-06,
+      "loss": 0.6453,
+      "step": 8270
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41823305818354367,
+      "learning_rate": 6.704563448553248e-06,
+      "loss": 0.6826,
+      "step": 8271
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4479733707333689,
+      "learning_rate": 6.700237096712598e-06,
+      "loss": 0.6765,
+      "step": 8272
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3761018348595254,
+      "learning_rate": 6.695911860306996e-06,
+      "loss": 0.642,
+      "step": 8273
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.38566367371039595,
+      "learning_rate": 6.691587739699199e-06,
+      "loss": 0.6423,
+      "step": 8274
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.37732458101541794,
+      "learning_rate": 6.6872647352518635e-06,
+      "loss": 0.6312,
+      "step": 8275
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.30957130605203353,
+      "learning_rate": 6.6829428473275604e-06,
+      "loss": 0.6371,
+      "step": 8276
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.42756882874132657,
+      "learning_rate": 6.678622076288757e-06,
+      "loss": 0.6819,
+      "step": 8277
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.40896346082321766,
+      "learning_rate": 6.674302422497836e-06,
+      "loss": 0.6248,
+      "step": 8278
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.32690406230981445,
+      "learning_rate": 6.669983886317073e-06,
+      "loss": 0.6084,
+      "step": 8279
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.42589869870718555,
+      "learning_rate": 6.665666468108663e-06,
+      "loss": 0.6953,
+      "step": 8280
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.40334480481546015,
+      "learning_rate": 6.6613501682347085e-06,
+      "loss": 0.5955,
+      "step": 8281
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41298852693629545,
+      "learning_rate": 6.657034987057192e-06,
+      "loss": 0.6516,
+      "step": 8282
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39187203009870686,
+      "learning_rate": 6.652720924938052e-06,
+      "loss": 0.6669,
+      "step": 8283
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.35380451537107965,
+      "learning_rate": 6.64840798223908e-06,
+      "loss": 0.665,
+      "step": 8284
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.38924017765390834,
+      "learning_rate": 6.644096159322009e-06,
+      "loss": 0.6454,
+      "step": 8285
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3592810896485927,
+      "learning_rate": 6.639785456548467e-06,
+      "loss": 0.6391,
+      "step": 8286
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39422000750593333,
+      "learning_rate": 6.635475874279977e-06,
+      "loss": 0.6354,
+      "step": 8287
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.35976501074780376,
+      "learning_rate": 6.631167412877988e-06,
+      "loss": 0.6335,
+      "step": 8288
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41390144795173295,
+      "learning_rate": 6.62686007270384e-06,
+      "loss": 0.6485,
+      "step": 8289
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3986068851991779,
+      "learning_rate": 6.6225538541187875e-06,
+      "loss": 0.6621,
+      "step": 8290
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3870072535044078,
+      "learning_rate": 6.61824875748399e-06,
+      "loss": 0.6196,
+      "step": 8291
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.36135979060208384,
+      "learning_rate": 6.613944783160511e-06,
+      "loss": 0.6339,
+      "step": 8292
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4042667844952443,
+      "learning_rate": 6.6096419315093115e-06,
+      "loss": 0.629,
+      "step": 8293
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.38800198615246384,
+      "learning_rate": 6.605340202891273e-06,
+      "loss": 0.6687,
+      "step": 8294
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3726487063399307,
+      "learning_rate": 6.60103959766718e-06,
+      "loss": 0.6818,
+      "step": 8295
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.37904863911763065,
+      "learning_rate": 6.596740116197702e-06,
+      "loss": 0.622,
+      "step": 8296
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3697109656603032,
+      "learning_rate": 6.592441758843455e-06,
+      "loss": 0.6332,
+      "step": 8297
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41913567458711104,
+      "learning_rate": 6.5881445259649214e-06,
+      "loss": 0.6615,
+      "step": 8298
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4089050603597523,
+      "learning_rate": 6.583848417922507e-06,
+      "loss": 0.6357,
+      "step": 8299
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.49202578726456053,
+      "learning_rate": 6.579553435076525e-06,
+      "loss": 0.7062,
+      "step": 8300
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3566952265454526,
+      "learning_rate": 6.5752595777871915e-06,
+      "loss": 0.6461,
+      "step": 8301
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4286184391527851,
+      "learning_rate": 6.5709668464146214e-06,
+      "loss": 0.6736,
+      "step": 8302
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41192029510188527,
+      "learning_rate": 6.566675241318843e-06,
+      "loss": 0.665,
+      "step": 8303
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41430718311537046,
+      "learning_rate": 6.562384762859788e-06,
+      "loss": 0.6354,
+      "step": 8304
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.40688989407483916,
+      "learning_rate": 6.5580954113972935e-06,
+      "loss": 0.6801,
+      "step": 8305
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.35516151254509354,
+      "learning_rate": 6.5538071872911106e-06,
+      "loss": 0.6769,
+      "step": 8306
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.345980965223104,
+      "learning_rate": 6.549520090900874e-06,
+      "loss": 0.6281,
+      "step": 8307
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3805588576864593,
+      "learning_rate": 6.545234122586144e-06,
+      "loss": 0.6159,
+      "step": 8308
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4463658648563318,
+      "learning_rate": 6.540949282706383e-06,
+      "loss": 0.6739,
+      "step": 8309
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3383300090987438,
+      "learning_rate": 6.53666557162094e-06,
+      "loss": 0.6417,
+      "step": 8310
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.42987979140731164,
+      "learning_rate": 6.532382989689103e-06,
+      "loss": 0.6816,
+      "step": 8311
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3724655543934008,
+      "learning_rate": 6.528101537270044e-06,
+      "loss": 0.6747,
+      "step": 8312
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4440561126698394,
+      "learning_rate": 6.523821214722834e-06,
+      "loss": 0.6812,
+      "step": 8313
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3925232392050374,
+      "learning_rate": 6.519542022406464e-06,
+      "loss": 0.6004,
+      "step": 8314
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3847625521517589,
+      "learning_rate": 6.515263960679832e-06,
+      "loss": 0.6998,
+      "step": 8315
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3601992962499892,
+      "learning_rate": 6.510987029901716e-06,
+      "loss": 0.6229,
+      "step": 8316
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3783081052012872,
+      "learning_rate": 6.506711230430831e-06,
+      "loss": 0.6487,
+      "step": 8317
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4304011053082631,
+      "learning_rate": 6.502436562625778e-06,
+      "loss": 0.6765,
+      "step": 8318
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.35573528008042893,
+      "learning_rate": 6.498163026845068e-06,
+      "loss": 0.669,
+      "step": 8319
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3592953828961496,
+      "learning_rate": 6.493890623447128e-06,
+      "loss": 0.6359,
+      "step": 8320
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.46102266155440635,
+      "learning_rate": 6.489619352790262e-06,
+      "loss": 0.6729,
+      "step": 8321
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3548166229176227,
+      "learning_rate": 6.4853492152327036e-06,
+      "loss": 0.631,
+      "step": 8322
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.38298546233010655,
+      "learning_rate": 6.481080211132587e-06,
+      "loss": 0.6275,
+      "step": 8323
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.36386813343997837,
+      "learning_rate": 6.476812340847944e-06,
+      "loss": 0.6322,
+      "step": 8324
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4324730627056812,
+      "learning_rate": 6.47254560473672e-06,
+      "loss": 0.6813,
+      "step": 8325
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3718019325170493,
+      "learning_rate": 6.468280003156766e-06,
+      "loss": 0.6197,
+      "step": 8326
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3672926232035359,
+      "learning_rate": 6.46401553646582e-06,
+      "loss": 0.6517,
+      "step": 8327
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3582472635114001,
+      "learning_rate": 6.4597522050215414e-06,
+      "loss": 0.6389,
+      "step": 8328
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.38889544464705694,
+      "learning_rate": 6.4554900091815e-06,
+      "loss": 0.6792,
+      "step": 8329
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.388614376518832,
+      "learning_rate": 6.451228949303148e-06,
+      "loss": 0.6472,
+      "step": 8330
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39645470325448917,
+      "learning_rate": 6.446969025743854e-06,
+      "loss": 0.641,
+      "step": 8331
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4028705667736027,
+      "learning_rate": 6.442710238860912e-06,
+      "loss": 0.6305,
+      "step": 8332
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.33416780169674437,
+      "learning_rate": 6.438452589011486e-06,
+      "loss": 0.6514,
+      "step": 8333
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.36317353217906007,
+      "learning_rate": 6.434196076552661e-06,
+      "loss": 0.6192,
+      "step": 8334
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3624036097746673,
+      "learning_rate": 6.4299407018414326e-06,
+      "loss": 0.6588,
+      "step": 8335
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4127607002770979,
+      "learning_rate": 6.425686465234686e-06,
+      "loss": 0.6333,
+      "step": 8336
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3996730267566744,
+      "learning_rate": 6.42143336708922e-06,
+      "loss": 0.6614,
+      "step": 8337
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3669936203297201,
+      "learning_rate": 6.417181407761739e-06,
+      "loss": 0.6491,
+      "step": 8338
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.41393751683987595,
+      "learning_rate": 6.41293058760885e-06,
+      "loss": 0.6888,
+      "step": 8339
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38212769740605657,
+      "learning_rate": 6.408680906987068e-06,
+      "loss": 0.6902,
+      "step": 8340
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3968724220413139,
+      "learning_rate": 6.4044323662527994e-06,
+      "loss": 0.6504,
+      "step": 8341
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38198106873436805,
+      "learning_rate": 6.400184965762368e-06,
+      "loss": 0.6583,
+      "step": 8342
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38368158534232777,
+      "learning_rate": 6.395938705872007e-06,
+      "loss": 0.6283,
+      "step": 8343
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.243158032093463,
+      "learning_rate": 6.39169358693783e-06,
+      "loss": 0.6614,
+      "step": 8344
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4090255922009346,
+      "learning_rate": 6.387449609315872e-06,
+      "loss": 0.6245,
+      "step": 8345
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.40383555831379603,
+      "learning_rate": 6.383206773362087e-06,
+      "loss": 0.6992,
+      "step": 8346
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.37155780105005093,
+      "learning_rate": 6.378965079432298e-06,
+      "loss": 0.6192,
+      "step": 8347
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.36247814242040083,
+      "learning_rate": 6.374724527882259e-06,
+      "loss": 0.6622,
+      "step": 8348
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3495775060507593,
+      "learning_rate": 6.370485119067624e-06,
+      "loss": 0.6512,
+      "step": 8349
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.8387195307811046,
+      "learning_rate": 6.366246853343936e-06,
+      "loss": 6.5472,
+      "step": 8350
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.36271335248643555,
+      "learning_rate": 6.362009731066658e-06,
+      "loss": 0.6157,
+      "step": 8351
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.42932428317219407,
+      "learning_rate": 6.357773752591156e-06,
+      "loss": 0.694,
+      "step": 8352
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.36297737996243234,
+      "learning_rate": 6.353538918272692e-06,
+      "loss": 0.6256,
+      "step": 8353
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3845841188958863,
+      "learning_rate": 6.349305228466444e-06,
+      "loss": 0.6645,
+      "step": 8354
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3668017561197356,
+      "learning_rate": 6.345072683527476e-06,
+      "loss": 0.6567,
+      "step": 8355
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3760405289949169,
+      "learning_rate": 6.340841283810768e-06,
+      "loss": 0.6527,
+      "step": 8356
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.40960064888926084,
+      "learning_rate": 6.336611029671209e-06,
+      "loss": 0.6774,
+      "step": 8357
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3825175267256722,
+      "learning_rate": 6.33238192146358e-06,
+      "loss": 0.6186,
+      "step": 8358
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.37739084384150584,
+      "learning_rate": 6.328153959542573e-06,
+      "loss": 0.6512,
+      "step": 8359
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.37395895090950054,
+      "learning_rate": 6.323927144262787e-06,
+      "loss": 0.6161,
+      "step": 8360
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38197968834147966,
+      "learning_rate": 6.31970147597871e-06,
+      "loss": 0.6421,
+      "step": 8361
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3779606958092785,
+      "learning_rate": 6.315476955044751e-06,
+      "loss": 0.6652,
+      "step": 8362
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3691256036995055,
+      "learning_rate": 6.311253581815215e-06,
+      "loss": 0.633,
+      "step": 8363
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3880594125125574,
+      "learning_rate": 6.307031356644304e-06,
+      "loss": 0.6611,
+      "step": 8364
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.41631605913703884,
+      "learning_rate": 6.302810279886131e-06,
+      "loss": 0.6513,
+      "step": 8365
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3702712425413035,
+      "learning_rate": 6.2985903518947315e-06,
+      "loss": 0.6528,
+      "step": 8366
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3652783209155449,
+      "learning_rate": 6.294371573024003e-06,
+      "loss": 0.5965,
+      "step": 8367
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4259496293412795,
+      "learning_rate": 6.290153943627782e-06,
+      "loss": 0.6282,
+      "step": 8368
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3683234127896203,
+      "learning_rate": 6.285937464059794e-06,
+      "loss": 0.6348,
+      "step": 8369
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38299632831621105,
+      "learning_rate": 6.281722134673667e-06,
+      "loss": 0.6673,
+      "step": 8370
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.37399968779362297,
+      "learning_rate": 6.277507955822935e-06,
+      "loss": 0.6482,
+      "step": 8371
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.34190370230478945,
+      "learning_rate": 6.2732949278610374e-06,
+      "loss": 0.6294,
+      "step": 8372
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3555754520738616,
+      "learning_rate": 6.269083051141318e-06,
+      "loss": 0.6286,
+      "step": 8373
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.40204371862773214,
+      "learning_rate": 6.264872326017024e-06,
+      "loss": 0.6356,
+      "step": 8374
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.35318997419123543,
+      "learning_rate": 6.260662752841298e-06,
+      "loss": 0.6706,
+      "step": 8375
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3882693148438636,
+      "learning_rate": 6.25645433196719e-06,
+      "loss": 0.6814,
+      "step": 8376
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38404957958422475,
+      "learning_rate": 6.252247063747668e-06,
+      "loss": 0.6465,
+      "step": 8377
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.380088693934544,
+      "learning_rate": 6.248040948535574e-06,
+      "loss": 0.6744,
+      "step": 8378
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38990544464549426,
+      "learning_rate": 6.243835986683673e-06,
+      "loss": 0.6548,
+      "step": 8379
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.37744784152623,
+      "learning_rate": 6.2396321785446456e-06,
+      "loss": 0.6129,
+      "step": 8380
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.35643983376311456,
+      "learning_rate": 6.235429524471044e-06,
+      "loss": 0.6243,
+      "step": 8381
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4189868167783136,
+      "learning_rate": 6.2312280248153435e-06,
+      "loss": 0.6768,
+      "step": 8382
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3442884197244239,
+      "learning_rate": 6.227027679929927e-06,
+      "loss": 0.6525,
+      "step": 8383
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3659967134275989,
+      "learning_rate": 6.222828490167059e-06,
+      "loss": 0.6335,
+      "step": 8384
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.42982373015672914,
+      "learning_rate": 6.218630455878926e-06,
+      "loss": 0.6701,
+      "step": 8385
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3611164518409403,
+      "learning_rate": 6.214433577417614e-06,
+      "loss": 0.6489,
+      "step": 8386
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3535728847441064,
+      "learning_rate": 6.210237855135109e-06,
+      "loss": 0.6177,
+      "step": 8387
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38660179141218365,
+      "learning_rate": 6.206043289383301e-06,
+      "loss": 0.6385,
+      "step": 8388
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4045386464542476,
+      "learning_rate": 6.201849880513991e-06,
+      "loss": 0.6892,
+      "step": 8389
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3987958039167017,
+      "learning_rate": 6.197657628878859e-06,
+      "loss": 0.6494,
+      "step": 8390
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39911045338420903,
+      "learning_rate": 6.1934665348295155e-06,
+      "loss": 0.6311,
+      "step": 8391
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3747966312323323,
+      "learning_rate": 6.189276598717464e-06,
+      "loss": 0.6673,
+      "step": 8392
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4100451839252102,
+      "learning_rate": 6.185087820894091e-06,
+      "loss": 0.669,
+      "step": 8393
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39068753774541426,
+      "learning_rate": 6.180900201710733e-06,
+      "loss": 0.6763,
+      "step": 8394
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3497850229222053,
+      "learning_rate": 6.176713741518581e-06,
+      "loss": 0.6161,
+      "step": 8395
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39899189007627883,
+      "learning_rate": 6.17252844066875e-06,
+      "loss": 0.6397,
+      "step": 8396
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39210234412952816,
+      "learning_rate": 6.1683442995122675e-06,
+      "loss": 0.6081,
+      "step": 8397
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.36050736955203333,
+      "learning_rate": 6.1641613184000374e-06,
+      "loss": 0.6437,
+      "step": 8398
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.376596082105252,
+      "learning_rate": 6.159979497682889e-06,
+      "loss": 0.6311,
+      "step": 8399
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3555032071066738,
+      "learning_rate": 6.1557988377115444e-06,
+      "loss": 0.6228,
+      "step": 8400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.37457243915687843,
+      "learning_rate": 6.1516193388366315e-06,
+      "loss": 0.5962,
+      "step": 8401
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.423170613793651,
+      "learning_rate": 6.1474410014086805e-06,
+      "loss": 0.6646,
+      "step": 8402
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.45101574171166614,
+      "learning_rate": 6.14326382577813e-06,
+      "loss": 0.635,
+      "step": 8403
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3862185290174527,
+      "learning_rate": 6.139087812295302e-06,
+      "loss": 0.6501,
+      "step": 8404
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3874949999308747,
+      "learning_rate": 6.134912961310438e-06,
+      "loss": 0.6094,
+      "step": 8405
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.37819364962757657,
+      "learning_rate": 6.130739273173685e-06,
+      "loss": 0.6262,
+      "step": 8406
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3885753666742401,
+      "learning_rate": 6.126566748235068e-06,
+      "loss": 0.6518,
+      "step": 8407
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39622921860779375,
+      "learning_rate": 6.122395386844555e-06,
+      "loss": 0.6551,
+      "step": 8408
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4274413518808349,
+      "learning_rate": 6.118225189351974e-06,
+      "loss": 0.6868,
+      "step": 8409
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3880247952137593,
+      "learning_rate": 6.114056156107082e-06,
+      "loss": 0.6993,
+      "step": 8410
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3992537439643188,
+      "learning_rate": 6.1098882874595375e-06,
+      "loss": 0.6388,
+      "step": 8411
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38147219539660226,
+      "learning_rate": 6.105721583758879e-06,
+      "loss": 0.6709,
+      "step": 8412
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.938381478077772,
+      "learning_rate": 6.101556045354573e-06,
+      "loss": 6.6819,
+      "step": 8413
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.400351533575736,
+      "learning_rate": 6.0973916725959755e-06,
+      "loss": 0.6168,
+      "step": 8414
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.44392212314430946,
+      "learning_rate": 6.0932284658323486e-06,
+      "loss": 0.6462,
+      "step": 8415
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.40404922834768797,
+      "learning_rate": 6.089066425412855e-06,
+      "loss": 0.6564,
+      "step": 8416
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39836426342294834,
+      "learning_rate": 6.084905551686569e-06,
+      "loss": 0.6496,
+      "step": 8417
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5325522946323632,
+      "learning_rate": 6.080745845002438e-06,
+      "loss": 0.6477,
+      "step": 8418
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.34153233158893703,
+      "learning_rate": 6.076587305709347e-06,
+      "loss": 0.6385,
+      "step": 8419
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.41708800181418154,
+      "learning_rate": 6.0724299341560676e-06,
+      "loss": 0.6478,
+      "step": 8420
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4108781809606789,
+      "learning_rate": 6.068273730691256e-06,
+      "loss": 0.5951,
+      "step": 8421
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38503493553698165,
+      "learning_rate": 6.06411869566351e-06,
+      "loss": 0.6516,
+      "step": 8422
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.40005610808491315,
+      "learning_rate": 6.059964829421303e-06,
+      "loss": 0.6509,
+      "step": 8423
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3530206928766137,
+      "learning_rate": 6.055812132313006e-06,
+      "loss": 0.6158,
+      "step": 8424
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3466724250957284,
+      "learning_rate": 6.051660604686905e-06,
+      "loss": 0.6274,
+      "step": 8425
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4149678576827248,
+      "learning_rate": 6.047510246891188e-06,
+      "loss": 0.665,
+      "step": 8426
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.810260972752004,
+      "learning_rate": 6.043361059273931e-06,
+      "loss": 6.7908,
+      "step": 8427
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4411446892058144,
+      "learning_rate": 6.039213042183127e-06,
+      "loss": 0.681,
+      "step": 8428
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.419981852836698,
+      "learning_rate": 6.035066195966666e-06,
+      "loss": 0.6439,
+      "step": 8429
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.43610330452718127,
+      "learning_rate": 6.030920520972336e-06,
+      "loss": 0.7059,
+      "step": 8430
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3366922868276612,
+      "learning_rate": 6.026776017547837e-06,
+      "loss": 0.6551,
+      "step": 8431
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3709947679209238,
+      "learning_rate": 6.022632686040755e-06,
+      "loss": 0.6535,
+      "step": 8432
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.35285329257316567,
+      "learning_rate": 6.018490526798586e-06,
+      "loss": 0.6357,
+      "step": 8433
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3979440347005264,
+      "learning_rate": 6.014349540168734e-06,
+      "loss": 0.6516,
+      "step": 8434
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39972141814522155,
+      "learning_rate": 6.010209726498493e-06,
+      "loss": 0.6442,
+      "step": 8435
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3840891716754943,
+      "learning_rate": 6.006071086135068e-06,
+      "loss": 0.6868,
+      "step": 8436
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39033848711611085,
+      "learning_rate": 6.0019336194255665e-06,
+      "loss": 0.6446,
+      "step": 8437
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.42726914649382636,
+      "learning_rate": 5.997797326716981e-06,
+      "loss": 0.6731,
+      "step": 8438
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.36243414574242633,
+      "learning_rate": 5.993662208356223e-06,
+      "loss": 0.616,
+      "step": 8439
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3852512420093456,
+      "learning_rate": 5.989528264690107e-06,
+      "loss": 0.6488,
+      "step": 8440
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38153002406465203,
+      "learning_rate": 5.985395496065322e-06,
+      "loss": 0.6155,
+      "step": 8441
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.34114077921984115,
+      "learning_rate": 5.9812639028285045e-06,
+      "loss": 0.6195,
+      "step": 8442
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39517276405933166,
+      "learning_rate": 5.977133485326148e-06,
+      "loss": 0.6683,
+      "step": 8443
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3593754646086528,
+      "learning_rate": 5.973004243904667e-06,
+      "loss": 0.6526,
+      "step": 8444
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4231000802326297,
+      "learning_rate": 5.968876178910383e-06,
+      "loss": 0.6847,
+      "step": 8445
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.41831125410007497,
+      "learning_rate": 5.964749290689513e-06,
+      "loss": 0.6364,
+      "step": 8446
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.41942494550330106,
+      "learning_rate": 5.960623579588165e-06,
+      "loss": 0.6641,
+      "step": 8447
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.34965844351913206,
+      "learning_rate": 5.956499045952362e-06,
+      "loss": 0.6453,
+      "step": 8448
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6369749820809316,
+      "learning_rate": 5.952375690128023e-06,
+      "loss": 0.6334,
+      "step": 8449
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.44446425939799883,
+      "learning_rate": 5.948253512460971e-06,
+      "loss": 0.6684,
+      "step": 8450
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.38371779719925725,
+      "learning_rate": 5.94413251329693e-06,
+      "loss": 0.6273,
+      "step": 8451
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4085728777323221,
+      "learning_rate": 5.940012692981516e-06,
+      "loss": 0.6565,
+      "step": 8452
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.34832709267277195,
+      "learning_rate": 5.935894051860256e-06,
+      "loss": 0.6083,
+      "step": 8453
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.37074558341735436,
+      "learning_rate": 5.931776590278584e-06,
+      "loss": 0.6327,
+      "step": 8454
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.41079512372692983,
+      "learning_rate": 5.927660308581806e-06,
+      "loss": 0.662,
+      "step": 8455
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3642102258713191,
+      "learning_rate": 5.923545207115171e-06,
+      "loss": 0.6662,
+      "step": 8456
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.381304581064202,
+      "learning_rate": 5.9194312862238045e-06,
+      "loss": 0.6255,
+      "step": 8457
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3945463763247022,
+      "learning_rate": 5.915318546252724e-06,
+      "loss": 0.6406,
+      "step": 8458
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3653325034945878,
+      "learning_rate": 5.911206987546867e-06,
+      "loss": 0.6407,
+      "step": 8459
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3785194186006019,
+      "learning_rate": 5.907096610451073e-06,
+      "loss": 0.6374,
+      "step": 8460
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4043372972345936,
+      "learning_rate": 5.902987415310057e-06,
+      "loss": 0.6345,
+      "step": 8461
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4001380892561691,
+      "learning_rate": 5.898879402468465e-06,
+      "loss": 0.6423,
+      "step": 8462
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3663133361617184,
+      "learning_rate": 5.894772572270826e-06,
+      "loss": 0.6508,
+      "step": 8463
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.35739963591893364,
+      "learning_rate": 5.890666925061576e-06,
+      "loss": 0.6299,
+      "step": 8464
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4240780908255833,
+      "learning_rate": 5.8865624611850595e-06,
+      "loss": 0.709,
+      "step": 8465
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.379582713846009,
+      "learning_rate": 5.8824591809854955e-06,
+      "loss": 0.6567,
+      "step": 8466
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4165679580948523,
+      "learning_rate": 5.878357084807034e-06,
+      "loss": 0.6592,
+      "step": 8467
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3637259161829086,
+      "learning_rate": 5.874256172993713e-06,
+      "loss": 0.6121,
+      "step": 8468
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3722266863103741,
+      "learning_rate": 5.8701564458894545e-06,
+      "loss": 0.6253,
+      "step": 8469
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.39366622913194615,
+      "learning_rate": 5.866057903838116e-06,
+      "loss": 0.624,
+      "step": 8470
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3897300324519267,
+      "learning_rate": 5.86196054718344e-06,
+      "loss": 0.6529,
+      "step": 8471
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3817666122384533,
+      "learning_rate": 5.857864376269051e-06,
+      "loss": 0.6075,
+      "step": 8472
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3726946403622256,
+      "learning_rate": 5.853769391438497e-06,
+      "loss": 0.6593,
+      "step": 8473
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3330703776394616,
+      "learning_rate": 5.849675593035226e-06,
+      "loss": 0.6171,
+      "step": 8474
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.43236408147994215,
+      "learning_rate": 5.845582981402569e-06,
+      "loss": 0.622,
+      "step": 8475
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.37649489874406833,
+      "learning_rate": 5.841491556883767e-06,
+      "loss": 0.5817,
+      "step": 8476
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3631104600003091,
+      "learning_rate": 5.837401319821981e-06,
+      "loss": 0.6294,
+      "step": 8477
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3674844437307317,
+      "learning_rate": 5.8333122705602364e-06,
+      "loss": 0.6456,
+      "step": 8478
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3519942216722405,
+      "learning_rate": 5.82922440944148e-06,
+      "loss": 0.6152,
+      "step": 8479
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.39100648218353673,
+      "learning_rate": 5.825137736808568e-06,
+      "loss": 0.6554,
+      "step": 8480
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3752868068443191,
+      "learning_rate": 5.821052253004227e-06,
+      "loss": 0.6753,
+      "step": 8481
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.38147381079229997,
+      "learning_rate": 5.816967958371109e-06,
+      "loss": 0.6646,
+      "step": 8482
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4236295310598535,
+      "learning_rate": 5.812884853251761e-06,
+      "loss": 0.6553,
+      "step": 8483
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.40633412487637005,
+      "learning_rate": 5.8088029379886265e-06,
+      "loss": 0.6661,
+      "step": 8484
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.42957905472198804,
+      "learning_rate": 5.804722212924056e-06,
+      "loss": 0.6706,
+      "step": 8485
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3640484215123666,
+      "learning_rate": 5.800642678400286e-06,
+      "loss": 0.6029,
+      "step": 8486
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.39926913247417445,
+      "learning_rate": 5.796564334759464e-06,
+      "loss": 0.6381,
+      "step": 8487
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.41529338691608186,
+      "learning_rate": 5.792487182343645e-06,
+      "loss": 0.6444,
+      "step": 8488
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.36777358492178086,
+      "learning_rate": 5.788411221494761e-06,
+      "loss": 0.6583,
+      "step": 8489
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.35092658753764,
+      "learning_rate": 5.7843364525546595e-06,
+      "loss": 0.6304,
+      "step": 8490
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4144577192477767,
+      "learning_rate": 5.780262875865103e-06,
+      "loss": 0.6423,
+      "step": 8491
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.41432888805937185,
+      "learning_rate": 5.7761904917677195e-06,
+      "loss": 0.658,
+      "step": 8492
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.37085976002657334,
+      "learning_rate": 5.772119300604062e-06,
+      "loss": 0.6452,
+      "step": 8493
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.40766124405318915,
+      "learning_rate": 5.768049302715582e-06,
+      "loss": 0.634,
+      "step": 8494
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.43783664196072014,
+      "learning_rate": 5.763980498443614e-06,
+      "loss": 0.6857,
+      "step": 8495
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3696913129202887,
+      "learning_rate": 5.759912888129406e-06,
+      "loss": 0.63,
+      "step": 8496
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.37086885090046817,
+      "learning_rate": 5.755846472114108e-06,
+      "loss": 0.5925,
+      "step": 8497
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.43929572438887354,
+      "learning_rate": 5.751781250738764e-06,
+      "loss": 0.6427,
+      "step": 8498
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 4.679075658743053,
+      "learning_rate": 5.747717224344327e-06,
+      "loss": 6.6215,
+      "step": 8499
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.34669936203222035,
+      "learning_rate": 5.743654393271625e-06,
+      "loss": 0.6033,
+      "step": 8500
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3683147924418471,
+      "learning_rate": 5.739592757861412e-06,
+      "loss": 0.662,
+      "step": 8501
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3533795414978267,
+      "learning_rate": 5.735532318454333e-06,
+      "loss": 0.6458,
+      "step": 8502
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.42345066421020594,
+      "learning_rate": 5.731473075390939e-06,
+      "loss": 0.628,
+      "step": 8503
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.42732590475912074,
+      "learning_rate": 5.7274150290116516e-06,
+      "loss": 0.6807,
+      "step": 8504
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4211473275096684,
+      "learning_rate": 5.723358179656842e-06,
+      "loss": 0.6672,
+      "step": 8505
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.38699247989504365,
+      "learning_rate": 5.719302527666733e-06,
+      "loss": 0.6545,
+      "step": 8506
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.36362125110239896,
+      "learning_rate": 5.715248073381474e-06,
+      "loss": 0.6554,
+      "step": 8507
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.398166263173972,
+      "learning_rate": 5.711194817141115e-06,
+      "loss": 0.6364,
+      "step": 8508
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.37373885098782256,
+      "learning_rate": 5.7071427592855845e-06,
+      "loss": 0.6592,
+      "step": 8509
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.39502108690681714,
+      "learning_rate": 5.703091900154727e-06,
+      "loss": 0.6329,
+      "step": 8510
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.32117799037078,
+      "learning_rate": 5.699042240088289e-06,
+      "loss": 0.6116,
+      "step": 8511
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.38535320327810074,
+      "learning_rate": 5.6949937794259035e-06,
+      "loss": 0.6529,
+      "step": 8512
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.38763490247856575,
+      "learning_rate": 5.690946518507115e-06,
+      "loss": 0.6247,
+      "step": 8513
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.42355099717265215,
+      "learning_rate": 5.686900457671367e-06,
+      "loss": 0.6727,
+      "step": 8514
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3842248037878429,
+      "learning_rate": 5.682855597257986e-06,
+      "loss": 0.6526,
+      "step": 8515
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4202733272184111,
+      "learning_rate": 5.6788119376062166e-06,
+      "loss": 0.6339,
+      "step": 8516
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3718118474474032,
+      "learning_rate": 5.674769479055193e-06,
+      "loss": 0.6315,
+      "step": 8517
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4005524794206198,
+      "learning_rate": 5.670728221943953e-06,
+      "loss": 0.6512,
+      "step": 8518
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.35236064353068847,
+      "learning_rate": 5.666688166611438e-06,
+      "loss": 0.6665,
+      "step": 8519
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.39604430537277463,
+      "learning_rate": 5.6626493133964684e-06,
+      "loss": 0.6764,
+      "step": 8520
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3822532302331933,
+      "learning_rate": 5.6586116626377875e-06,
+      "loss": 0.602,
+      "step": 8521
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.2159436283418876,
+      "learning_rate": 5.654575214674032e-06,
+      "loss": 6.6097,
+      "step": 8522
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.38023475343735347,
+      "learning_rate": 5.650539969843723e-06,
+      "loss": 0.6009,
+      "step": 8523
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.37120602461346025,
+      "learning_rate": 5.6465059284852884e-06,
+      "loss": 0.6311,
+      "step": 8524
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4262561262126476,
+      "learning_rate": 5.642473090937082e-06,
+      "loss": 0.6528,
+      "step": 8525
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.45168201388168155,
+      "learning_rate": 5.638441457537311e-06,
+      "loss": 0.6543,
+      "step": 8526
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3433231812293748,
+      "learning_rate": 5.6344110286241095e-06,
+      "loss": 0.627,
+      "step": 8527
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3959379187081622,
+      "learning_rate": 5.630381804535514e-06,
+      "loss": 0.6095,
+      "step": 8528
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3560369722069603,
+      "learning_rate": 5.6263537856094355e-06,
+      "loss": 0.6399,
+      "step": 8529
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3735662814786476,
+      "learning_rate": 5.622326972183705e-06,
+      "loss": 0.6366,
+      "step": 8530
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3897416040476952,
+      "learning_rate": 5.618301364596048e-06,
+      "loss": 0.6384,
+      "step": 8531
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.35728645783616636,
+      "learning_rate": 5.614276963184087e-06,
+      "loss": 0.6052,
+      "step": 8532
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3914795865464401,
+      "learning_rate": 5.610253768285343e-06,
+      "loss": 0.6663,
+      "step": 8533
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4111472221094283,
+      "learning_rate": 5.606231780237246e-06,
+      "loss": 0.6489,
+      "step": 8534
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3552060053832923,
+      "learning_rate": 5.602210999377096e-06,
+      "loss": 0.6404,
+      "step": 8535
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4039138995476978,
+      "learning_rate": 5.598191426042123e-06,
+      "loss": 0.6275,
+      "step": 8536
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.36128421835255525,
+      "learning_rate": 5.59417306056945e-06,
+      "loss": 0.6533,
+      "step": 8537
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4270991783221654,
+      "learning_rate": 5.59015590329607e-06,
+      "loss": 0.637,
+      "step": 8538
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3991524689462374,
+      "learning_rate": 5.5861399545589245e-06,
+      "loss": 0.656,
+      "step": 8539
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.39217882836777174,
+      "learning_rate": 5.582125214694809e-06,
+      "loss": 0.6624,
+      "step": 8540
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3730425965135344,
+      "learning_rate": 5.578111684040437e-06,
+      "loss": 0.6488,
+      "step": 8541
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.35595748937515737,
+      "learning_rate": 5.574099362932428e-06,
+      "loss": 0.6426,
+      "step": 8542
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.38025819754956747,
+      "learning_rate": 5.570088251707278e-06,
+      "loss": 0.6107,
+      "step": 8543
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4170526035723036,
+      "learning_rate": 5.566078350701398e-06,
+      "loss": 0.6765,
+      "step": 8544
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.36117482419923536,
+      "learning_rate": 5.562069660251097e-06,
+      "loss": 0.618,
+      "step": 8545
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.35337901913410064,
+      "learning_rate": 5.558062180692574e-06,
+      "loss": 0.6234,
+      "step": 8546
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.40013124485147383,
+      "learning_rate": 5.554055912361937e-06,
+      "loss": 0.644,
+      "step": 8547
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.41365321201469835,
+      "learning_rate": 5.550050855595189e-06,
+      "loss": 0.6334,
+      "step": 8548
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4076961688709984,
+      "learning_rate": 5.54604701072822e-06,
+      "loss": 0.6113,
+      "step": 8549
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4366181747321968,
+      "learning_rate": 5.542044378096831e-06,
+      "loss": 0.6668,
+      "step": 8550
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3420507105706567,
+      "learning_rate": 5.538042958036724e-06,
+      "loss": 0.5938,
+      "step": 8551
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3677067992350431,
+      "learning_rate": 5.534042750883479e-06,
+      "loss": 0.6573,
+      "step": 8552
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3826474878433704,
+      "learning_rate": 5.530043756972609e-06,
+      "loss": 0.6483,
+      "step": 8553
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5048502643449844,
+      "learning_rate": 5.526045976639487e-06,
+      "loss": 0.7213,
+      "step": 8554
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4122917181415525,
+      "learning_rate": 5.52204941021941e-06,
+      "loss": 0.5975,
+      "step": 8555
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.39269266942765985,
+      "learning_rate": 5.518054058047571e-06,
+      "loss": 0.6401,
+      "step": 8556
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4151010916270037,
+      "learning_rate": 5.514059920459041e-06,
+      "loss": 0.6678,
+      "step": 8557
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4005786242743759,
+      "learning_rate": 5.51006699778881e-06,
+      "loss": 0.6165,
+      "step": 8558
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.41262436317457074,
+      "learning_rate": 5.5060752903717615e-06,
+      "loss": 0.5983,
+      "step": 8559
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3404957279383971,
+      "learning_rate": 5.502084798542675e-06,
+      "loss": 0.6531,
+      "step": 8560
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3848832049422809,
+      "learning_rate": 5.498095522636226e-06,
+      "loss": 0.6262,
+      "step": 8561
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3249923407419005,
+      "learning_rate": 5.494107462987e-06,
+      "loss": 0.6123,
+      "step": 8562
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3742791803102254,
+      "learning_rate": 5.490120619929453e-06,
+      "loss": 0.6035,
+      "step": 8563
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.417201435884242,
+      "learning_rate": 5.486134993797967e-06,
+      "loss": 0.6683,
+      "step": 8564
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3312272023551598,
+      "learning_rate": 5.482150584926818e-06,
+      "loss": 0.6409,
+      "step": 8565
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.39395214380900634,
+      "learning_rate": 5.478167393650153e-06,
+      "loss": 0.6914,
+      "step": 8566
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.377853941293888,
+      "learning_rate": 5.474185420302058e-06,
+      "loss": 0.6799,
+      "step": 8567
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.39343461255469975,
+      "learning_rate": 5.470204665216494e-06,
+      "loss": 0.6532,
+      "step": 8568
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3829576226693883,
+      "learning_rate": 5.466225128727314e-06,
+      "loss": 0.6608,
+      "step": 8569
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3880644370296402,
+      "learning_rate": 5.462246811168277e-06,
+      "loss": 0.6645,
+      "step": 8570
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3229398396116435,
+      "learning_rate": 5.45826971287305e-06,
+      "loss": 0.6536,
+      "step": 8571
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.37886081251391573,
+      "learning_rate": 5.454293834175173e-06,
+      "loss": 0.6324,
+      "step": 8572
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3550995299928253,
+      "learning_rate": 5.450319175408107e-06,
+      "loss": 0.619,
+      "step": 8573
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.36201212843132635,
+      "learning_rate": 5.4463457369051984e-06,
+      "loss": 0.6181,
+      "step": 8574
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3929808004586688,
+      "learning_rate": 5.4423735189996976e-06,
+      "loss": 0.6839,
+      "step": 8575
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3655985852566005,
+      "learning_rate": 5.438402522024755e-06,
+      "loss": 0.5833,
+      "step": 8576
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3654052235359208,
+      "learning_rate": 5.4344327463134005e-06,
+      "loss": 0.628,
+      "step": 8577
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3714368173839614,
+      "learning_rate": 5.4304641921985814e-06,
+      "loss": 0.6506,
+      "step": 8578
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3474252753015229,
+      "learning_rate": 5.42649686001314e-06,
+      "loss": 0.6467,
+      "step": 8579
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38620651020672303,
+      "learning_rate": 5.422530750089794e-06,
+      "loss": 0.6422,
+      "step": 8580
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3871331507728079,
+      "learning_rate": 5.418565862761196e-06,
+      "loss": 0.6647,
+      "step": 8581
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.40760762103588377,
+      "learning_rate": 5.414602198359875e-06,
+      "loss": 0.6556,
+      "step": 8582
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.37695353543802296,
+      "learning_rate": 5.410639757218248e-06,
+      "loss": 0.6457,
+      "step": 8583
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3581208952060035,
+      "learning_rate": 5.406678539668644e-06,
+      "loss": 0.6661,
+      "step": 8584
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4174255145827041,
+      "learning_rate": 5.402718546043293e-06,
+      "loss": 0.6355,
+      "step": 8585
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.39894683589478713,
+      "learning_rate": 5.3987597766743006e-06,
+      "loss": 0.6659,
+      "step": 8586
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.43614508089381504,
+      "learning_rate": 5.394802231893694e-06,
+      "loss": 0.6844,
+      "step": 8587
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.37845902368021833,
+      "learning_rate": 5.390845912033385e-06,
+      "loss": 0.6639,
+      "step": 8588
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3576339089667608,
+      "learning_rate": 5.386890817425185e-06,
+      "loss": 0.6507,
+      "step": 8589
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4127690301902639,
+      "learning_rate": 5.382936948400805e-06,
+      "loss": 0.6993,
+      "step": 8590
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.550388366350205,
+      "learning_rate": 5.378984305291857e-06,
+      "loss": 0.666,
+      "step": 8591
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.45229887385169115,
+      "learning_rate": 5.3750328884298295e-06,
+      "loss": 0.6545,
+      "step": 8592
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3714133698470979,
+      "learning_rate": 5.3710826981461325e-06,
+      "loss": 0.6481,
+      "step": 8593
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3461418833602548,
+      "learning_rate": 5.367133734772063e-06,
+      "loss": 0.654,
+      "step": 8594
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38728223995877836,
+      "learning_rate": 5.363185998638814e-06,
+      "loss": 0.6418,
+      "step": 8595
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4175107948050827,
+      "learning_rate": 5.359239490077486e-06,
+      "loss": 0.62,
+      "step": 8596
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3732257743612378,
+      "learning_rate": 5.355294209419053e-06,
+      "loss": 0.6406,
+      "step": 8597
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3782011971986034,
+      "learning_rate": 5.351350156994408e-06,
+      "loss": 0.6608,
+      "step": 8598
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.36443411192748626,
+      "learning_rate": 5.34740733313434e-06,
+      "loss": 0.6195,
+      "step": 8599
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4008747785561861,
+      "learning_rate": 5.343465738169513e-06,
+      "loss": 0.6197,
+      "step": 8600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3788221381969775,
+      "learning_rate": 5.3395253724305185e-06,
+      "loss": 0.654,
+      "step": 8601
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.599305774356224,
+      "learning_rate": 5.3355862362478315e-06,
+      "loss": 6.6034,
+      "step": 8602
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.568768356190587,
+      "learning_rate": 5.331648329951813e-06,
+      "loss": 6.6429,
+      "step": 8603
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.42016838050827604,
+      "learning_rate": 5.3277116538727316e-06,
+      "loss": 0.6492,
+      "step": 8604
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.36935952017009777,
+      "learning_rate": 5.32377620834076e-06,
+      "loss": 0.644,
+      "step": 8605
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.516522026319349,
+      "learning_rate": 5.319841993685951e-06,
+      "loss": 6.7698,
+      "step": 8606
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.41895287490879335,
+      "learning_rate": 5.315909010238261e-06,
+      "loss": 0.6595,
+      "step": 8607
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3631484705919941,
+      "learning_rate": 5.31197725832755e-06,
+      "loss": 0.6418,
+      "step": 8608
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4483930003381706,
+      "learning_rate": 5.308046738283568e-06,
+      "loss": 0.6325,
+      "step": 8609
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.39385571441850825,
+      "learning_rate": 5.304117450435966e-06,
+      "loss": 0.6648,
+      "step": 8610
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3940129195759514,
+      "learning_rate": 5.300189395114281e-06,
+      "loss": 0.6255,
+      "step": 8611
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4041116176603845,
+      "learning_rate": 5.296262572647961e-06,
+      "loss": 0.6444,
+      "step": 8612
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.3913277455004525,
+      "learning_rate": 5.2923369833663444e-06,
+      "loss": 0.6284,
+      "step": 8613
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3777636862510042,
+      "learning_rate": 5.288412627598652e-06,
+      "loss": 0.6294,
+      "step": 8614
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38161590202360823,
+      "learning_rate": 5.284489505674031e-06,
+      "loss": 0.6351,
+      "step": 8615
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3512919757083623,
+      "learning_rate": 5.280567617921508e-06,
+      "loss": 0.6456,
+      "step": 8616
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.41030082853007677,
+      "learning_rate": 5.276646964669998e-06,
+      "loss": 0.6856,
+      "step": 8617
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4033127262578794,
+      "learning_rate": 5.2727275462483265e-06,
+      "loss": 0.6635,
+      "step": 8618
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3785574257539552,
+      "learning_rate": 5.268809362985215e-06,
+      "loss": 0.6429,
+      "step": 8619
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.40061969292148275,
+      "learning_rate": 5.264892415209264e-06,
+      "loss": 0.6361,
+      "step": 8620
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3479903248562543,
+      "learning_rate": 5.260976703248992e-06,
+      "loss": 0.6356,
+      "step": 8621
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.43305070983751554,
+      "learning_rate": 5.257062227432803e-06,
+      "loss": 0.6367,
+      "step": 8622
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3874956667640722,
+      "learning_rate": 5.253148988089001e-06,
+      "loss": 0.639,
+      "step": 8623
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4267179199634148,
+      "learning_rate": 5.249236985545783e-06,
+      "loss": 0.6437,
+      "step": 8624
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3419327598836084,
+      "learning_rate": 5.2453262201312505e-06,
+      "loss": 0.6547,
+      "step": 8625
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.37050886810571027,
+      "learning_rate": 5.241416692173382e-06,
+      "loss": 0.6002,
+      "step": 8626
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38939272096780586,
+      "learning_rate": 5.237508402000073e-06,
+      "loss": 0.6481,
+      "step": 8627
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3715126178295252,
+      "learning_rate": 5.233601349939106e-06,
+      "loss": 0.6222,
+      "step": 8628
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38114790156094347,
+      "learning_rate": 5.22969553631816e-06,
+      "loss": 0.653,
+      "step": 8629
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4131653283816001,
+      "learning_rate": 5.225790961464816e-06,
+      "loss": 0.6596,
+      "step": 8630
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.40740641343673945,
+      "learning_rate": 5.221887625706537e-06,
+      "loss": 0.6698,
+      "step": 8631
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4333440988997623,
+      "learning_rate": 5.217985529370695e-06,
+      "loss": 0.6806,
+      "step": 8632
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.36574861511441026,
+      "learning_rate": 5.214084672784561e-06,
+      "loss": 0.6257,
+      "step": 8633
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.45402166999312915,
+      "learning_rate": 5.210185056275281e-06,
+      "loss": 0.6746,
+      "step": 8634
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38622839733314124,
+      "learning_rate": 5.206286680169914e-06,
+      "loss": 0.6485,
+      "step": 8635
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3850600578336008,
+      "learning_rate": 5.20238954479543e-06,
+      "loss": 0.6622,
+      "step": 8636
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 3.613636888167501,
+      "learning_rate": 5.198493650478656e-06,
+      "loss": 6.7541,
+      "step": 8637
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.39107933615512913,
+      "learning_rate": 5.194598997546343e-06,
+      "loss": 0.6631,
+      "step": 8638
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.36633940439991536,
+      "learning_rate": 5.19070558632514e-06,
+      "loss": 0.6205,
+      "step": 8639
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.39287338837355806,
+      "learning_rate": 5.186813417141567e-06,
+      "loss": 0.6303,
+      "step": 8640
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4087446204642752,
+      "learning_rate": 5.182922490322064e-06,
+      "loss": 0.6404,
+      "step": 8641
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38015101987228744,
+      "learning_rate": 5.179032806192956e-06,
+      "loss": 0.6219,
+      "step": 8642
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3480246371123081,
+      "learning_rate": 5.175144365080469e-06,
+      "loss": 0.62,
+      "step": 8643
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38489045020974244,
+      "learning_rate": 5.171257167310724e-06,
+      "loss": 0.6325,
+      "step": 8644
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.40684587781410275,
+      "learning_rate": 5.1673712132097266e-06,
+      "loss": 0.6768,
+      "step": 8645
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3575054992420779,
+      "learning_rate": 5.163486503103394e-06,
+      "loss": 0.6203,
+      "step": 8646
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.373052392688242,
+      "learning_rate": 5.1596030373175285e-06,
+      "loss": 0.6398,
+      "step": 8647
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38407668411790047,
+      "learning_rate": 5.15572081617784e-06,
+      "loss": 0.6381,
+      "step": 8648
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.198545144564807,
+      "learning_rate": 5.151839840009907e-06,
+      "loss": 6.6081,
+      "step": 8649
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38393169574786634,
+      "learning_rate": 5.147960109139248e-06,
+      "loss": 0.6591,
+      "step": 8650
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38191276270787494,
+      "learning_rate": 5.144081623891231e-06,
+      "loss": 0.6369,
+      "step": 8651
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3791946483684646,
+      "learning_rate": 5.140204384591145e-06,
+      "loss": 0.6532,
+      "step": 8652
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3878168429145577,
+      "learning_rate": 5.13632839156418e-06,
+      "loss": 0.6343,
+      "step": 8653
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3877719877095599,
+      "learning_rate": 5.1324536451353955e-06,
+      "loss": 0.6205,
+      "step": 8654
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3831482616022292,
+      "learning_rate": 5.128580145629767e-06,
+      "loss": 0.6505,
+      "step": 8655
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3942981069479881,
+      "learning_rate": 5.1247078933721625e-06,
+      "loss": 0.6496,
+      "step": 8656
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4013344198558401,
+      "learning_rate": 5.120836888687342e-06,
+      "loss": 0.651,
+      "step": 8657
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3631584783816973,
+      "learning_rate": 5.116967131899961e-06,
+      "loss": 0.6215,
+      "step": 8658
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3821820827080799,
+      "learning_rate": 5.1130986233345805e-06,
+      "loss": 0.6537,
+      "step": 8659
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.36597661498072726,
+      "learning_rate": 5.109231363315632e-06,
+      "loss": 0.6703,
+      "step": 8660
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3935852880254783,
+      "learning_rate": 5.105365352167466e-06,
+      "loss": 0.6403,
+      "step": 8661
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3814190194146751,
+      "learning_rate": 5.101500590214328e-06,
+      "loss": 0.6605,
+      "step": 8662
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.39212599567585316,
+      "learning_rate": 5.097637077780329e-06,
+      "loss": 0.6282,
+      "step": 8663
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3555820273827148,
+      "learning_rate": 5.09377481518952e-06,
+      "loss": 0.6159,
+      "step": 8664
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3919711693614353,
+      "learning_rate": 5.0899138027658136e-06,
+      "loss": 0.6559,
+      "step": 8665
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3974516638509639,
+      "learning_rate": 5.0860540408330286e-06,
+      "loss": 0.6522,
+      "step": 8666
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.44293834654284053,
+      "learning_rate": 5.0821955297148835e-06,
+      "loss": 0.6792,
+      "step": 8667
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.7734776962191106,
+      "learning_rate": 5.078338269734979e-06,
+      "loss": 6.5077,
+      "step": 8668
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.36096197646396194,
+      "learning_rate": 5.074482261216822e-06,
+      "loss": 0.6208,
+      "step": 8669
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38281520838757394,
+      "learning_rate": 5.070627504483813e-06,
+      "loss": 0.6405,
+      "step": 8670
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.35001422453648795,
+      "learning_rate": 5.066773999859244e-06,
+      "loss": 0.6403,
+      "step": 8671
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3884477628894712,
+      "learning_rate": 5.062921747666307e-06,
+      "loss": 0.637,
+      "step": 8672
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.40537391422129904,
+      "learning_rate": 5.059070748228088e-06,
+      "loss": 0.6815,
+      "step": 8673
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4255131125109314,
+      "learning_rate": 5.055221001867556e-06,
+      "loss": 0.6424,
+      "step": 8674
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6107152811037204,
+      "learning_rate": 5.05137250890759e-06,
+      "loss": 0.6539,
+      "step": 8675
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.44948258692867626,
+      "learning_rate": 5.047525269670963e-06,
+      "loss": 0.6076,
+      "step": 8676
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3836131566540395,
+      "learning_rate": 5.043679284480323e-06,
+      "loss": 0.6318,
+      "step": 8677
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.42962230561897996,
+      "learning_rate": 5.0398345536582495e-06,
+      "loss": 0.695,
+      "step": 8678
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3362979774573123,
+      "learning_rate": 5.03599107752718e-06,
+      "loss": 0.6392,
+      "step": 8679
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4537875414725047,
+      "learning_rate": 5.032148856409467e-06,
+      "loss": 0.6841,
+      "step": 8680
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.392309348695706,
+      "learning_rate": 5.028307890627351e-06,
+      "loss": 0.613,
+      "step": 8681
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3683262528097134,
+      "learning_rate": 5.024468180502979e-06,
+      "loss": 0.6183,
+      "step": 8682
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.35687073714348827,
+      "learning_rate": 5.020629726358368e-06,
+      "loss": 0.6088,
+      "step": 8683
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4113225467593377,
+      "learning_rate": 5.016792528515451e-06,
+      "loss": 0.6243,
+      "step": 8684
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38005959282716373,
+      "learning_rate": 5.012956587296049e-06,
+      "loss": 0.6622,
+      "step": 8685
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.39530438615660374,
+      "learning_rate": 5.009121903021881e-06,
+      "loss": 0.6421,
+      "step": 8686
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.396854221253168,
+      "learning_rate": 5.005288476014558e-06,
+      "loss": 0.6424,
+      "step": 8687
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.32757260751021483,
+      "learning_rate": 5.001456306595576e-06,
+      "loss": 0.6096,
+      "step": 8688
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.36377835993855334,
+      "learning_rate": 4.997625395086338e-06,
+      "loss": 0.6608,
+      "step": 8689
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4251453242754243,
+      "learning_rate": 4.993795741808143e-06,
+      "loss": 0.6453,
+      "step": 8690
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38251919202495144,
+      "learning_rate": 4.9899673470821765e-06,
+      "loss": 0.6141,
+      "step": 8691
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38926995398890274,
+      "learning_rate": 4.986140211229518e-06,
+      "loss": 0.6177,
+      "step": 8692
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3798084043613346,
+      "learning_rate": 4.982314334571156e-06,
+      "loss": 0.6103,
+      "step": 8693
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3810596343241818,
+      "learning_rate": 4.978489717427948e-06,
+      "loss": 0.6041,
+      "step": 8694
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3319306706547514,
+      "learning_rate": 4.974666360120666e-06,
+      "loss": 0.6054,
+      "step": 8695
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3739009126034507,
+      "learning_rate": 4.970844262969974e-06,
+      "loss": 0.6359,
+      "step": 8696
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4041915693602485,
+      "learning_rate": 4.967023426296413e-06,
+      "loss": 0.6683,
+      "step": 8697
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3169604529224674,
+      "learning_rate": 4.963203850420455e-06,
+      "loss": 0.6408,
+      "step": 8698
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38415857040002166,
+      "learning_rate": 4.959385535662422e-06,
+      "loss": 0.6425,
+      "step": 8699
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.42587841157477396,
+      "learning_rate": 4.955568482342561e-06,
+      "loss": 0.6539,
+      "step": 8700
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38858220371911173,
+      "learning_rate": 4.951752690781009e-06,
+      "loss": 0.6166,
+      "step": 8701
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.37164535527335807,
+      "learning_rate": 4.947938161297778e-06,
+      "loss": 0.6589,
+      "step": 8702
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.37120432560850625,
+      "learning_rate": 4.944124894212796e-06,
+      "loss": 0.6391,
+      "step": 8703
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3839767317296217,
+      "learning_rate": 4.9403128898458755e-06,
+      "loss": 0.6106,
+      "step": 8704
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3508767636537284,
+      "learning_rate": 4.936502148516724e-06,
+      "loss": 0.6345,
+      "step": 8705
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3747600460098119,
+      "learning_rate": 4.932692670544947e-06,
+      "loss": 0.637,
+      "step": 8706
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3991008830355653,
+      "learning_rate": 4.928884456250045e-06,
+      "loss": 0.6116,
+      "step": 8707
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.40408102431646375,
+      "learning_rate": 4.925077505951396e-06,
+      "loss": 0.6514,
+      "step": 8708
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3701787008463124,
+      "learning_rate": 4.921271819968292e-06,
+      "loss": 0.6472,
+      "step": 8709
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4059897510372293,
+      "learning_rate": 4.917467398619915e-06,
+      "loss": 0.686,
+      "step": 8710
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4050934570216399,
+      "learning_rate": 4.91366424222532e-06,
+      "loss": 0.6543,
+      "step": 8711
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.407899264954549,
+      "learning_rate": 4.909862351103492e-06,
+      "loss": 0.6593,
+      "step": 8712
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3667459254199296,
+      "learning_rate": 4.9060617255732875e-06,
+      "loss": 0.6508,
+      "step": 8713
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3579272223365348,
+      "learning_rate": 4.902262365953454e-06,
+      "loss": 0.6307,
+      "step": 8714
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4037603190216707,
+      "learning_rate": 4.898464272562642e-06,
+      "loss": 0.6334,
+      "step": 8715
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3577057912564387,
+      "learning_rate": 4.894667445719399e-06,
+      "loss": 0.6505,
+      "step": 8716
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38843004626706606,
+      "learning_rate": 4.890871885742148e-06,
+      "loss": 0.6371,
+      "step": 8717
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.42826751642972494,
+      "learning_rate": 4.887077592949225e-06,
+      "loss": 0.6944,
+      "step": 8718
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4253737908567468,
+      "learning_rate": 4.883284567658854e-06,
+      "loss": 0.6947,
+      "step": 8719
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3505584118480086,
+      "learning_rate": 4.8794928101891505e-06,
+      "loss": 0.6297,
+      "step": 8720
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38009953801920987,
+      "learning_rate": 4.875702320858127e-06,
+      "loss": 0.6564,
+      "step": 8721
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.42832739131702335,
+      "learning_rate": 4.871913099983681e-06,
+      "loss": 0.6925,
+      "step": 8722
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.45502589491197915,
+      "learning_rate": 4.868125147883611e-06,
+      "loss": 0.641,
+      "step": 8723
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.42014068958901407,
+      "learning_rate": 4.864338464875618e-06,
+      "loss": 0.645,
+      "step": 8724
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3590441378305698,
+      "learning_rate": 4.860553051277266e-06,
+      "loss": 0.6441,
+      "step": 8725
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.37864984155997305,
+      "learning_rate": 4.856768907406051e-06,
+      "loss": 0.675,
+      "step": 8726
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3838156236440292,
+      "learning_rate": 4.8529860335793456e-06,
+      "loss": 0.6373,
+      "step": 8727
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4469190487194006,
+      "learning_rate": 4.849204430114405e-06,
+      "loss": 0.6465,
+      "step": 8728
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.32325459355043873,
+      "learning_rate": 4.845424097328389e-06,
+      "loss": 0.6307,
+      "step": 8729
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3961843348905865,
+      "learning_rate": 4.841645035538358e-06,
+      "loss": 0.639,
+      "step": 8730
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38746486492771504,
+      "learning_rate": 4.837867245061245e-06,
+      "loss": 0.6666,
+      "step": 8731
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4440335693377254,
+      "learning_rate": 4.834090726213896e-06,
+      "loss": 0.6632,
+      "step": 8732
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3635202407567604,
+      "learning_rate": 4.8303154793130414e-06,
+      "loss": 0.6317,
+      "step": 8733
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.33475740083393224,
+      "learning_rate": 4.826541504675306e-06,
+      "loss": 0.6368,
+      "step": 8734
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3937321201254525,
+      "learning_rate": 4.822768802617217e-06,
+      "loss": 0.6205,
+      "step": 8735
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.34220780777351884,
+      "learning_rate": 4.818997373455174e-06,
+      "loss": 0.6373,
+      "step": 8736
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3801902180225607,
+      "learning_rate": 4.815227217505487e-06,
+      "loss": 0.6199,
+      "step": 8737
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3706089391335168,
+      "learning_rate": 4.811458335084353e-06,
+      "loss": 0.6189,
+      "step": 8738
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3748443905039026,
+      "learning_rate": 4.807690726507868e-06,
+      "loss": 0.6679,
+      "step": 8739
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.36949813141056154,
+      "learning_rate": 4.803924392092012e-06,
+      "loss": 0.6525,
+      "step": 8740
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.36808323136744564,
+      "learning_rate": 4.80015933215267e-06,
+      "loss": 0.6741,
+      "step": 8741
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3687205403880471,
+      "learning_rate": 4.7963955470056035e-06,
+      "loss": 0.6368,
+      "step": 8742
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3747429535223751,
+      "learning_rate": 4.792633036966481e-06,
+      "loss": 0.6391,
+      "step": 8743
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38719607551094093,
+      "learning_rate": 4.788871802350867e-06,
+      "loss": 0.6854,
+      "step": 8744
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3842875872786595,
+      "learning_rate": 4.785111843474197e-06,
+      "loss": 0.6483,
+      "step": 8745
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4095471132199678,
+      "learning_rate": 4.7813531606518156e-06,
+      "loss": 0.6358,
+      "step": 8746
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38359015383335615,
+      "learning_rate": 4.7775957541989785e-06,
+      "loss": 0.6629,
+      "step": 8747
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.37126437499496023,
+      "learning_rate": 4.773839624430796e-06,
+      "loss": 0.6458,
+      "step": 8748
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3917370724378442,
+      "learning_rate": 4.770084771662294e-06,
+      "loss": 0.65,
+      "step": 8749
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3408022071742591,
+      "learning_rate": 4.766331196208396e-06,
+      "loss": 0.6271,
+      "step": 8750
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.41862896805460376,
+      "learning_rate": 4.762578898383896e-06,
+      "loss": 0.6499,
+      "step": 8751
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38438216523404295,
+      "learning_rate": 4.758827878503502e-06,
+      "loss": 0.6696,
+      "step": 8752
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3924064414957232,
+      "learning_rate": 4.755078136881808e-06,
+      "loss": 0.6554,
+      "step": 8753
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.37843145110780185,
+      "learning_rate": 4.751329673833298e-06,
+      "loss": 0.5784,
+      "step": 8754
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3894370133005494,
+      "learning_rate": 4.74758248967236e-06,
+      "loss": 0.6881,
+      "step": 8755
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38592835998857133,
+      "learning_rate": 4.7438365847132525e-06,
+      "loss": 0.6732,
+      "step": 8756
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.41328262298163854,
+      "learning_rate": 4.740091959270146e-06,
+      "loss": 0.638,
+      "step": 8757
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3535434708934818,
+      "learning_rate": 4.7363486136571e-06,
+      "loss": 0.6226,
+      "step": 8758
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3509961474121817,
+      "learning_rate": 4.732606548188059e-06,
+      "loss": 0.5934,
+      "step": 8759
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.42941238402192017,
+      "learning_rate": 4.728865763176862e-06,
+      "loss": 0.6682,
+      "step": 8760
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.40583926571055134,
+      "learning_rate": 4.725126258937262e-06,
+      "loss": 0.6017,
+      "step": 8761
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4037262135389521,
+      "learning_rate": 4.721388035782872e-06,
+      "loss": 0.6911,
+      "step": 8762
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.39999203029630376,
+      "learning_rate": 4.717651094027214e-06,
+      "loss": 0.6648,
+      "step": 8763
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.36017473061969574,
+      "learning_rate": 4.713915433983707e-06,
+      "loss": 0.633,
+      "step": 8764
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.43061972068632975,
+      "learning_rate": 4.710181055965646e-06,
+      "loss": 0.6975,
+      "step": 8765
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.36968511291107214,
+      "learning_rate": 4.706447960286236e-06,
+      "loss": 0.6514,
+      "step": 8766
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.42986994256536093,
+      "learning_rate": 4.702716147258564e-06,
+      "loss": 0.6456,
+      "step": 8767
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.39229066403429136,
+      "learning_rate": 4.698985617195615e-06,
+      "loss": 0.6568,
+      "step": 8768
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.36988703565051345,
+      "learning_rate": 4.695256370410262e-06,
+      "loss": 0.5986,
+      "step": 8769
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.38472426152128775,
+      "learning_rate": 4.691528407215282e-06,
+      "loss": 0.6363,
+      "step": 8770
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3574377043575622,
+      "learning_rate": 4.687801727923318e-06,
+      "loss": 0.6095,
+      "step": 8771
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3538517814691135,
+      "learning_rate": 4.684076332846932e-06,
+      "loss": 0.6405,
+      "step": 8772
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3898250711300614,
+      "learning_rate": 4.680352222298568e-06,
+      "loss": 0.6152,
+      "step": 8773
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.39372879099352376,
+      "learning_rate": 4.676629396590562e-06,
+      "loss": 0.6372,
+      "step": 8774
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.363802151205609,
+      "learning_rate": 4.672907856035147e-06,
+      "loss": 0.6335,
+      "step": 8775
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3218351324122911,
+      "learning_rate": 4.669187600944433e-06,
+      "loss": 0.6066,
+      "step": 8776
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.39535932155671594,
+      "learning_rate": 4.6654686316304406e-06,
+      "loss": 0.6175,
+      "step": 8777
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.37745604467160326,
+      "learning_rate": 4.661750948405082e-06,
+      "loss": 0.6803,
+      "step": 8778
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4042627407859719,
+      "learning_rate": 4.658034551580139e-06,
+      "loss": 0.68,
+      "step": 8779
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.40627129707096765,
+      "learning_rate": 4.654319441467306e-06,
+      "loss": 0.6324,
+      "step": 8780
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3568129037841283,
+      "learning_rate": 4.650605618378179e-06,
+      "loss": 0.6264,
+      "step": 8781
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4430620526964161,
+      "learning_rate": 4.646893082624217e-06,
+      "loss": 0.7119,
+      "step": 8782
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37149071261566646,
+      "learning_rate": 4.64318183451679e-06,
+      "loss": 0.6109,
+      "step": 8783
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.42845742243800017,
+      "learning_rate": 4.6394718743671605e-06,
+      "loss": 0.6461,
+      "step": 8784
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.36823967609032365,
+      "learning_rate": 4.6357632024864695e-06,
+      "loss": 0.664,
+      "step": 8785
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3429048523803549,
+      "learning_rate": 4.632055819185764e-06,
+      "loss": 0.6117,
+      "step": 8786
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.38009004524865836,
+      "learning_rate": 4.628349724775976e-06,
+      "loss": 0.5928,
+      "step": 8787
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3581536845987714,
+      "learning_rate": 4.624644919567934e-06,
+      "loss": 0.6833,
+      "step": 8788
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4181569029951705,
+      "learning_rate": 4.620941403872356e-06,
+      "loss": 0.7025,
+      "step": 8789
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6308460334637238,
+      "learning_rate": 4.617239177999846e-06,
+      "loss": 0.6609,
+      "step": 8790
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.42059221789858253,
+      "learning_rate": 4.6135382422609086e-06,
+      "loss": 0.6521,
+      "step": 8791
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3597896958882799,
+      "learning_rate": 4.609838596965943e-06,
+      "loss": 0.6282,
+      "step": 8792
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37226380285947464,
+      "learning_rate": 4.6061402424252215e-06,
+      "loss": 0.6298,
+      "step": 8793
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.39331893820850583,
+      "learning_rate": 4.602443178948919e-06,
+      "loss": 0.6503,
+      "step": 8794
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.39783263171191313,
+      "learning_rate": 4.598747406847126e-06,
+      "loss": 0.676,
+      "step": 8795
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37230850766103074,
+      "learning_rate": 4.595052926429779e-06,
+      "loss": 0.6238,
+      "step": 8796
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.36855398357096775,
+      "learning_rate": 4.5913597380067424e-06,
+      "loss": 0.5898,
+      "step": 8797
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.417373378052024,
+      "learning_rate": 4.58766784188776e-06,
+      "loss": 0.6138,
+      "step": 8798
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3551984294849844,
+      "learning_rate": 4.583977238382455e-06,
+      "loss": 0.6092,
+      "step": 8799
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.41837884934398245,
+      "learning_rate": 4.580287927800362e-06,
+      "loss": 0.6871,
+      "step": 8800
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3700490694596553,
+      "learning_rate": 4.576599910450899e-06,
+      "loss": 0.6251,
+      "step": 8801
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.39356118941873697,
+      "learning_rate": 4.572913186643373e-06,
+      "loss": 0.7027,
+      "step": 8802
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.367109196016017,
+      "learning_rate": 4.569227756686989e-06,
+      "loss": 0.6454,
+      "step": 8803
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.42341224728387056,
+      "learning_rate": 4.565543620890842e-06,
+      "loss": 0.6627,
+      "step": 8804
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37277930389643377,
+      "learning_rate": 4.561860779563905e-06,
+      "loss": 0.6639,
+      "step": 8805
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.38236368348512095,
+      "learning_rate": 4.558179233015061e-06,
+      "loss": 0.6408,
+      "step": 8806
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37751701923179165,
+      "learning_rate": 4.554498981553082e-06,
+      "loss": 0.6118,
+      "step": 8807
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.41584808545733787,
+      "learning_rate": 4.550820025486607e-06,
+      "loss": 0.6365,
+      "step": 8808
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.399020679638242,
+      "learning_rate": 4.547142365124211e-06,
+      "loss": 0.6534,
+      "step": 8809
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3810601849017077,
+      "learning_rate": 4.5434660007743144e-06,
+      "loss": 0.6219,
+      "step": 8810
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.39128917786612916,
+      "learning_rate": 4.53979093274526e-06,
+      "loss": 0.6059,
+      "step": 8811
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.33952665074960764,
+      "learning_rate": 4.536117161345275e-06,
+      "loss": 0.6199,
+      "step": 8812
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3781759897263874,
+      "learning_rate": 4.532444686882462e-06,
+      "loss": 0.6256,
+      "step": 8813
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3651062866330933,
+      "learning_rate": 4.528773509664832e-06,
+      "loss": 0.6391,
+      "step": 8814
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3751777963085341,
+      "learning_rate": 4.5251036300002845e-06,
+      "loss": 0.6426,
+      "step": 8815
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4026621231353928,
+      "learning_rate": 4.521435048196605e-06,
+      "loss": 0.6367,
+      "step": 8816
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4070318583229455,
+      "learning_rate": 4.517767764561476e-06,
+      "loss": 0.6508,
+      "step": 8817
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3532973004613104,
+      "learning_rate": 4.514101779402471e-06,
+      "loss": 0.6137,
+      "step": 8818
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.42919286028837056,
+      "learning_rate": 4.510437093027045e-06,
+      "loss": 0.6578,
+      "step": 8819
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.38860914535822566,
+      "learning_rate": 4.506773705742551e-06,
+      "loss": 0.6367,
+      "step": 8820
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.36755920743542836,
+      "learning_rate": 4.50311161785624e-06,
+      "loss": 0.6504,
+      "step": 8821
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37541013312664256,
+      "learning_rate": 4.499450829675232e-06,
+      "loss": 0.6659,
+      "step": 8822
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.38413778654710945,
+      "learning_rate": 4.495791341506575e-06,
+      "loss": 0.6469,
+      "step": 8823
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.39757180257855057,
+      "learning_rate": 4.4921331536571655e-06,
+      "loss": 0.6462,
+      "step": 8824
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4453064322770506,
+      "learning_rate": 4.4884762664338236e-06,
+      "loss": 0.6679,
+      "step": 8825
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37261352833416744,
+      "learning_rate": 4.48482068014324e-06,
+      "loss": 0.6518,
+      "step": 8826
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3604749902469332,
+      "learning_rate": 4.481166395092016e-06,
+      "loss": 0.622,
+      "step": 8827
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.42791149402315404,
+      "learning_rate": 4.477513411586621e-06,
+      "loss": 0.6418,
+      "step": 8828
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4048727626549714,
+      "learning_rate": 4.473861729933426e-06,
+      "loss": 0.6439,
+      "step": 8829
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.42305731433966826,
+      "learning_rate": 4.470211350438698e-06,
+      "loss": 0.6745,
+      "step": 8830
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.41989883115967347,
+      "learning_rate": 4.4665622734085905e-06,
+      "loss": 0.6127,
+      "step": 8831
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37919567173650615,
+      "learning_rate": 4.462914499149149e-06,
+      "loss": 0.6402,
+      "step": 8832
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.41504168384985307,
+      "learning_rate": 4.459268027966299e-06,
+      "loss": 0.6673,
+      "step": 8833
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4087887273816279,
+      "learning_rate": 4.455622860165872e-06,
+      "loss": 0.6489,
+      "step": 8834
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4085546938538315,
+      "learning_rate": 4.451978996053588e-06,
+      "loss": 0.6396,
+      "step": 8835
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4062773661805865,
+      "learning_rate": 4.448336435935037e-06,
+      "loss": 0.6273,
+      "step": 8836
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3912838174488429,
+      "learning_rate": 4.444695180115734e-06,
+      "loss": 0.6544,
+      "step": 8837
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.40610386406649696,
+      "learning_rate": 4.4410552289010655e-06,
+      "loss": 0.6137,
+      "step": 8838
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.34184796815256985,
+      "learning_rate": 4.437416582596299e-06,
+      "loss": 0.623,
+      "step": 8839
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.35736987826740224,
+      "learning_rate": 4.433779241506608e-06,
+      "loss": 0.6454,
+      "step": 8840
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3593547350801666,
+      "learning_rate": 4.430143205937058e-06,
+      "loss": 0.6347,
+      "step": 8841
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3514792334189806,
+      "learning_rate": 4.4265084761925904e-06,
+      "loss": 0.608,
+      "step": 8842
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37378049039852396,
+      "learning_rate": 4.422875052578046e-06,
+      "loss": 0.6209,
+      "step": 8843
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.469396164512376,
+      "learning_rate": 4.41924293539816e-06,
+      "loss": 6.6306,
+      "step": 8844
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.39168159472224545,
+      "learning_rate": 4.415612124957551e-06,
+      "loss": 0.6469,
+      "step": 8845
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.38287387746631585,
+      "learning_rate": 4.41198262156074e-06,
+      "loss": 0.6227,
+      "step": 8846
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3457903050443944,
+      "learning_rate": 4.408354425512114e-06,
+      "loss": 0.6113,
+      "step": 8847
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3917766165220241,
+      "learning_rate": 4.404727537115971e-06,
+      "loss": 0.6636,
+      "step": 8848
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3901929715613469,
+      "learning_rate": 4.401101956676503e-06,
+      "loss": 0.6066,
+      "step": 8849
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4332975410552945,
+      "learning_rate": 4.397477684497764e-06,
+      "loss": 0.661,
+      "step": 8850
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3695243135302884,
+      "learning_rate": 4.393854720883732e-06,
+      "loss": 0.6147,
+      "step": 8851
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3724721820368473,
+      "learning_rate": 4.390233066138266e-06,
+      "loss": 0.6635,
+      "step": 8852
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3889639542992012,
+      "learning_rate": 4.386612720565091e-06,
+      "loss": 0.6672,
+      "step": 8853
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.39347074699493073,
+      "learning_rate": 4.382993684467853e-06,
+      "loss": 0.6772,
+      "step": 8854
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.41827725426035073,
+      "learning_rate": 4.379375958150081e-06,
+      "loss": 0.6518,
+      "step": 8855
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.36503237752398143,
+      "learning_rate": 4.37575954191517e-06,
+      "loss": 0.647,
+      "step": 8856
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3827289354922543,
+      "learning_rate": 4.372144436066448e-06,
+      "loss": 0.6628,
+      "step": 8857
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.40895658928089734,
+      "learning_rate": 4.3685306409070935e-06,
+      "loss": 0.6117,
+      "step": 8858
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4013415405473945,
+      "learning_rate": 4.364918156740196e-06,
+      "loss": 0.6366,
+      "step": 8859
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37250858713585594,
+      "learning_rate": 4.361306983868729e-06,
+      "loss": 0.6422,
+      "step": 8860
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.36728594271221165,
+      "learning_rate": 4.357697122595566e-06,
+      "loss": 0.6482,
+      "step": 8861
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3740671138039441,
+      "learning_rate": 4.354088573223448e-06,
+      "loss": 0.6424,
+      "step": 8862
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3613467816764837,
+      "learning_rate": 4.350481336055026e-06,
+      "loss": 0.6095,
+      "step": 8863
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3786606243157344,
+      "learning_rate": 4.346875411392835e-06,
+      "loss": 0.6099,
+      "step": 8864
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3719123863811093,
+      "learning_rate": 4.343270799539297e-06,
+      "loss": 0.6383,
+      "step": 8865
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.392468767079211,
+      "learning_rate": 4.339667500796736e-06,
+      "loss": 0.6306,
+      "step": 8866
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 3.3974397099629097,
+      "learning_rate": 4.336065515467342e-06,
+      "loss": 6.5476,
+      "step": 8867
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3812477205167007,
+      "learning_rate": 4.332464843853217e-06,
+      "loss": 0.6442,
+      "step": 8868
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.40838515820723154,
+      "learning_rate": 4.328865486256348e-06,
+      "loss": 0.6425,
+      "step": 8869
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4130348705629128,
+      "learning_rate": 4.325267442978596e-06,
+      "loss": 0.6661,
+      "step": 8870
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.36861884946271084,
+      "learning_rate": 4.321670714321738e-06,
+      "loss": 0.6195,
+      "step": 8871
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.39179226967161773,
+      "learning_rate": 4.3180753005874275e-06,
+      "loss": 0.6501,
+      "step": 8872
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.41455719046349554,
+      "learning_rate": 4.314481202077196e-06,
+      "loss": 0.6655,
+      "step": 8873
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.38869741788043394,
+      "learning_rate": 4.310888419092485e-06,
+      "loss": 0.6478,
+      "step": 8874
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.39841909085050475,
+      "learning_rate": 4.307296951934619e-06,
+      "loss": 0.6356,
+      "step": 8875
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.36617035262151265,
+      "learning_rate": 4.303706800904799e-06,
+      "loss": 0.6213,
+      "step": 8876
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3358231199267879,
+      "learning_rate": 4.3001179663041355e-06,
+      "loss": 0.6744,
+      "step": 8877
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4186577246693441,
+      "learning_rate": 4.296530448433615e-06,
+      "loss": 0.6347,
+      "step": 8878
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3834086184242422,
+      "learning_rate": 4.292944247594124e-06,
+      "loss": 0.6829,
+      "step": 8879
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.35367441674326555,
+      "learning_rate": 4.289359364086434e-06,
+      "loss": 0.6192,
+      "step": 8880
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.43313876096509046,
+      "learning_rate": 4.2857757982111935e-06,
+      "loss": 0.6976,
+      "step": 8881
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3591236086958627,
+      "learning_rate": 4.282193550268961e-06,
+      "loss": 0.6023,
+      "step": 8882
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3576361589853354,
+      "learning_rate": 4.27861262056017e-06,
+      "loss": 0.6292,
+      "step": 8883
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.36824253261971673,
+      "learning_rate": 4.275033009385155e-06,
+      "loss": 0.6163,
+      "step": 8884
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3296844825391675,
+      "learning_rate": 4.2714547170441275e-06,
+      "loss": 0.6179,
+      "step": 8885
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3925486403581314,
+      "learning_rate": 4.267877743837205e-06,
+      "loss": 0.6713,
+      "step": 8886
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4344201744551478,
+      "learning_rate": 4.264302090064372e-06,
+      "loss": 0.6583,
+      "step": 8887
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4132019672488705,
+      "learning_rate": 4.260727756025518e-06,
+      "loss": 0.7129,
+      "step": 8888
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4123870304735449,
+      "learning_rate": 4.257154742020424e-06,
+      "loss": 0.6487,
+      "step": 8889
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.5455736309407215,
+      "learning_rate": 4.2535830483487435e-06,
+      "loss": 6.6975,
+      "step": 8890
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3975467602320378,
+      "learning_rate": 4.250012675310031e-06,
+      "loss": 0.658,
+      "step": 8891
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.41284392368523987,
+      "learning_rate": 4.2464436232037445e-06,
+      "loss": 0.648,
+      "step": 8892
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4075487788343105,
+      "learning_rate": 4.242875892329201e-06,
+      "loss": 0.6733,
+      "step": 8893
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.40578715451204594,
+      "learning_rate": 4.2393094829856275e-06,
+      "loss": 0.6669,
+      "step": 8894
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.38795260401363535,
+      "learning_rate": 4.235744395472136e-06,
+      "loss": 0.6329,
+      "step": 8895
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4021796307313543,
+      "learning_rate": 4.2321806300877214e-06,
+      "loss": 0.6269,
+      "step": 8896
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4202383070141119,
+      "learning_rate": 4.2286181871312745e-06,
+      "loss": 0.661,
+      "step": 8897
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.35817887541037136,
+      "learning_rate": 4.225057066901572e-06,
+      "loss": 0.6251,
+      "step": 8898
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.39773407473831396,
+      "learning_rate": 4.221497269697281e-06,
+      "loss": 0.6097,
+      "step": 8899
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.202793921608466,
+      "learning_rate": 4.217938795816965e-06,
+      "loss": 6.5771,
+      "step": 8900
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4491622446421977,
+      "learning_rate": 4.214381645559058e-06,
+      "loss": 0.6741,
+      "step": 8901
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3893147850669898,
+      "learning_rate": 4.210825819221897e-06,
+      "loss": 0.6364,
+      "step": 8902
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3694744040661987,
+      "learning_rate": 4.207271317103712e-06,
+      "loss": 0.6392,
+      "step": 8903
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3940518419383747,
+      "learning_rate": 4.203718139502601e-06,
+      "loss": 0.6782,
+      "step": 8904
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3216833560633937,
+      "learning_rate": 4.200166286716571e-06,
+      "loss": 0.6226,
+      "step": 8905
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3723818202062178,
+      "learning_rate": 4.19661575904352e-06,
+      "loss": 0.6454,
+      "step": 8906
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.35296400825681695,
+      "learning_rate": 4.193066556781218e-06,
+      "loss": 0.6287,
+      "step": 8907
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.397442504605861,
+      "learning_rate": 4.189518680227331e-06,
+      "loss": 0.6372,
+      "step": 8908
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3683356487379272,
+      "learning_rate": 4.185972129679426e-06,
+      "loss": 0.6454,
+      "step": 8909
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.33326729797356786,
+      "learning_rate": 4.182426905434933e-06,
+      "loss": 0.6132,
+      "step": 8910
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.34315807508919166,
+      "learning_rate": 4.178883007791192e-06,
+      "loss": 0.662,
+      "step": 8911
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.42165604615898,
+      "learning_rate": 4.175340437045428e-06,
+      "loss": 0.6751,
+      "step": 8912
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.39140027841072866,
+      "learning_rate": 4.17179919349475e-06,
+      "loss": 0.6495,
+      "step": 8913
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.36058924685414273,
+      "learning_rate": 4.168259277436162e-06,
+      "loss": 0.6288,
+      "step": 8914
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.34398387893827514,
+      "learning_rate": 4.164720689166544e-06,
+      "loss": 0.6083,
+      "step": 8915
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3784406429124795,
+      "learning_rate": 4.161183428982678e-06,
+      "loss": 0.6035,
+      "step": 8916
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.38466342860515557,
+      "learning_rate": 4.157647497181227e-06,
+      "loss": 0.6486,
+      "step": 8917
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3696932141454656,
+      "learning_rate": 4.154112894058755e-06,
+      "loss": 0.5962,
+      "step": 8918
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.393997477203331,
+      "learning_rate": 4.150579619911687e-06,
+      "loss": 0.6443,
+      "step": 8919
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3435798799627419,
+      "learning_rate": 4.1470476750363755e-06,
+      "loss": 0.6367,
+      "step": 8920
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3964980352146365,
+      "learning_rate": 4.1435170597290255e-06,
+      "loss": 0.6671,
+      "step": 8921
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3354232843495785,
+      "learning_rate": 4.1399877742857475e-06,
+      "loss": 0.5941,
+      "step": 8922
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.35545807251717615,
+      "learning_rate": 4.13645981900255e-06,
+      "loss": 0.6343,
+      "step": 8923
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3973904690254908,
+      "learning_rate": 4.132933194175299e-06,
+      "loss": 0.683,
+      "step": 8924
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3778616731531157,
+      "learning_rate": 4.129407900099782e-06,
+      "loss": 0.6262,
+      "step": 8925
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.40734686243356316,
+      "learning_rate": 4.1258839370716575e-06,
+      "loss": 0.6501,
+      "step": 8926
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3546959523229586,
+      "learning_rate": 4.1223613053864734e-06,
+      "loss": 0.6264,
+      "step": 8927
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4007277314662672,
+      "learning_rate": 4.1188400053396725e-06,
+      "loss": 0.6505,
+      "step": 8928
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3903016332077508,
+      "learning_rate": 4.115320037226586e-06,
+      "loss": 0.6664,
+      "step": 8929
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.41279209846067033,
+      "learning_rate": 4.111801401342421e-06,
+      "loss": 0.6398,
+      "step": 8930
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3944773925708867,
+      "learning_rate": 4.1082840979822824e-06,
+      "loss": 0.6351,
+      "step": 8931
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3722352471678451,
+      "learning_rate": 4.10476812744117e-06,
+      "loss": 0.6337,
+      "step": 8932
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.39917815615213814,
+      "learning_rate": 4.101253490013948e-06,
+      "loss": 0.632,
+      "step": 8933
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3293419626924151,
+      "learning_rate": 4.0977401859954045e-06,
+      "loss": 0.6292,
+      "step": 8934
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4307332442605112,
+      "learning_rate": 4.094228215680183e-06,
+      "loss": 0.6567,
+      "step": 8935
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3730171642474426,
+      "learning_rate": 4.090717579362828e-06,
+      "loss": 0.6322,
+      "step": 8936
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4290561454870628,
+      "learning_rate": 4.0872082773377864e-06,
+      "loss": 0.6421,
+      "step": 8937
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3997343267741867,
+      "learning_rate": 4.083700309899359e-06,
+      "loss": 0.6574,
+      "step": 8938
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3611993102666683,
+      "learning_rate": 4.080193677341762e-06,
+      "loss": 0.6321,
+      "step": 8939
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.41741854579093857,
+      "learning_rate": 4.076688379959104e-06,
+      "loss": 0.658,
+      "step": 8940
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3553068301744984,
+      "learning_rate": 4.073184418045357e-06,
+      "loss": 0.6408,
+      "step": 8941
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.43827728119840714,
+      "learning_rate": 4.069681791894397e-06,
+      "loss": 0.681,
+      "step": 8942
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3981035725069993,
+      "learning_rate": 4.066180501799994e-06,
+      "loss": 0.6445,
+      "step": 8943
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.38432127880944,
+      "learning_rate": 4.062680548055782e-06,
+      "loss": 0.6585,
+      "step": 8944
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.39249539831305696,
+      "learning_rate": 4.059181930955305e-06,
+      "loss": 0.6382,
+      "step": 8945
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.44936266348479115,
+      "learning_rate": 4.055684650791987e-06,
+      "loss": 0.6921,
+      "step": 8946
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.41701452323889293,
+      "learning_rate": 4.0521887078591436e-06,
+      "loss": 0.6418,
+      "step": 8947
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3671415489743687,
+      "learning_rate": 4.04869410244997e-06,
+      "loss": 0.6219,
+      "step": 8948
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3895287790961018,
+      "learning_rate": 4.045200834857566e-06,
+      "loss": 0.6496,
+      "step": 8949
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.39536083341264455,
+      "learning_rate": 4.0417089053748945e-06,
+      "loss": 0.6415,
+      "step": 8950
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3501321535107621,
+      "learning_rate": 4.0382183142948216e-06,
+      "loss": 0.6091,
+      "step": 8951
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.31512516252394995,
+      "learning_rate": 4.03472906191011e-06,
+      "loss": 0.6378,
+      "step": 8952
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3403061465005393,
+      "learning_rate": 4.0312411485133784e-06,
+      "loss": 0.6182,
+      "step": 8953
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.38466970040713894,
+      "learning_rate": 4.027754574397176e-06,
+      "loss": 0.6503,
+      "step": 8954
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.37390171058859856,
+      "learning_rate": 4.024269339853904e-06,
+      "loss": 0.6406,
+      "step": 8955
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3894801359472867,
+      "learning_rate": 4.020785445175868e-06,
+      "loss": 0.6581,
+      "step": 8956
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.39031639319947814,
+      "learning_rate": 4.017302890655266e-06,
+      "loss": 0.6601,
+      "step": 8957
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3650793591383457,
+      "learning_rate": 4.0138216765841595e-06,
+      "loss": 0.6463,
+      "step": 8958
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4059154160658839,
+      "learning_rate": 4.010341803254525e-06,
+      "loss": 0.6388,
+      "step": 8959
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.37653939541624787,
+      "learning_rate": 4.0068632709582125e-06,
+      "loss": 0.6141,
+      "step": 8960
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4392170126788798,
+      "learning_rate": 4.003386079986964e-06,
+      "loss": 0.6611,
+      "step": 8961
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3639152673573287,
+      "learning_rate": 3.999910230632405e-06,
+      "loss": 0.629,
+      "step": 8962
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.39062855134642616,
+      "learning_rate": 3.996435723186056e-06,
+      "loss": 0.677,
+      "step": 8963
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3720295660976755,
+      "learning_rate": 3.9929625579393106e-06,
+      "loss": 0.6486,
+      "step": 8964
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.34959865316549676,
+      "learning_rate": 3.9894907351834636e-06,
+      "loss": 0.6101,
+      "step": 8965
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4640915725714556,
+      "learning_rate": 3.986020255209697e-06,
+      "loss": 0.6886,
+      "step": 8966
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.343733499206238,
+      "learning_rate": 3.982551118309063e-06,
+      "loss": 0.6245,
+      "step": 8967
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3573984370968992,
+      "learning_rate": 3.979083324772533e-06,
+      "loss": 0.6332,
+      "step": 8968
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.40742523903635675,
+      "learning_rate": 3.975616874890931e-06,
+      "loss": 0.663,
+      "step": 8969
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4194198981732673,
+      "learning_rate": 3.972151768954988e-06,
+      "loss": 0.6371,
+      "step": 8970
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.44758281699162333,
+      "learning_rate": 3.968688007255325e-06,
+      "loss": 0.6407,
+      "step": 8971
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4064964823795335,
+      "learning_rate": 3.965225590082433e-06,
+      "loss": 0.6757,
+      "step": 8972
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.39306975674386185,
+      "learning_rate": 3.9617645177267075e-06,
+      "loss": 0.6385,
+      "step": 8973
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.37016113579057824,
+      "learning_rate": 3.95830479047842e-06,
+      "loss": 0.6638,
+      "step": 8974
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.37767716413894414,
+      "learning_rate": 3.954846408627737e-06,
+      "loss": 0.6587,
+      "step": 8975
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.44737942472983283,
+      "learning_rate": 3.951389372464707e-06,
+      "loss": 0.6078,
+      "step": 8976
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.40206969363033834,
+      "learning_rate": 3.947933682279277e-06,
+      "loss": 0.6533,
+      "step": 8977
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.41649034347735897,
+      "learning_rate": 3.944479338361253e-06,
+      "loss": 0.6584,
+      "step": 8978
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.41058567808352336,
+      "learning_rate": 3.941026341000361e-06,
+      "loss": 0.6445,
+      "step": 8979
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.41794591764994826,
+      "learning_rate": 3.937574690486199e-06,
+      "loss": 0.6715,
+      "step": 8980
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.39532620647436606,
+      "learning_rate": 3.934124387108238e-06,
+      "loss": 0.6467,
+      "step": 8981
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3807369841922678,
+      "learning_rate": 3.930675431155868e-06,
+      "loss": 0.6757,
+      "step": 8982
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4218110822239347,
+      "learning_rate": 3.9272278229183495e-06,
+      "loss": 0.6362,
+      "step": 8983
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.41767947759391005,
+      "learning_rate": 3.923781562684816e-06,
+      "loss": 0.6434,
+      "step": 8984
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3648756996467117,
+      "learning_rate": 3.9203366507443075e-06,
+      "loss": 0.6626,
+      "step": 8985
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4322691977163243,
+      "learning_rate": 3.9168930873857495e-06,
+      "loss": 0.6553,
+      "step": 8986
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.393986729288283,
+      "learning_rate": 3.913450872897939e-06,
+      "loss": 0.6364,
+      "step": 8987
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.34124541552168286,
+      "learning_rate": 3.910010007569578e-06,
+      "loss": 0.6432,
+      "step": 8988
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4297140967552062,
+      "learning_rate": 3.906570491689246e-06,
+      "loss": 0.6606,
+      "step": 8989
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.36712440331247675,
+      "learning_rate": 3.903132325545409e-06,
+      "loss": 0.6184,
+      "step": 8990
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.41973476967618445,
+      "learning_rate": 3.899695509426431e-06,
+      "loss": 0.6693,
+      "step": 8991
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.40891766298396837,
+      "learning_rate": 3.896260043620539e-06,
+      "loss": 0.6813,
+      "step": 8992
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.39868284719002434,
+      "learning_rate": 3.8928259284158706e-06,
+      "loss": 0.6498,
+      "step": 8993
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4095935108598303,
+      "learning_rate": 3.8893931641004435e-06,
+      "loss": 0.6725,
+      "step": 8994
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.34725704091911347,
+      "learning_rate": 3.885961750962144e-06,
+      "loss": 0.6216,
+      "step": 8995
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.37719175454761467,
+      "learning_rate": 3.882531689288778e-06,
+      "loss": 0.6401,
+      "step": 8996
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3556317176687369,
+      "learning_rate": 3.8791029793680195e-06,
+      "loss": 0.632,
+      "step": 8997
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3624844744612656,
+      "learning_rate": 3.8756756214874204e-06,
+      "loss": 0.6207,
+      "step": 8998
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3942438595323976,
+      "learning_rate": 3.8722496159344315e-06,
+      "loss": 0.6761,
+      "step": 8999
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3585038490457709,
+      "learning_rate": 3.868824962996396e-06,
+      "loss": 0.6187,
+      "step": 9000
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.378902667403775,
+      "learning_rate": 3.865401662960524e-06,
+      "loss": 0.6419,
+      "step": 9001
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3710287085305736,
+      "learning_rate": 3.86197971611393e-06,
+      "loss": 0.6558,
+      "step": 9002
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3670663279769937,
+      "learning_rate": 3.8585591227436055e-06,
+      "loss": 0.6075,
+      "step": 9003
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4363525417274136,
+      "learning_rate": 3.855139883136436e-06,
+      "loss": 0.6661,
+      "step": 9004
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.373206768750203,
+      "learning_rate": 3.851721997579185e-06,
+      "loss": 0.6489,
+      "step": 9005
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3743985298636239,
+      "learning_rate": 3.848305466358513e-06,
+      "loss": 0.6233,
+      "step": 9006
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37907266782805615,
+      "learning_rate": 3.84489028976095e-06,
+      "loss": 0.6694,
+      "step": 9007
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.35689738645051017,
+      "learning_rate": 3.841476468072926e-06,
+      "loss": 0.6477,
+      "step": 9008
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4021550838174854,
+      "learning_rate": 3.838064001580759e-06,
+      "loss": 0.6617,
+      "step": 9009
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3358612600030957,
+      "learning_rate": 3.834652890570645e-06,
+      "loss": 0.6114,
+      "step": 9010
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.35288665972724625,
+      "learning_rate": 3.831243135328675e-06,
+      "loss": 0.6318,
+      "step": 9011
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.41089090403694967,
+      "learning_rate": 3.827834736140812e-06,
+      "loss": 0.6153,
+      "step": 9012
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.34726695670508234,
+      "learning_rate": 3.824427693292918e-06,
+      "loss": 0.6199,
+      "step": 9013
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3670891840316988,
+      "learning_rate": 3.821022007070743e-06,
+      "loss": 0.6369,
+      "step": 9014
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39019256303573835,
+      "learning_rate": 3.817617677759908e-06,
+      "loss": 0.6503,
+      "step": 9015
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4273042905964275,
+      "learning_rate": 3.814214705645931e-06,
+      "loss": 0.6571,
+      "step": 9016
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4220037239048434,
+      "learning_rate": 3.810813091014227e-06,
+      "loss": 0.6562,
+      "step": 9017
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4015399041049925,
+      "learning_rate": 3.807412834150075e-06,
+      "loss": 0.6598,
+      "step": 9018
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.403741853225389,
+      "learning_rate": 3.8040139353386507e-06,
+      "loss": 0.6884,
+      "step": 9019
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37691667171293336,
+      "learning_rate": 3.8006163948650244e-06,
+      "loss": 0.6633,
+      "step": 9020
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4072068010511335,
+      "learning_rate": 3.797220213014132e-06,
+      "loss": 0.6232,
+      "step": 9021
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.33065934154322346,
+      "learning_rate": 3.793825390070811e-06,
+      "loss": 0.6187,
+      "step": 9022
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.270159402449618,
+      "learning_rate": 3.7904319263197843e-06,
+      "loss": 6.5587,
+      "step": 9023
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.35809695626186894,
+      "learning_rate": 3.787039822045655e-06,
+      "loss": 0.6416,
+      "step": 9024
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3920428306330814,
+      "learning_rate": 3.7836490775329206e-06,
+      "loss": 0.6554,
+      "step": 9025
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39513996183354216,
+      "learning_rate": 3.7802596930659486e-06,
+      "loss": 0.6254,
+      "step": 9026
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.384210087811699,
+      "learning_rate": 3.7768716689290053e-06,
+      "loss": 0.6389,
+      "step": 9027
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4020842352326136,
+      "learning_rate": 3.77348500540625e-06,
+      "loss": 0.5987,
+      "step": 9028
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.38613960467606984,
+      "learning_rate": 3.7700997027816997e-06,
+      "loss": 0.6508,
+      "step": 9029
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4187152773889158,
+      "learning_rate": 3.7667157613392924e-06,
+      "loss": 0.6078,
+      "step": 9030
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.396681704907079,
+      "learning_rate": 3.763333181362836e-06,
+      "loss": 0.6249,
+      "step": 9031
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.34159786986004403,
+      "learning_rate": 3.759951963136008e-06,
+      "loss": 0.6387,
+      "step": 9032
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.40334504892463097,
+      "learning_rate": 3.7565721069424e-06,
+      "loss": 0.6372,
+      "step": 9033
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.38350276440094305,
+      "learning_rate": 3.7531936130654735e-06,
+      "loss": 0.6275,
+      "step": 9034
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3940544308935584,
+      "learning_rate": 3.749816481788575e-06,
+      "loss": 0.602,
+      "step": 9035
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39119912958715364,
+      "learning_rate": 3.746440713394941e-06,
+      "loss": 0.6111,
+      "step": 9036
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.41153664850689564,
+      "learning_rate": 3.7430663081676977e-06,
+      "loss": 0.6162,
+      "step": 9037
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3657431906091205,
+      "learning_rate": 3.7396932663898476e-06,
+      "loss": 0.6149,
+      "step": 9038
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37658832013288396,
+      "learning_rate": 3.736321588344285e-06,
+      "loss": 0.6378,
+      "step": 9039
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37112664814298535,
+      "learning_rate": 3.7329512743137962e-06,
+      "loss": 0.6653,
+      "step": 9040
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3799883914476725,
+      "learning_rate": 3.7295823245810338e-06,
+      "loss": 0.6047,
+      "step": 9041
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3806917648108557,
+      "learning_rate": 3.726214739428551e-06,
+      "loss": 0.6575,
+      "step": 9042
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3596418051615498,
+      "learning_rate": 3.7228485191387818e-06,
+      "loss": 0.618,
+      "step": 9043
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.38793135191949285,
+      "learning_rate": 3.7194836639940523e-06,
+      "loss": 0.6379,
+      "step": 9044
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3333238678284765,
+      "learning_rate": 3.716120174276572e-06,
+      "loss": 0.6264,
+      "step": 9045
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.40907286909149165,
+      "learning_rate": 3.712758050268419e-06,
+      "loss": 0.6217,
+      "step": 9046
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39680027593419076,
+      "learning_rate": 3.7093972922515797e-06,
+      "loss": 0.6058,
+      "step": 9047
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3213079871717637,
+      "learning_rate": 3.7060379005079216e-06,
+      "loss": 0.5852,
+      "step": 9048
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37157171684952256,
+      "learning_rate": 3.7026798753191795e-06,
+      "loss": 0.6432,
+      "step": 9049
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3701359031384973,
+      "learning_rate": 3.69932321696699e-06,
+      "loss": 0.6259,
+      "step": 9050
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3802535238478894,
+      "learning_rate": 3.6959679257328862e-06,
+      "loss": 0.6763,
+      "step": 9051
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3682784538187199,
+      "learning_rate": 3.692614001898258e-06,
+      "loss": 0.5984,
+      "step": 9052
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4142457312516836,
+      "learning_rate": 3.6892614457443986e-06,
+      "loss": 0.6364,
+      "step": 9053
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.34937471630795597,
+      "learning_rate": 3.6859102575524874e-06,
+      "loss": 0.6315,
+      "step": 9054
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.417146235304831,
+      "learning_rate": 3.682560437603577e-06,
+      "loss": 0.6063,
+      "step": 9055
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3986212854888357,
+      "learning_rate": 3.6792119861786146e-06,
+      "loss": 0.6212,
+      "step": 9056
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37749831717837107,
+      "learning_rate": 3.675864903558433e-06,
+      "loss": 0.6159,
+      "step": 9057
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.384474757506861,
+      "learning_rate": 3.6725191900237465e-06,
+      "loss": 0.6461,
+      "step": 9058
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.35075635808502686,
+      "learning_rate": 3.6691748458551633e-06,
+      "loss": 0.6187,
+      "step": 9059
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4090954722359648,
+      "learning_rate": 3.6658318713331586e-06,
+      "loss": 0.638,
+      "step": 9060
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 7.619564825600123,
+      "learning_rate": 3.662490266738108e-06,
+      "loss": 6.5488,
+      "step": 9061
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3775336001226615,
+      "learning_rate": 3.6591500323502672e-06,
+      "loss": 0.6477,
+      "step": 9062
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.385160801809114,
+      "learning_rate": 3.655811168449783e-06,
+      "loss": 0.639,
+      "step": 9063
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.40261742033942904,
+      "learning_rate": 3.652473675316668e-06,
+      "loss": 0.6459,
+      "step": 9064
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3608491146250568,
+      "learning_rate": 3.6491375532308526e-06,
+      "loss": 0.6139,
+      "step": 9065
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3900702289381849,
+      "learning_rate": 3.645802802472118e-06,
+      "loss": 0.6451,
+      "step": 9066
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4240568471731022,
+      "learning_rate": 3.642469423320154e-06,
+      "loss": 0.6731,
+      "step": 9067
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39743997124731206,
+      "learning_rate": 3.639137416054528e-06,
+      "loss": 0.68,
+      "step": 9068
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39361312165620665,
+      "learning_rate": 3.635806780954683e-06,
+      "loss": 0.594,
+      "step": 9069
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37979404471754635,
+      "learning_rate": 3.6324775182999593e-06,
+      "loss": 0.6388,
+      "step": 9070
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3800912659734229,
+      "learning_rate": 3.6291496283695814e-06,
+      "loss": 0.6096,
+      "step": 9071
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3875227854244644,
+      "learning_rate": 3.6258231114426524e-06,
+      "loss": 0.6511,
+      "step": 9072
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39272234378310544,
+      "learning_rate": 3.622497967798164e-06,
+      "loss": 0.6255,
+      "step": 9073
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39640407501609887,
+      "learning_rate": 3.6191741977149965e-06,
+      "loss": 0.6466,
+      "step": 9074
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.43687848416012853,
+      "learning_rate": 3.6158518014719015e-06,
+      "loss": 0.6661,
+      "step": 9075
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39541960232711426,
+      "learning_rate": 3.6125307793475297e-06,
+      "loss": 0.628,
+      "step": 9076
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3924682384400229,
+      "learning_rate": 3.609211131620414e-06,
+      "loss": 0.6174,
+      "step": 9077
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 7.344447003925667,
+      "learning_rate": 3.6058928585689556e-06,
+      "loss": 6.6396,
+      "step": 9078
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4185075062096365,
+      "learning_rate": 3.602575960471475e-06,
+      "loss": 0.6964,
+      "step": 9079
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4022153060199269,
+      "learning_rate": 3.59926043760614e-06,
+      "loss": 0.6318,
+      "step": 9080
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37405122234758814,
+      "learning_rate": 3.5959462902510243e-06,
+      "loss": 0.6277,
+      "step": 9081
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4342695181833861,
+      "learning_rate": 3.592633518684088e-06,
+      "loss": 0.6908,
+      "step": 9082
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4073362627597823,
+      "learning_rate": 3.5893221231831586e-06,
+      "loss": 0.6524,
+      "step": 9083
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37687437110348876,
+      "learning_rate": 3.5860121040259623e-06,
+      "loss": 0.6552,
+      "step": 9084
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.401324039569692,
+      "learning_rate": 3.582703461490109e-06,
+      "loss": 0.6496,
+      "step": 9085
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4636955360328137,
+      "learning_rate": 3.5793961958530886e-06,
+      "loss": 0.6687,
+      "step": 9086
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.35346737176815357,
+      "learning_rate": 3.5760903073922792e-06,
+      "loss": 0.6263,
+      "step": 9087
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3510902872277225,
+      "learning_rate": 3.5727857963849456e-06,
+      "loss": 0.6516,
+      "step": 9088
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3643306927314413,
+      "learning_rate": 3.5694826631082235e-06,
+      "loss": 0.6485,
+      "step": 9089
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3948810752399277,
+      "learning_rate": 3.566180907839145e-06,
+      "loss": 0.6111,
+      "step": 9090
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.43321726421452905,
+      "learning_rate": 3.562880530854633e-06,
+      "loss": 0.6689,
+      "step": 9091
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39727950486833546,
+      "learning_rate": 3.559581532431471e-06,
+      "loss": 0.6512,
+      "step": 9092
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.44054260048151456,
+      "learning_rate": 3.556283912846361e-06,
+      "loss": 0.6297,
+      "step": 9093
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.43690152451259706,
+      "learning_rate": 3.552987672375854e-06,
+      "loss": 0.6753,
+      "step": 9094
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.44428505974660115,
+      "learning_rate": 3.549692811296408e-06,
+      "loss": 0.6526,
+      "step": 9095
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37695163465340814,
+      "learning_rate": 3.54639932988436e-06,
+      "loss": 0.5923,
+      "step": 9096
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.41168513019904596,
+      "learning_rate": 3.5431072284159342e-06,
+      "loss": 0.6401,
+      "step": 9097
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4186194006628618,
+      "learning_rate": 3.539816507167224e-06,
+      "loss": 0.6488,
+      "step": 9098
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3912706292714738,
+      "learning_rate": 3.5365271664142254e-06,
+      "loss": 0.62,
+      "step": 9099
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3850451839306901,
+      "learning_rate": 3.53323920643281e-06,
+      "loss": 0.6443,
+      "step": 9100
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3546457123814211,
+      "learning_rate": 3.5299526274987363e-06,
+      "loss": 0.632,
+      "step": 9101
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.40528246269329055,
+      "learning_rate": 3.526667429887649e-06,
+      "loss": 0.6221,
+      "step": 9102
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.44412862753982235,
+      "learning_rate": 3.5233836138750644e-06,
+      "loss": 0.6521,
+      "step": 9103
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3761667475760257,
+      "learning_rate": 3.520101179736399e-06,
+      "loss": 0.6411,
+      "step": 9104
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3676212814419851,
+      "learning_rate": 3.5168201277469494e-06,
+      "loss": 0.6134,
+      "step": 9105
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4377193537571678,
+      "learning_rate": 3.5135404581818786e-06,
+      "loss": 0.6732,
+      "step": 9106
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4101992617217282,
+      "learning_rate": 3.5102621713162632e-06,
+      "loss": 0.6428,
+      "step": 9107
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4327655941776556,
+      "learning_rate": 3.5069852674250514e-06,
+      "loss": 0.6891,
+      "step": 9108
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.40148558242040766,
+      "learning_rate": 3.5037097467830616e-06,
+      "loss": 0.6228,
+      "step": 9109
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.36102984732668697,
+      "learning_rate": 3.500435609665012e-06,
+      "loss": 0.6422,
+      "step": 9110
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.37447193685059527,
+      "learning_rate": 3.497162856345506e-06,
+      "loss": 0.5909,
+      "step": 9111
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4518964052446267,
+      "learning_rate": 3.4938914870990125e-06,
+      "loss": 0.6659,
+      "step": 9112
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39929012878867237,
+      "learning_rate": 3.4906215021999133e-06,
+      "loss": 0.681,
+      "step": 9113
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39492445008292526,
+      "learning_rate": 3.4873529019224472e-06,
+      "loss": 0.6703,
+      "step": 9114
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4245832516303558,
+      "learning_rate": 3.4840856865407503e-06,
+      "loss": 0.6754,
+      "step": 9115
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3934711349136147,
+      "learning_rate": 3.4808198563288476e-06,
+      "loss": 0.6529,
+      "step": 9116
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4074030752882321,
+      "learning_rate": 3.477555411560627e-06,
+      "loss": 0.6446,
+      "step": 9117
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4118212306893609,
+      "learning_rate": 3.474292352509878e-06,
+      "loss": 0.6233,
+      "step": 9118
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.38559693754698465,
+      "learning_rate": 3.4710306794502733e-06,
+      "loss": 0.6452,
+      "step": 9119
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3541447338997895,
+      "learning_rate": 3.4677703926553653e-06,
+      "loss": 0.6542,
+      "step": 9120
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3924973457456763,
+      "learning_rate": 3.4645114923985855e-06,
+      "loss": 0.6402,
+      "step": 9121
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4523432777805097,
+      "learning_rate": 3.4612539789532627e-06,
+      "loss": 0.6516,
+      "step": 9122
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3711111132731942,
+      "learning_rate": 3.457997852592592e-06,
+      "loss": 0.6033,
+      "step": 9123
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3990752691681,
+      "learning_rate": 3.4547431135896626e-06,
+      "loss": 0.6268,
+      "step": 9124
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.41861991507091817,
+      "learning_rate": 3.4514897622174505e-06,
+      "loss": 0.6442,
+      "step": 9125
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4473815571821962,
+      "learning_rate": 3.448237798748797e-06,
+      "loss": 0.6469,
+      "step": 9126
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3851748467586013,
+      "learning_rate": 3.444987223456455e-06,
+      "loss": 0.6495,
+      "step": 9127
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.35280789651535965,
+      "learning_rate": 3.4417380366130472e-06,
+      "loss": 0.6232,
+      "step": 9128
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4226555290983706,
+      "learning_rate": 3.438490238491068e-06,
+      "loss": 0.6528,
+      "step": 9129
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.35983464275831994,
+      "learning_rate": 3.4352438293629085e-06,
+      "loss": 0.6419,
+      "step": 9130
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.414214014259284,
+      "learning_rate": 3.4319988095008516e-06,
+      "loss": 0.6573,
+      "step": 9131
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.34539392328814467,
+      "learning_rate": 3.428755179177039e-06,
+      "loss": 0.6226,
+      "step": 9132
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.46174722136447643,
+      "learning_rate": 3.425512938663518e-06,
+      "loss": 0.6559,
+      "step": 9133
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.40764080993826596,
+      "learning_rate": 3.422272088232208e-06,
+      "loss": 0.6426,
+      "step": 9134
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4145267868801667,
+      "learning_rate": 3.419032628154919e-06,
+      "loss": 0.602,
+      "step": 9135
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3894654762785187,
+      "learning_rate": 3.4157945587033426e-06,
+      "loss": 0.6274,
+      "step": 9136
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 8.099855605009466,
+      "learning_rate": 3.4125578801490433e-06,
+      "loss": 6.5093,
+      "step": 9137
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.45547094431367036,
+      "learning_rate": 3.409322592763482e-06,
+      "loss": 0.6571,
+      "step": 9138
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3612614856919645,
+      "learning_rate": 3.406088696818004e-06,
+      "loss": 0.6413,
+      "step": 9139
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3932472819380058,
+      "learning_rate": 3.4028561925838166e-06,
+      "loss": 0.6745,
+      "step": 9140
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.712313714403619,
+      "learning_rate": 3.3996250803320407e-06,
+      "loss": 6.5572,
+      "step": 9141
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.38699120545275567,
+      "learning_rate": 3.396395360333664e-06,
+      "loss": 0.6116,
+      "step": 9142
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3791985556920159,
+      "learning_rate": 3.3931670328595544e-06,
+      "loss": 0.6111,
+      "step": 9143
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3955016982838123,
+      "learning_rate": 3.3899400981804688e-06,
+      "loss": 0.6483,
+      "step": 9144
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.44179087692214963,
+      "learning_rate": 3.386714556567052e-06,
+      "loss": 0.6969,
+      "step": 9145
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4152403926976845,
+      "learning_rate": 3.3834904082898155e-06,
+      "loss": 0.6518,
+      "step": 9146
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3887120799257228,
+      "learning_rate": 3.380267653619174e-06,
+      "loss": 0.6424,
+      "step": 9147
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3464387720641035,
+      "learning_rate": 3.37704629282541e-06,
+      "loss": 0.6433,
+      "step": 9148
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.36799989180117576,
+      "learning_rate": 3.3738263261786996e-06,
+      "loss": 0.6056,
+      "step": 9149
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.37811981502631925,
+      "learning_rate": 3.3706077539490933e-06,
+      "loss": 0.6038,
+      "step": 9150
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.40862960664560216,
+      "learning_rate": 3.3673905764065397e-06,
+      "loss": 0.6373,
+      "step": 9151
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.37839732921966773,
+      "learning_rate": 3.3641747938208447e-06,
+      "loss": 0.625,
+      "step": 9152
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.41638694279369476,
+      "learning_rate": 3.3609604064617196e-06,
+      "loss": 0.6525,
+      "step": 9153
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.46357341045642536,
+      "learning_rate": 3.357747414598749e-06,
+      "loss": 0.6243,
+      "step": 9154
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.33621965466492376,
+      "learning_rate": 3.3545358185014053e-06,
+      "loss": 0.6358,
+      "step": 9155
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.39174114915927494,
+      "learning_rate": 3.351325618439043e-06,
+      "loss": 0.6404,
+      "step": 9156
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.36798543336723355,
+      "learning_rate": 3.348116814680891e-06,
+      "loss": 0.6417,
+      "step": 9157
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3852189602807659,
+      "learning_rate": 3.3449094074960706e-06,
+      "loss": 0.645,
+      "step": 9158
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.39151691771072605,
+      "learning_rate": 3.3417033971535904e-06,
+      "loss": 0.6245,
+      "step": 9159
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4112789291179433,
+      "learning_rate": 3.3384987839223217e-06,
+      "loss": 0.6227,
+      "step": 9160
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.38891107824456855,
+      "learning_rate": 3.335295568071033e-06,
+      "loss": 0.6568,
+      "step": 9161
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.39108606697960874,
+      "learning_rate": 3.3320937498683904e-06,
+      "loss": 0.6255,
+      "step": 9162
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.38505638469644776,
+      "learning_rate": 3.3288933295829097e-06,
+      "loss": 0.657,
+      "step": 9163
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4217446038022115,
+      "learning_rate": 3.3256943074830116e-06,
+      "loss": 0.639,
+      "step": 9164
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3718265553509155,
+      "learning_rate": 3.3224966838369997e-06,
+      "loss": 0.5882,
+      "step": 9165
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4125829301513956,
+      "learning_rate": 3.319300458913044e-06,
+      "loss": 0.6115,
+      "step": 9166
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.36369438946549554,
+      "learning_rate": 3.316105632979214e-06,
+      "loss": 0.6054,
+      "step": 9167
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3972806037679336,
+      "learning_rate": 3.3129122063034556e-06,
+      "loss": 0.6255,
+      "step": 9168
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3715601736313751,
+      "learning_rate": 3.309720179153599e-06,
+      "loss": 0.6247,
+      "step": 9169
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3903762793361397,
+      "learning_rate": 3.306529551797357e-06,
+      "loss": 0.6383,
+      "step": 9170
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.41836384825539624,
+      "learning_rate": 3.3033403245023176e-06,
+      "loss": 0.616,
+      "step": 9171
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.39628334824425615,
+      "learning_rate": 3.30015249753596e-06,
+      "loss": 0.6644,
+      "step": 9172
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.36656832053491245,
+      "learning_rate": 3.296966071165648e-06,
+      "loss": 0.6097,
+      "step": 9173
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4127349767527317,
+      "learning_rate": 3.293781045658615e-06,
+      "loss": 0.654,
+      "step": 9174
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.41209472824109505,
+      "learning_rate": 3.2905974212819868e-06,
+      "loss": 0.6551,
+      "step": 9175
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.3906048980490875,
+      "learning_rate": 3.287415198302781e-06,
+      "loss": 0.6972,
+      "step": 9176
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4347705517581697,
+      "learning_rate": 3.2842343769878738e-06,
+      "loss": 0.6501,
+      "step": 9177
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4420049041628869,
+      "learning_rate": 3.281054957604042e-06,
+      "loss": 0.6604,
+      "step": 9178
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3553935113666295,
+      "learning_rate": 3.2778769404179457e-06,
+      "loss": 0.6528,
+      "step": 9179
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3955017308820569,
+      "learning_rate": 3.2747003256961073e-06,
+      "loss": 0.6586,
+      "step": 9180
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.403380259658854,
+      "learning_rate": 3.2715251137049566e-06,
+      "loss": 0.601,
+      "step": 9181
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.36039592365837375,
+      "learning_rate": 3.2683513047107905e-06,
+      "loss": 0.6246,
+      "step": 9182
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.40011595045144116,
+      "learning_rate": 3.265178898979795e-06,
+      "loss": 0.6209,
+      "step": 9183
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3978135491931749,
+      "learning_rate": 3.262007896778032e-06,
+      "loss": 0.6342,
+      "step": 9184
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4255544985790053,
+      "learning_rate": 3.2588382983714585e-06,
+      "loss": 0.6866,
+      "step": 9185
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3681951464182538,
+      "learning_rate": 3.255670104025894e-06,
+      "loss": 0.6235,
+      "step": 9186
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.35897050434855887,
+      "learning_rate": 3.252503314007056e-06,
+      "loss": 0.6096,
+      "step": 9187
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4929684869206844,
+      "learning_rate": 3.249337928580545e-06,
+      "loss": 0.665,
+      "step": 9188
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3505530088684208,
+      "learning_rate": 3.2461739480118215e-06,
+      "loss": 0.6256,
+      "step": 9189
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.36533962223857924,
+      "learning_rate": 3.2430113725662647e-06,
+      "loss": 0.6251,
+      "step": 9190
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.39473318846757977,
+      "learning_rate": 3.239850202509103e-06,
+      "loss": 0.6573,
+      "step": 9191
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4348930469484813,
+      "learning_rate": 3.2366904381054633e-06,
+      "loss": 0.6674,
+      "step": 9192
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.415072092842408,
+      "learning_rate": 3.2335320796203564e-06,
+      "loss": 0.6322,
+      "step": 9193
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3575750848349233,
+      "learning_rate": 3.230375127318661e-06,
+      "loss": 0.6194,
+      "step": 9194
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.40466065015547736,
+      "learning_rate": 3.2272195814651443e-06,
+      "loss": 0.6541,
+      "step": 9195
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3670167913397534,
+      "learning_rate": 3.2240654423244754e-06,
+      "loss": 0.5887,
+      "step": 9196
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.41160808902249274,
+      "learning_rate": 3.220912710161175e-06,
+      "loss": 0.6578,
+      "step": 9197
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.40039384534998435,
+      "learning_rate": 3.2177613852396594e-06,
+      "loss": 0.6489,
+      "step": 9198
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.38772069984771734,
+      "learning_rate": 3.2146114678242337e-06,
+      "loss": 0.5965,
+      "step": 9199
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.39165070090693027,
+      "learning_rate": 3.211462958179068e-06,
+      "loss": 0.6073,
+      "step": 9200
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.382044068033421,
+      "learning_rate": 3.2083158565682293e-06,
+      "loss": 0.6199,
+      "step": 9201
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4002241944957013,
+      "learning_rate": 3.2051701632556596e-06,
+      "loss": 0.6311,
+      "step": 9202
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.40763361556023414,
+      "learning_rate": 3.202025878505184e-06,
+      "loss": 0.6406,
+      "step": 9203
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3948219331780201,
+      "learning_rate": 3.1988830025805174e-06,
+      "loss": 0.6657,
+      "step": 9204
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3881742048003058,
+      "learning_rate": 3.195741535745238e-06,
+      "loss": 0.63,
+      "step": 9205
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.40318890142503444,
+      "learning_rate": 3.1926014782628223e-06,
+      "loss": 0.6346,
+      "step": 9206
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4227612475393893,
+      "learning_rate": 3.1894628303966212e-06,
+      "loss": 0.6594,
+      "step": 9207
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3875239449681248,
+      "learning_rate": 3.186325592409876e-06,
+      "loss": 0.645,
+      "step": 9208
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.37797667178259337,
+      "learning_rate": 3.1831897645656884e-06,
+      "loss": 0.6147,
+      "step": 9209
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8373823465218468,
+      "learning_rate": 3.1800553471270733e-06,
+      "loss": 0.6361,
+      "step": 9210
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.362932576642273,
+      "learning_rate": 3.1769223403569005e-06,
+      "loss": 0.6252,
+      "step": 9211
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.43111453948625156,
+      "learning_rate": 3.1737907445179326e-06,
+      "loss": 0.639,
+      "step": 9212
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4089848677986117,
+      "learning_rate": 3.170660559872818e-06,
+      "loss": 0.6589,
+      "step": 9213
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4098576494071317,
+      "learning_rate": 3.167531786684075e-06,
+      "loss": 0.6424,
+      "step": 9214
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.34543169978876126,
+      "learning_rate": 3.1644044252141117e-06,
+      "loss": 0.6177,
+      "step": 9215
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.42936825264122197,
+      "learning_rate": 3.1612784757252157e-06,
+      "loss": 0.6636,
+      "step": 9216
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.37427062574890163,
+      "learning_rate": 3.15815393847956e-06,
+      "loss": 0.6389,
+      "step": 9217
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.37163969190842155,
+      "learning_rate": 3.1550308137391927e-06,
+      "loss": 0.5898,
+      "step": 9218
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4038416615271089,
+      "learning_rate": 3.151909101766051e-06,
+      "loss": 0.6109,
+      "step": 9219
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4072333602314851,
+      "learning_rate": 3.1487888028219426e-06,
+      "loss": 0.6382,
+      "step": 9220
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3560377191788296,
+      "learning_rate": 3.1456699171685657e-06,
+      "loss": 0.6194,
+      "step": 9221
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3677902018537821,
+      "learning_rate": 3.1425524450675017e-06,
+      "loss": 0.6356,
+      "step": 9222
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.36624936494063387,
+      "learning_rate": 3.139436386780197e-06,
+      "loss": 0.6308,
+      "step": 9223
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.42176810454161623,
+      "learning_rate": 3.136321742568009e-06,
+      "loss": 0.634,
+      "step": 9224
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4655652216191283,
+      "learning_rate": 3.1332085126921473e-06,
+      "loss": 0.6273,
+      "step": 9225
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3377075988073819,
+      "learning_rate": 3.130096697413716e-06,
+      "loss": 0.5809,
+      "step": 9226
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3265850828708335,
+      "learning_rate": 3.1269862969937057e-06,
+      "loss": 0.5834,
+      "step": 9227
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3913618712765124,
+      "learning_rate": 3.123877311692973e-06,
+      "loss": 0.5903,
+      "step": 9228
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.36131168831837146,
+      "learning_rate": 3.1207697417722692e-06,
+      "loss": 0.5979,
+      "step": 9229
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3785869227956532,
+      "learning_rate": 3.1176635874922214e-06,
+      "loss": 0.6399,
+      "step": 9230
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3970110716378601,
+      "learning_rate": 3.1145588491133405e-06,
+      "loss": 0.62,
+      "step": 9231
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.33836255304837,
+      "learning_rate": 3.111455526896017e-06,
+      "loss": 0.6111,
+      "step": 9232
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.38548070967239806,
+      "learning_rate": 3.1083536211005263e-06,
+      "loss": 0.6665,
+      "step": 9233
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.34897556463406515,
+      "learning_rate": 3.1052531319870137e-06,
+      "loss": 0.6607,
+      "step": 9234
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.35444517992378693,
+      "learning_rate": 3.1021540598155166e-06,
+      "loss": 0.5927,
+      "step": 9235
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5131757944546821,
+      "learning_rate": 3.099056404845955e-06,
+      "loss": 0.6662,
+      "step": 9236
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4171860457847278,
+      "learning_rate": 3.0959601673381124e-06,
+      "loss": 0.6359,
+      "step": 9237
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3876532765836063,
+      "learning_rate": 3.0928653475516855e-06,
+      "loss": 0.6366,
+      "step": 9238
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.35067153785723765,
+      "learning_rate": 3.0897719457462185e-06,
+      "loss": 0.612,
+      "step": 9239
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3378560851611541,
+      "learning_rate": 3.086679962181156e-06,
+      "loss": 0.5871,
+      "step": 9240
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4064699695653542,
+      "learning_rate": 3.083589397115818e-06,
+      "loss": 0.6355,
+      "step": 9241
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3849007563327919,
+      "learning_rate": 3.0805002508094127e-06,
+      "loss": 0.6589,
+      "step": 9242
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3856391331067986,
+      "learning_rate": 3.0774125235210127e-06,
+      "loss": 0.6394,
+      "step": 9243
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.41991785308669666,
+      "learning_rate": 3.074326215509589e-06,
+      "loss": 0.6364,
+      "step": 9244
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.40344401194496676,
+      "learning_rate": 3.0712413270339826e-06,
+      "loss": 0.6559,
+      "step": 9245
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.40896275636726864,
+      "learning_rate": 3.0681578583529226e-06,
+      "loss": 0.6347,
+      "step": 9246
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4454640160432805,
+      "learning_rate": 3.0650758097250178e-06,
+      "loss": 0.6607,
+      "step": 9247
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.34877616964634434,
+      "learning_rate": 3.06199518140875e-06,
+      "loss": 0.5801,
+      "step": 9248
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4119015909316201,
+      "learning_rate": 3.0589159736624885e-06,
+      "loss": 0.6688,
+      "step": 9249
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3910745292604737,
+      "learning_rate": 3.0558381867444884e-06,
+      "loss": 0.6534,
+      "step": 9250
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4529716996608093,
+      "learning_rate": 3.0527618209128705e-06,
+      "loss": 0.6886,
+      "step": 9251
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.37316762815021065,
+      "learning_rate": 3.0496868764256527e-06,
+      "loss": 0.6289,
+      "step": 9252
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.39395999923336794,
+      "learning_rate": 3.0466133535407325e-06,
+      "loss": 0.6453,
+      "step": 9253
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.39479146303663787,
+      "learning_rate": 3.0435412525158713e-06,
+      "loss": 0.625,
+      "step": 9254
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.34386636971256174,
+      "learning_rate": 3.0404705736087247e-06,
+      "loss": 0.6561,
+      "step": 9255
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3944436609898028,
+      "learning_rate": 3.0374013170768356e-06,
+      "loss": 0.6452,
+      "step": 9256
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.32832729648441394,
+      "learning_rate": 3.0343334831776096e-06,
+      "loss": 0.6243,
+      "step": 9257
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.38911557339005987,
+      "learning_rate": 3.0312670721683424e-06,
+      "loss": 0.6563,
+      "step": 9258
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.45840515504349255,
+      "learning_rate": 3.028202084306213e-06,
+      "loss": 0.6125,
+      "step": 9259
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.38106398629880844,
+      "learning_rate": 3.0251385198482786e-06,
+      "loss": 0.6363,
+      "step": 9260
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3986056741572743,
+      "learning_rate": 3.0220763790514816e-06,
+      "loss": 0.6476,
+      "step": 9261
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.36729665558403635,
+      "learning_rate": 3.0190156621726284e-06,
+      "loss": 0.6399,
+      "step": 9262
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.38768682994870857,
+      "learning_rate": 3.0159563694684245e-06,
+      "loss": 0.646,
+      "step": 9263
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3956594143244139,
+      "learning_rate": 3.012898501195447e-06,
+      "loss": 0.6695,
+      "step": 9264
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4205702690920152,
+      "learning_rate": 3.0098420576101593e-06,
+      "loss": 0.6458,
+      "step": 9265
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.354659143157351,
+      "learning_rate": 3.006787038968897e-06,
+      "loss": 0.6192,
+      "step": 9266
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3784844976953588,
+      "learning_rate": 3.00373344552789e-06,
+      "loss": 0.6561,
+      "step": 9267
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.38464768665043453,
+      "learning_rate": 3.000681277543227e-06,
+      "loss": 0.6482,
+      "step": 9268
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3613586082255094,
+      "learning_rate": 2.9976305352708945e-06,
+      "loss": 0.6009,
+      "step": 9269
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3727202592712636,
+      "learning_rate": 2.994581218966761e-06,
+      "loss": 0.6493,
+      "step": 9270
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.41815173666086775,
+      "learning_rate": 2.991533328886558e-06,
+      "loss": 0.6665,
+      "step": 9271
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3890006349157081,
+      "learning_rate": 2.9884868652859087e-06,
+      "loss": 0.6278,
+      "step": 9272
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3741666042845425,
+      "learning_rate": 2.9854418284203324e-06,
+      "loss": 0.6366,
+      "step": 9273
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3993373702390092,
+      "learning_rate": 2.982398218545197e-06,
+      "loss": 0.6382,
+      "step": 9274
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.37530645215495256,
+      "learning_rate": 2.979356035915768e-06,
+      "loss": 0.6247,
+      "step": 9275
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5216668258195091,
+      "learning_rate": 2.976315280787201e-06,
+      "loss": 0.6347,
+      "step": 9276
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.39944099403026556,
+      "learning_rate": 2.973275953414505e-06,
+      "loss": 0.6547,
+      "step": 9277
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.36804871318459215,
+      "learning_rate": 2.970238054052592e-06,
+      "loss": 0.6282,
+      "step": 9278
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4096260768250567,
+      "learning_rate": 2.967201582956245e-06,
+      "loss": 0.671,
+      "step": 9279
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3996202464758894,
+      "learning_rate": 2.9641665403801332e-06,
+      "loss": 0.6457,
+      "step": 9280
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3845378663711345,
+      "learning_rate": 2.961132926578805e-06,
+      "loss": 0.6437,
+      "step": 9281
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3429353885944317,
+      "learning_rate": 2.9581007418066732e-06,
+      "loss": 0.6424,
+      "step": 9282
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3775231967213199,
+      "learning_rate": 2.9550699863180533e-06,
+      "loss": 0.6493,
+      "step": 9283
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.37007857670693667,
+      "learning_rate": 2.9520406603671303e-06,
+      "loss": 0.6394,
+      "step": 9284
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3438809249742163,
+      "learning_rate": 2.949012764207961e-06,
+      "loss": 0.6242,
+      "step": 9285
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.42604520682490565,
+      "learning_rate": 2.945986298094503e-06,
+      "loss": 0.6583,
+      "step": 9286
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6343592973395211,
+      "learning_rate": 2.942961262280584e-06,
+      "loss": 0.6731,
+      "step": 9287
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3560792320352897,
+      "learning_rate": 2.9399376570198978e-06,
+      "loss": 0.6293,
+      "step": 9288
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4352211501916987,
+      "learning_rate": 2.9369154825660385e-06,
+      "loss": 0.6635,
+      "step": 9289
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4142534990303221,
+      "learning_rate": 2.9338947391724738e-06,
+      "loss": 0.6521,
+      "step": 9290
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.35865596774387354,
+      "learning_rate": 2.930875427092543e-06,
+      "loss": 0.6248,
+      "step": 9291
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.33525297943986193,
+      "learning_rate": 2.9278575465794755e-06,
+      "loss": 0.5964,
+      "step": 9292
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3994684746056048,
+      "learning_rate": 2.9248410978863775e-06,
+      "loss": 0.6293,
+      "step": 9293
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3918233291184679,
+      "learning_rate": 2.9218260812662345e-06,
+      "loss": 0.634,
+      "step": 9294
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3425203535744534,
+      "learning_rate": 2.918812496971919e-06,
+      "loss": 0.6081,
+      "step": 9295
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4183982395681645,
+      "learning_rate": 2.9158003452561656e-06,
+      "loss": 0.67,
+      "step": 9296
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.37424277361243896,
+      "learning_rate": 2.9127896263716037e-06,
+      "loss": 0.6259,
+      "step": 9297
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3850942095174865,
+      "learning_rate": 2.9097803405707405e-06,
+      "loss": 0.6234,
+      "step": 9298
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3765846551152216,
+      "learning_rate": 2.90677248810596e-06,
+      "loss": 0.6469,
+      "step": 9299
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4034408221494101,
+      "learning_rate": 2.903766069229528e-06,
+      "loss": 0.6496,
+      "step": 9300
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.37212610091683507,
+      "learning_rate": 2.900761084193593e-06,
+      "loss": 0.6132,
+      "step": 9301
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.37642595479436786,
+      "learning_rate": 2.8977575332501716e-06,
+      "loss": 0.6381,
+      "step": 9302
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3629764573931412,
+      "learning_rate": 2.8947554166511717e-06,
+      "loss": 0.6329,
+      "step": 9303
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3818713437355654,
+      "learning_rate": 2.89175473464838e-06,
+      "loss": 0.6548,
+      "step": 9304
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.32632888074572786,
+      "learning_rate": 2.888755487493453e-06,
+      "loss": 0.6089,
+      "step": 9305
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.37643489156913523,
+      "learning_rate": 2.8857576754379345e-06,
+      "loss": 0.6416,
+      "step": 9306
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4429056862851212,
+      "learning_rate": 2.882761298733259e-06,
+      "loss": 0.621,
+      "step": 9307
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3614749252345933,
+      "learning_rate": 2.879766357630718e-06,
+      "loss": 0.5985,
+      "step": 9308
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.40115042136574985,
+      "learning_rate": 2.876772852381495e-06,
+      "loss": 0.6466,
+      "step": 9309
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4048980668743939,
+      "learning_rate": 2.8737807832366595e-06,
+      "loss": 0.654,
+      "step": 9310
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3763527769435453,
+      "learning_rate": 2.8707901504471404e-06,
+      "loss": 0.6059,
+      "step": 9311
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.401404170703451,
+      "learning_rate": 2.867800954263764e-06,
+      "loss": 0.6302,
+      "step": 9312
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.38167383708346214,
+      "learning_rate": 2.8648131949372325e-06,
+      "loss": 0.6694,
+      "step": 9313
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3939214457166032,
+      "learning_rate": 2.861826872718123e-06,
+      "loss": 0.6092,
+      "step": 9314
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.41681924083779864,
+      "learning_rate": 2.8588419878569017e-06,
+      "loss": 0.663,
+      "step": 9315
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3539783620002946,
+      "learning_rate": 2.8558585406038976e-06,
+      "loss": 0.6358,
+      "step": 9316
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3896330817645872,
+      "learning_rate": 2.8528765312093323e-06,
+      "loss": 0.6358,
+      "step": 9317
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.43440612137590623,
+      "learning_rate": 2.8498959599233077e-06,
+      "loss": 0.6638,
+      "step": 9318
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.40165917885987434,
+      "learning_rate": 2.8469168269957915e-06,
+      "loss": 0.6432,
+      "step": 9319
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 9.336297135497562,
+      "learning_rate": 2.843939132676643e-06,
+      "loss": 6.6582,
+      "step": 9320
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4012837152471021,
+      "learning_rate": 2.840962877215607e-06,
+      "loss": 0.701,
+      "step": 9321
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3637398619242294,
+      "learning_rate": 2.837988060862289e-06,
+      "loss": 0.5919,
+      "step": 9322
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3613731759363745,
+      "learning_rate": 2.835014683866184e-06,
+      "loss": 0.6082,
+      "step": 9323
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3641640357865708,
+      "learning_rate": 2.8320427464766733e-06,
+      "loss": 0.6165,
+      "step": 9324
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3945342096714525,
+      "learning_rate": 2.829072248942999e-06,
+      "loss": 0.6012,
+      "step": 9325
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.31469379223771776,
+      "learning_rate": 2.826103191514298e-06,
+      "loss": 0.621,
+      "step": 9326
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3867492931545694,
+      "learning_rate": 2.8231355744395815e-06,
+      "loss": 0.6449,
+      "step": 9327
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.38821880692826916,
+      "learning_rate": 2.82016939796774e-06,
+      "loss": 0.6135,
+      "step": 9328
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3741225437188807,
+      "learning_rate": 2.8172046623475436e-06,
+      "loss": 0.6286,
+      "step": 9329
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.37334264099417774,
+      "learning_rate": 2.814241367827644e-06,
+      "loss": 0.6393,
+      "step": 9330
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3606652168129716,
+      "learning_rate": 2.811279514656562e-06,
+      "loss": 0.6212,
+      "step": 9331
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.39399203391589005,
+      "learning_rate": 2.8083191030827083e-06,
+      "loss": 0.6403,
+      "step": 9332
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3651725284415953,
+      "learning_rate": 2.8053601333543736e-06,
+      "loss": 0.6129,
+      "step": 9333
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 3.6877387386531977,
+      "learning_rate": 2.8024026057197096e-06,
+      "loss": 6.5991,
+      "step": 9334
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.41072356640169233,
+      "learning_rate": 2.7994465204267784e-06,
+      "loss": 0.6732,
+      "step": 9335
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.35388200085510574,
+      "learning_rate": 2.7964918777234905e-06,
+      "loss": 0.6327,
+      "step": 9336
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.390161465234441,
+      "learning_rate": 2.7935386778576522e-06,
+      "loss": 0.6888,
+      "step": 9337
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.37331533581152054,
+      "learning_rate": 2.79058692107695e-06,
+      "loss": 0.5896,
+      "step": 9338
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3902378800231429,
+      "learning_rate": 2.787636607628934e-06,
+      "loss": 0.6791,
+      "step": 9339
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.39544971199528495,
+      "learning_rate": 2.784687737761047e-06,
+      "loss": 0.6341,
+      "step": 9340
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4019698277632915,
+      "learning_rate": 2.781740311720611e-06,
+      "loss": 0.6217,
+      "step": 9341
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4492160643436231,
+      "learning_rate": 2.778794329754819e-06,
+      "loss": 0.6395,
+      "step": 9342
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4059318266663716,
+      "learning_rate": 2.7758497921107474e-06,
+      "loss": 0.6374,
+      "step": 9343
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4163231794304062,
+      "learning_rate": 2.772906699035358e-06,
+      "loss": 0.6511,
+      "step": 9344
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4176127109392462,
+      "learning_rate": 2.7699650507754763e-06,
+      "loss": 0.7046,
+      "step": 9345
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3735990441989358,
+      "learning_rate": 2.767024847577815e-06,
+      "loss": 0.6285,
+      "step": 9346
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 3.8786215699837094,
+      "learning_rate": 2.7640860896889734e-06,
+      "loss": 6.4785,
+      "step": 9347
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3886155737599547,
+      "learning_rate": 2.761148777355405e-06,
+      "loss": 0.6626,
+      "step": 9348
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3892325000994909,
+      "learning_rate": 2.758212910823479e-06,
+      "loss": 0.6224,
+      "step": 9349
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4045976823637585,
+      "learning_rate": 2.755278490339408e-06,
+      "loss": 0.6506,
+      "step": 9350
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3707113806969339,
+      "learning_rate": 2.7523455161493063e-06,
+      "loss": 0.6335,
+      "step": 9351
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.39431176567249343,
+      "learning_rate": 2.749413988499159e-06,
+      "loss": 0.6533,
+      "step": 9352
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.44762967854620683,
+      "learning_rate": 2.7464839076348224e-06,
+      "loss": 0.669,
+      "step": 9353
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.39332843122514427,
+      "learning_rate": 2.7435552738020434e-06,
+      "loss": 0.5958,
+      "step": 9354
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38150528052697386,
+      "learning_rate": 2.7406280872464442e-06,
+      "loss": 0.627,
+      "step": 9355
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3391937615045822,
+      "learning_rate": 2.7377023482135223e-06,
+      "loss": 0.6334,
+      "step": 9356
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3539476779885909,
+      "learning_rate": 2.734778056948657e-06,
+      "loss": 0.6053,
+      "step": 9357
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.41157980381415643,
+      "learning_rate": 2.73185521369711e-06,
+      "loss": 0.674,
+      "step": 9358
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.42241605786272274,
+      "learning_rate": 2.728933818704007e-06,
+      "loss": 0.6265,
+      "step": 9359
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3935168735587659,
+      "learning_rate": 2.726013872214366e-06,
+      "loss": 0.641,
+      "step": 9360
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.36933551929692554,
+      "learning_rate": 2.7230953744730838e-06,
+      "loss": 0.6249,
+      "step": 9361
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.243527856511405,
+      "learning_rate": 2.7201783257249203e-06,
+      "loss": 6.5206,
+      "step": 9362
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.404340881065244,
+      "learning_rate": 2.7172627262145334e-06,
+      "loss": 0.6373,
+      "step": 9363
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.346724845648041,
+      "learning_rate": 2.714348576186456e-06,
+      "loss": 0.6451,
+      "step": 9364
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.36936239740664156,
+      "learning_rate": 2.7114358758850823e-06,
+      "loss": 0.6311,
+      "step": 9365
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38890668304455933,
+      "learning_rate": 2.7085246255547048e-06,
+      "loss": 0.6632,
+      "step": 9366
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.42402091214847937,
+      "learning_rate": 2.7056148254394863e-06,
+      "loss": 0.5973,
+      "step": 9367
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.39349646543555744,
+      "learning_rate": 2.7027064757834566e-06,
+      "loss": 0.6116,
+      "step": 9368
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.45326688540449683,
+      "learning_rate": 2.6997995768305553e-06,
+      "loss": 0.647,
+      "step": 9369
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3794514766261119,
+      "learning_rate": 2.6968941288245652e-06,
+      "loss": 0.6391,
+      "step": 9370
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4115205760620854,
+      "learning_rate": 2.6939901320091676e-06,
+      "loss": 0.6343,
+      "step": 9371
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3639227829600642,
+      "learning_rate": 2.691087586627923e-06,
+      "loss": 0.6329,
+      "step": 9372
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4053053198214743,
+      "learning_rate": 2.688186492924252e-06,
+      "loss": 0.6269,
+      "step": 9373
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3872490536821609,
+      "learning_rate": 2.6852868511414733e-06,
+      "loss": 0.672,
+      "step": 9374
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3483661137764451,
+      "learning_rate": 2.6823886615227767e-06,
+      "loss": 0.627,
+      "step": 9375
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4183471793161821,
+      "learning_rate": 2.679491924311226e-06,
+      "loss": 0.6175,
+      "step": 9376
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3640167974303574,
+      "learning_rate": 2.6765966397497712e-06,
+      "loss": 0.6311,
+      "step": 9377
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.33515699276671107,
+      "learning_rate": 2.67370280808124e-06,
+      "loss": 0.6337,
+      "step": 9378
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4505812366457755,
+      "learning_rate": 2.670810429548325e-06,
+      "loss": 0.6506,
+      "step": 9379
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3860648219402363,
+      "learning_rate": 2.66791950439361e-06,
+      "loss": 0.5896,
+      "step": 9380
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.39168911937772055,
+      "learning_rate": 2.6650300328595593e-06,
+      "loss": 0.6355,
+      "step": 9381
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.41290610640070435,
+      "learning_rate": 2.6621420151884957e-06,
+      "loss": 0.6451,
+      "step": 9382
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.40796716831573687,
+      "learning_rate": 2.659255451622651e-06,
+      "loss": 0.6321,
+      "step": 9383
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4251611620641384,
+      "learning_rate": 2.6563703424041045e-06,
+      "loss": 0.6757,
+      "step": 9384
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.41594846775806465,
+      "learning_rate": 2.653486687774831e-06,
+      "loss": 0.631,
+      "step": 9385
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.45810162510646146,
+      "learning_rate": 2.6506044879766823e-06,
+      "loss": 0.6654,
+      "step": 9386
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4017656999520155,
+      "learning_rate": 2.647723743251385e-06,
+      "loss": 0.6574,
+      "step": 9387
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3890397582614846,
+      "learning_rate": 2.644844453840536e-06,
+      "loss": 0.6592,
+      "step": 9388
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4406913920125624,
+      "learning_rate": 2.641966619985623e-06,
+      "loss": 0.6753,
+      "step": 9389
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3954405238647306,
+      "learning_rate": 2.6390902419280084e-06,
+      "loss": 0.6557,
+      "step": 9390
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.37947863818935706,
+      "learning_rate": 2.6362153199089257e-06,
+      "loss": 0.6358,
+      "step": 9391
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3882156117292672,
+      "learning_rate": 2.633341854169502e-06,
+      "loss": 0.6477,
+      "step": 9392
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3927357335536313,
+      "learning_rate": 2.630469844950716e-06,
+      "loss": 0.663,
+      "step": 9393
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4034904295494623,
+      "learning_rate": 2.627599292493448e-06,
+      "loss": 0.6252,
+      "step": 9394
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.39169472173648334,
+      "learning_rate": 2.624730197038452e-06,
+      "loss": 0.6588,
+      "step": 9395
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3985526930869564,
+      "learning_rate": 2.62186255882634e-06,
+      "loss": 0.6606,
+      "step": 9396
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3772370784010237,
+      "learning_rate": 2.6189963780976337e-06,
+      "loss": 0.5885,
+      "step": 9397
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4129798429195629,
+      "learning_rate": 2.616131655092713e-06,
+      "loss": 0.6114,
+      "step": 9398
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3711686795637038,
+      "learning_rate": 2.613268390051833e-06,
+      "loss": 0.6383,
+      "step": 9399
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.37419749522262363,
+      "learning_rate": 2.610406583215135e-06,
+      "loss": 0.646,
+      "step": 9400
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.46175806688795495,
+      "learning_rate": 2.6075462348226378e-06,
+      "loss": 0.697,
+      "step": 9401
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38074471762431844,
+      "learning_rate": 2.6046873451142317e-06,
+      "loss": 0.6202,
+      "step": 9402
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3705097101054331,
+      "learning_rate": 2.6018299143296878e-06,
+      "loss": 0.6381,
+      "step": 9403
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.36498443941861214,
+      "learning_rate": 2.598973942708658e-06,
+      "loss": 0.6606,
+      "step": 9404
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38746837696960884,
+      "learning_rate": 2.596119430490669e-06,
+      "loss": 0.5961,
+      "step": 9405
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3368402323271819,
+      "learning_rate": 2.593266377915129e-06,
+      "loss": 0.6302,
+      "step": 9406
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3229061460998936,
+      "learning_rate": 2.5904147852213115e-06,
+      "loss": 0.6129,
+      "step": 9407
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3699437397374817,
+      "learning_rate": 2.58756465264838e-06,
+      "loss": 0.6461,
+      "step": 9408
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.40395257733578205,
+      "learning_rate": 2.584715980435377e-06,
+      "loss": 0.6464,
+      "step": 9409
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.39440805913277205,
+      "learning_rate": 2.581868768821203e-06,
+      "loss": 0.6393,
+      "step": 9410
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.409809075177583,
+      "learning_rate": 2.579023018044664e-06,
+      "loss": 0.6323,
+      "step": 9411
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38026766143928803,
+      "learning_rate": 2.5761787283444293e-06,
+      "loss": 0.6446,
+      "step": 9412
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.37008449728493803,
+      "learning_rate": 2.57333589995904e-06,
+      "loss": 0.6368,
+      "step": 9413
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3401349771167816,
+      "learning_rate": 2.5704945331269217e-06,
+      "loss": 0.6486,
+      "step": 9414
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4169727777202019,
+      "learning_rate": 2.56765462808638e-06,
+      "loss": 0.6472,
+      "step": 9415
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.37869756105012803,
+      "learning_rate": 2.5648161850755894e-06,
+      "loss": 0.6525,
+      "step": 9416
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.39329818472147404,
+      "learning_rate": 2.5619792043326075e-06,
+      "loss": 0.6512,
+      "step": 9417
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38111731929237175,
+      "learning_rate": 2.5591436860953713e-06,
+      "loss": 0.6468,
+      "step": 9418
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3725825570719603,
+      "learning_rate": 2.55630963060169e-06,
+      "loss": 0.6275,
+      "step": 9419
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3961844997793939,
+      "learning_rate": 2.5534770380892516e-06,
+      "loss": 0.6586,
+      "step": 9420
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.37678634557087654,
+      "learning_rate": 2.55064590879563e-06,
+      "loss": 0.652,
+      "step": 9421
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4634338225377279,
+      "learning_rate": 2.5478162429582588e-06,
+      "loss": 0.647,
+      "step": 9422
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3555523581953869,
+      "learning_rate": 2.5449880408144624e-06,
+      "loss": 0.6462,
+      "step": 9423
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.33928418473660216,
+      "learning_rate": 2.542161302601438e-06,
+      "loss": 0.5976,
+      "step": 9424
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.45082469333590613,
+      "learning_rate": 2.539336028556263e-06,
+      "loss": 0.6693,
+      "step": 9425
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.37636694519162883,
+      "learning_rate": 2.5365122189158898e-06,
+      "loss": 0.6651,
+      "step": 9426
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4211963312399071,
+      "learning_rate": 2.5336898739171444e-06,
+      "loss": 0.6547,
+      "step": 9427
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.41891826993120657,
+      "learning_rate": 2.5308689937967334e-06,
+      "loss": 0.6347,
+      "step": 9428
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3404853975493192,
+      "learning_rate": 2.528049578791247e-06,
+      "loss": 0.624,
+      "step": 9429
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3770982762006466,
+      "learning_rate": 2.525231629137139e-06,
+      "loss": 0.5966,
+      "step": 9430
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3859235188117319,
+      "learning_rate": 2.5224151450707446e-06,
+      "loss": 0.6417,
+      "step": 9431
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4339976017764264,
+      "learning_rate": 2.519600126828292e-06,
+      "loss": 0.647,
+      "step": 9432
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38373765429262685,
+      "learning_rate": 2.5167865746458643e-06,
+      "loss": 0.6518,
+      "step": 9433
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.41215467887784,
+      "learning_rate": 2.5139744887594296e-06,
+      "loss": 0.6091,
+      "step": 9434
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3320814706298555,
+      "learning_rate": 2.5111638694048423e-06,
+      "loss": 0.6072,
+      "step": 9435
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4004058291271533,
+      "learning_rate": 2.508354716817816e-06,
+      "loss": 0.6563,
+      "step": 9436
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3528460089730077,
+      "learning_rate": 2.5055470312339548e-06,
+      "loss": 0.6588,
+      "step": 9437
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4085569535064913,
+      "learning_rate": 2.5027408128887354e-06,
+      "loss": 0.6411,
+      "step": 9438
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38731401761152884,
+      "learning_rate": 2.499936062017514e-06,
+      "loss": 0.6215,
+      "step": 9439
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.43442860921784654,
+      "learning_rate": 2.4971327788555222e-06,
+      "loss": 0.669,
+      "step": 9440
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.37217968812494134,
+      "learning_rate": 2.494330963637863e-06,
+      "loss": 0.6228,
+      "step": 9441
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3991257860716944,
+      "learning_rate": 2.491530616599527e-06,
+      "loss": 0.6587,
+      "step": 9442
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38323554675388766,
+      "learning_rate": 2.4887317379753697e-06,
+      "loss": 0.6623,
+      "step": 9443
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4155476953349503,
+      "learning_rate": 2.485934328000139e-06,
+      "loss": 0.6525,
+      "step": 9444
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.44024634680441266,
+      "learning_rate": 2.483138386908437e-06,
+      "loss": 0.6685,
+      "step": 9445
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3772566242655452,
+      "learning_rate": 2.4803439149347708e-06,
+      "loss": 0.636,
+      "step": 9446
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.38075378266005483,
+      "learning_rate": 2.477550912313498e-06,
+      "loss": 0.6104,
+      "step": 9447
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3941110057536342,
+      "learning_rate": 2.4747593792788684e-06,
+      "loss": 0.6087,
+      "step": 9448
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.40522822239112904,
+      "learning_rate": 2.4719693160650084e-06,
+      "loss": 0.6174,
+      "step": 9449
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.41854310452396287,
+      "learning_rate": 2.469180722905911e-06,
+      "loss": 0.6643,
+      "step": 9450
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3382834947245743,
+      "learning_rate": 2.466393600035448e-06,
+      "loss": 0.6144,
+      "step": 9451
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3655156252030006,
+      "learning_rate": 2.4636079476873877e-06,
+      "loss": 0.6661,
+      "step": 9452
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3759944715671386,
+      "learning_rate": 2.460823766095346e-06,
+      "loss": 0.6252,
+      "step": 9453
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4356747268889239,
+      "learning_rate": 2.4580410554928325e-06,
+      "loss": 0.6334,
+      "step": 9454
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3672920244138808,
+      "learning_rate": 2.4552598161132355e-06,
+      "loss": 0.6313,
+      "step": 9455
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.37189385253008594,
+      "learning_rate": 2.4524800481898047e-06,
+      "loss": 0.6533,
+      "step": 9456
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.36107972793265164,
+      "learning_rate": 2.44970175195568e-06,
+      "loss": 0.6094,
+      "step": 9457
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 4.238691562811427,
+      "learning_rate": 2.4469249276438743e-06,
+      "loss": 6.5495,
+      "step": 9458
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4123958733966021,
+      "learning_rate": 2.4441495754872756e-06,
+      "loss": 0.6843,
+      "step": 9459
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.42763953026511825,
+      "learning_rate": 2.441375695718655e-06,
+      "loss": 0.6461,
+      "step": 9460
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3841322484318592,
+      "learning_rate": 2.4386032885706447e-06,
+      "loss": 0.6129,
+      "step": 9461
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4351152968788737,
+      "learning_rate": 2.435832354275769e-06,
+      "loss": 0.6303,
+      "step": 9462
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.418884518163765,
+      "learning_rate": 2.4330628930664248e-06,
+      "loss": 0.6355,
+      "step": 9463
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4106247740239038,
+      "learning_rate": 2.430294905174875e-06,
+      "loss": 0.6416,
+      "step": 9464
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4261063271987929,
+      "learning_rate": 2.427528390833269e-06,
+      "loss": 0.6883,
+      "step": 9465
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3829697504245277,
+      "learning_rate": 2.4247633502736423e-06,
+      "loss": 0.6361,
+      "step": 9466
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.430519770791937,
+      "learning_rate": 2.4219997837278863e-06,
+      "loss": 0.6575,
+      "step": 9467
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40758425382602775,
+      "learning_rate": 2.419237691427778e-06,
+      "loss": 0.6033,
+      "step": 9468
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4185157160021633,
+      "learning_rate": 2.416477073604977e-06,
+      "loss": 0.654,
+      "step": 9469
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38695826337943606,
+      "learning_rate": 2.413717930491002e-06,
+      "loss": 0.6239,
+      "step": 9470
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40777642853685286,
+      "learning_rate": 2.410960262317268e-06,
+      "loss": 0.6101,
+      "step": 9471
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.378474145948402,
+      "learning_rate": 2.4082040693150542e-06,
+      "loss": 0.6042,
+      "step": 9472
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3642468804036223,
+      "learning_rate": 2.4054493517155187e-06,
+      "loss": 0.6489,
+      "step": 9473
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.42328006124088774,
+      "learning_rate": 2.4026961097497026e-06,
+      "loss": 0.7058,
+      "step": 9474
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.391224841629325,
+      "learning_rate": 2.399944343648506e-06,
+      "loss": 0.6309,
+      "step": 9475
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6300212791419207,
+      "learning_rate": 2.397194053642722e-06,
+      "loss": 0.6198,
+      "step": 9476
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.39954657940609706,
+      "learning_rate": 2.394445239963015e-06,
+      "loss": 0.6451,
+      "step": 9477
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38488524826326237,
+      "learning_rate": 2.391697902839927e-06,
+      "loss": 0.6585,
+      "step": 9478
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.42098278447959697,
+      "learning_rate": 2.3889520425038624e-06,
+      "loss": 0.6303,
+      "step": 9479
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40754054377983534,
+      "learning_rate": 2.3862076591851293e-06,
+      "loss": 0.632,
+      "step": 9480
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3827478969096716,
+      "learning_rate": 2.383464753113882e-06,
+      "loss": 0.5779,
+      "step": 9481
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.41518104065742034,
+      "learning_rate": 2.3807233245201733e-06,
+      "loss": 0.6649,
+      "step": 9482
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 9.407814741779955,
+      "learning_rate": 2.377983373633923e-06,
+      "loss": 6.3974,
+      "step": 9483
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38092485720918356,
+      "learning_rate": 2.375244900684923e-06,
+      "loss": 0.6362,
+      "step": 9484
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.384555491637943,
+      "learning_rate": 2.3725079059028456e-06,
+      "loss": 0.6339,
+      "step": 9485
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3898979528396,
+      "learning_rate": 2.369772389517242e-06,
+      "loss": 0.626,
+      "step": 9486
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.35385043901351904,
+      "learning_rate": 2.367038351757538e-06,
+      "loss": 0.6368,
+      "step": 9487
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4156512090639758,
+      "learning_rate": 2.36430579285303e-06,
+      "loss": 0.6667,
+      "step": 9488
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.42274255575422287,
+      "learning_rate": 2.3615747130329013e-06,
+      "loss": 0.6647,
+      "step": 9489
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3925139611243028,
+      "learning_rate": 2.358845112526196e-06,
+      "loss": 0.6773,
+      "step": 9490
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3739396792721214,
+      "learning_rate": 2.3561169915618455e-06,
+      "loss": 0.6106,
+      "step": 9491
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3546338187497382,
+      "learning_rate": 2.3533903503686585e-06,
+      "loss": 0.6306,
+      "step": 9492
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38680059433266667,
+      "learning_rate": 2.350665189175303e-06,
+      "loss": 0.5837,
+      "step": 9493
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3379235558882275,
+      "learning_rate": 2.347941508210352e-06,
+      "loss": 0.6039,
+      "step": 9494
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.36490715526300815,
+      "learning_rate": 2.3452193077022245e-06,
+      "loss": 0.6273,
+      "step": 9495
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3883896602371825,
+      "learning_rate": 2.3424985878792317e-06,
+      "loss": 0.6412,
+      "step": 9496
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38469575820288116,
+      "learning_rate": 2.339779348969562e-06,
+      "loss": 0.635,
+      "step": 9497
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3776101981174721,
+      "learning_rate": 2.337061591201264e-06,
+      "loss": 0.6129,
+      "step": 9498
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.36693042549112875,
+      "learning_rate": 2.33434531480228e-06,
+      "loss": 0.6225,
+      "step": 9499
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3767479068631991,
+      "learning_rate": 2.331630520000421e-06,
+      "loss": 0.6377,
+      "step": 9500
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3942075032407892,
+      "learning_rate": 2.328917207023371e-06,
+      "loss": 0.6163,
+      "step": 9501
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.41023335311992953,
+      "learning_rate": 2.3262053760986936e-06,
+      "loss": 0.6159,
+      "step": 9502
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3923381852825885,
+      "learning_rate": 2.3234950274538302e-06,
+      "loss": 0.6453,
+      "step": 9503
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.36518819236457706,
+      "learning_rate": 2.3207861613160885e-06,
+      "loss": 0.6376,
+      "step": 9504
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.32434877604073264,
+      "learning_rate": 2.31807877791266e-06,
+      "loss": 0.622,
+      "step": 9505
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.36107170948345435,
+      "learning_rate": 2.315372877470614e-06,
+      "loss": 0.6495,
+      "step": 9506
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.36652851842296286,
+      "learning_rate": 2.312668460216878e-06,
+      "loss": 0.644,
+      "step": 9507
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38631341768406224,
+      "learning_rate": 2.309965526378284e-06,
+      "loss": 0.6452,
+      "step": 9508
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.42002228438680955,
+      "learning_rate": 2.307264076181519e-06,
+      "loss": 0.6717,
+      "step": 9509
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.41618603307070906,
+      "learning_rate": 2.3045641098531448e-06,
+      "loss": 0.6164,
+      "step": 9510
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3447923911339994,
+      "learning_rate": 2.3018656276196083e-06,
+      "loss": 0.6229,
+      "step": 9511
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3905936696785141,
+      "learning_rate": 2.2991686297072314e-06,
+      "loss": 0.6159,
+      "step": 9512
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4581996982688068,
+      "learning_rate": 2.2964731163422016e-06,
+      "loss": 0.6904,
+      "step": 9513
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4203803592191196,
+      "learning_rate": 2.2937790877505894e-06,
+      "loss": 0.6813,
+      "step": 9514
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3403514101283238,
+      "learning_rate": 2.291086544158343e-06,
+      "loss": 0.6338,
+      "step": 9515
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.33915611435320825,
+      "learning_rate": 2.2883954857912796e-06,
+      "loss": 0.6553,
+      "step": 9516
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.31748240539499967,
+      "learning_rate": 2.2857059128751e-06,
+      "loss": 0.5824,
+      "step": 9517
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4473331451982576,
+      "learning_rate": 2.2830178256353698e-06,
+      "loss": 0.6189,
+      "step": 9518
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40082808649985846,
+      "learning_rate": 2.2803312242975383e-06,
+      "loss": 0.6551,
+      "step": 9519
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40432845942147544,
+      "learning_rate": 2.2776461090869304e-06,
+      "loss": 0.6132,
+      "step": 9520
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40908547696939523,
+      "learning_rate": 2.2749624802287307e-06,
+      "loss": 0.6637,
+      "step": 9521
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.37014175623076556,
+      "learning_rate": 2.272280337948027e-06,
+      "loss": 0.6041,
+      "step": 9522
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4170493218241037,
+      "learning_rate": 2.269599682469765e-06,
+      "loss": 0.6603,
+      "step": 9523
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4069666031890883,
+      "learning_rate": 2.26692051401876e-06,
+      "loss": 0.6693,
+      "step": 9524
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.44328141423166467,
+      "learning_rate": 2.2642428328197164e-06,
+      "loss": 0.623,
+      "step": 9525
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4439146062620673,
+      "learning_rate": 2.2615666390972126e-06,
+      "loss": 0.6475,
+      "step": 9526
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3863386099792964,
+      "learning_rate": 2.2588919330756866e-06,
+      "loss": 0.6477,
+      "step": 9527
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3676466922857041,
+      "learning_rate": 2.256218714979468e-06,
+      "loss": 0.6428,
+      "step": 9528
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4219376350700261,
+      "learning_rate": 2.2535469850327575e-06,
+      "loss": 0.6585,
+      "step": 9529
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38604052519226795,
+      "learning_rate": 2.2508767434596267e-06,
+      "loss": 0.6629,
+      "step": 9530
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.34897485964732494,
+      "learning_rate": 2.2482079904840348e-06,
+      "loss": 0.6087,
+      "step": 9531
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3285418068346252,
+      "learning_rate": 2.2455407263297957e-06,
+      "loss": 0.6298,
+      "step": 9532
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.35269145490977494,
+      "learning_rate": 2.2428749512206127e-06,
+      "loss": 0.597,
+      "step": 9533
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3981985481928001,
+      "learning_rate": 2.2402106653800626e-06,
+      "loss": 0.6082,
+      "step": 9534
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.39292548439353575,
+      "learning_rate": 2.2375478690315957e-06,
+      "loss": 0.6005,
+      "step": 9535
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.345526040799798,
+      "learning_rate": 2.234886562398537e-06,
+      "loss": 0.6119,
+      "step": 9536
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.35764279834342233,
+      "learning_rate": 2.2322267457040935e-06,
+      "loss": 0.6246,
+      "step": 9537
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3777026654810825,
+      "learning_rate": 2.2295684191713286e-06,
+      "loss": 0.6224,
+      "step": 9538
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.36220679817799833,
+      "learning_rate": 2.226911583023199e-06,
+      "loss": 0.6628,
+      "step": 9539
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.42174865388465665,
+      "learning_rate": 2.224256237482536e-06,
+      "loss": 0.5841,
+      "step": 9540
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.39432888440191843,
+      "learning_rate": 2.2216023827720256e-06,
+      "loss": 0.6368,
+      "step": 9541
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.44176275781190927,
+      "learning_rate": 2.2189500191142566e-06,
+      "loss": 0.6632,
+      "step": 9542
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4016118561136557,
+      "learning_rate": 2.216299146731682e-06,
+      "loss": 0.6618,
+      "step": 9543
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3831930529488113,
+      "learning_rate": 2.2136497658466127e-06,
+      "loss": 0.6395,
+      "step": 9544
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.379120653626365,
+      "learning_rate": 2.2110018766812602e-06,
+      "loss": 0.6487,
+      "step": 9545
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40474581350568317,
+      "learning_rate": 2.208355479457698e-06,
+      "loss": 0.6451,
+      "step": 9546
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3769990802816622,
+      "learning_rate": 2.2057105743978725e-06,
+      "loss": 0.616,
+      "step": 9547
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40108634560044687,
+      "learning_rate": 2.203067161723611e-06,
+      "loss": 0.6425,
+      "step": 9548
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4205569088439093,
+      "learning_rate": 2.200425241656614e-06,
+      "loss": 0.6505,
+      "step": 9549
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4996487367238256,
+      "learning_rate": 2.1977848144184535e-06,
+      "loss": 0.663,
+      "step": 9550
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3901392477249025,
+      "learning_rate": 2.195145880230587e-06,
+      "loss": 0.621,
+      "step": 9551
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.41609358247936745,
+      "learning_rate": 2.192508439314327e-06,
+      "loss": 0.6568,
+      "step": 9552
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.39695786805949707,
+      "learning_rate": 2.1898724918908786e-06,
+      "loss": 0.6432,
+      "step": 9553
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.33890839634288444,
+      "learning_rate": 2.1872380381813207e-06,
+      "loss": 0.6167,
+      "step": 9554
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.37676266840562045,
+      "learning_rate": 2.1846050784065875e-06,
+      "loss": 0.6403,
+      "step": 9555
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38831554595137374,
+      "learning_rate": 2.1819736127875137e-06,
+      "loss": 0.6462,
+      "step": 9556
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.37812418386453855,
+      "learning_rate": 2.179343641544798e-06,
+      "loss": 0.6705,
+      "step": 9557
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.407745440767974,
+      "learning_rate": 2.176715164899006e-06,
+      "loss": 0.647,
+      "step": 9558
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4070184912280225,
+      "learning_rate": 2.1740881830705884e-06,
+      "loss": 0.6509,
+      "step": 9559
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3903640679327684,
+      "learning_rate": 2.171462696279871e-06,
+      "loss": 0.6192,
+      "step": 9560
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.42017093182939874,
+      "learning_rate": 2.1688387047470404e-06,
+      "loss": 0.6831,
+      "step": 9561
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.33007984983055527,
+      "learning_rate": 2.1662162086921733e-06,
+      "loss": 0.6382,
+      "step": 9562
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4123286929817314,
+      "learning_rate": 2.163595208335216e-06,
+      "loss": 0.6576,
+      "step": 9563
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.40112834144432613,
+      "learning_rate": 2.1609757038959887e-06,
+      "loss": 0.6643,
+      "step": 9564
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3968323386774719,
+      "learning_rate": 2.158357695594182e-06,
+      "loss": 0.5987,
+      "step": 9565
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4460570678710844,
+      "learning_rate": 2.1557411836493735e-06,
+      "loss": 0.6349,
+      "step": 9566
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.419286059966957,
+      "learning_rate": 2.1531261682809988e-06,
+      "loss": 0.6625,
+      "step": 9567
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3531315186842496,
+      "learning_rate": 2.1505126497083783e-06,
+      "loss": 0.6767,
+      "step": 9568
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.399625826306492,
+      "learning_rate": 2.1479006281507032e-06,
+      "loss": 0.6633,
+      "step": 9569
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3791756630975571,
+      "learning_rate": 2.1452901038270445e-06,
+      "loss": 0.6232,
+      "step": 9570
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.37812934754751965,
+      "learning_rate": 2.1426810769563457e-06,
+      "loss": 0.6484,
+      "step": 9571
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4083173072556893,
+      "learning_rate": 2.140073547757413e-06,
+      "loss": 0.6392,
+      "step": 9572
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3930885261496717,
+      "learning_rate": 2.1374675164489454e-06,
+      "loss": 0.6655,
+      "step": 9573
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3591015806015641,
+      "learning_rate": 2.134862983249506e-06,
+      "loss": 0.5937,
+      "step": 9574
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4149846374129184,
+      "learning_rate": 2.132259948377531e-06,
+      "loss": 0.617,
+      "step": 9575
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3837406052026535,
+      "learning_rate": 2.12965841205133e-06,
+      "loss": 0.6098,
+      "step": 9576
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3917855371928496,
+      "learning_rate": 2.127058374489104e-06,
+      "loss": 0.6223,
+      "step": 9577
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38684197401078807,
+      "learning_rate": 2.124459835908903e-06,
+      "loss": 0.6458,
+      "step": 9578
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.42652726537425556,
+      "learning_rate": 2.1218627965286664e-06,
+      "loss": 0.6971,
+      "step": 9579
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 3.735680331364691,
+      "learning_rate": 2.119267256566211e-06,
+      "loss": 6.4755,
+      "step": 9580
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3624678351013458,
+      "learning_rate": 2.1166732162392134e-06,
+      "loss": 0.5807,
+      "step": 9581
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.39383369432256926,
+      "learning_rate": 2.1140806757652334e-06,
+      "loss": 0.6472,
+      "step": 9582
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3562615916348256,
+      "learning_rate": 2.1114896353617074e-06,
+      "loss": 0.6337,
+      "step": 9583
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.41170429012216403,
+      "learning_rate": 2.1089000952459427e-06,
+      "loss": 0.6596,
+      "step": 9584
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4300333790347588,
+      "learning_rate": 2.1063120556351225e-06,
+      "loss": 0.62,
+      "step": 9585
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3878567710065717,
+      "learning_rate": 2.1037255167462954e-06,
+      "loss": 0.6237,
+      "step": 9586
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.36354264069140685,
+      "learning_rate": 2.1011404787963975e-06,
+      "loss": 0.6037,
+      "step": 9587
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38495530442398057,
+      "learning_rate": 2.0985569420022366e-06,
+      "loss": 0.6231,
+      "step": 9588
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.40794327321274043,
+      "learning_rate": 2.0959749065804803e-06,
+      "loss": 0.6564,
+      "step": 9589
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4033417781924274,
+      "learning_rate": 2.093394372747681e-06,
+      "loss": 0.6783,
+      "step": 9590
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4039378925330645,
+      "learning_rate": 2.090815340720278e-06,
+      "loss": 0.6423,
+      "step": 9591
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3590576242820244,
+      "learning_rate": 2.088237810714562e-06,
+      "loss": 0.626,
+      "step": 9592
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4343455954356665,
+      "learning_rate": 2.085661782946706e-06,
+      "loss": 0.6512,
+      "step": 9593
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.42181732073056366,
+      "learning_rate": 2.0830872576327652e-06,
+      "loss": 0.6831,
+      "step": 9594
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4080507571306859,
+      "learning_rate": 2.080514234988655e-06,
+      "loss": 0.5997,
+      "step": 9595
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.36450008676586537,
+      "learning_rate": 2.0779427152301766e-06,
+      "loss": 0.6086,
+      "step": 9596
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.44937405539822617,
+      "learning_rate": 2.075372698572995e-06,
+      "loss": 0.6872,
+      "step": 9597
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.39161131213181155,
+      "learning_rate": 2.0728041852326597e-06,
+      "loss": 0.6639,
+      "step": 9598
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.39838615329623106,
+      "learning_rate": 2.0702371754245875e-06,
+      "loss": 0.6266,
+      "step": 9599
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.37977443772540714,
+      "learning_rate": 2.067671669364071e-06,
+      "loss": 0.6054,
+      "step": 9600
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4226992612311745,
+      "learning_rate": 2.0651076672662727e-06,
+      "loss": 0.622,
+      "step": 9601
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3916916498198411,
+      "learning_rate": 2.062545169346235e-06,
+      "loss": 0.6181,
+      "step": 9602
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4132096298812721,
+      "learning_rate": 2.0599841758188742e-06,
+      "loss": 0.6249,
+      "step": 9603
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3697682894798332,
+      "learning_rate": 2.0574246868989658e-06,
+      "loss": 0.6097,
+      "step": 9604
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.35394053521322244,
+      "learning_rate": 2.0548667028011883e-06,
+      "loss": 0.6114,
+      "step": 9605
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4312283223650054,
+      "learning_rate": 2.052310223740064e-06,
+      "loss": 0.631,
+      "step": 9606
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4098432962415797,
+      "learning_rate": 2.0497552499300054e-06,
+      "loss": 0.6198,
+      "step": 9607
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.43216040641132597,
+      "learning_rate": 2.047201781585302e-06,
+      "loss": 0.6394,
+      "step": 9608
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3481190902408991,
+      "learning_rate": 2.044649818920097e-06,
+      "loss": 0.6317,
+      "step": 9609
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.33850855071787994,
+      "learning_rate": 2.0420993621484285e-06,
+      "loss": 0.633,
+      "step": 9610
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38588259716139384,
+      "learning_rate": 2.039550411484199e-06,
+      "loss": 0.6512,
+      "step": 9611
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.99036842360886,
+      "learning_rate": 2.037002967141184e-06,
+      "loss": 6.4611,
+      "step": 9612
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3551380498581021,
+      "learning_rate": 2.034457029333039e-06,
+      "loss": 0.641,
+      "step": 9613
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38936644301467505,
+      "learning_rate": 2.031912598273289e-06,
+      "loss": 0.6526,
+      "step": 9614
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.37209428748198226,
+      "learning_rate": 2.0293696741753256e-06,
+      "loss": 0.619,
+      "step": 9615
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.43043389351637545,
+      "learning_rate": 2.0268282572524223e-06,
+      "loss": 0.6784,
+      "step": 9616
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.36655608001516543,
+      "learning_rate": 2.0242883477177287e-06,
+      "loss": 0.6239,
+      "step": 9617
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.052965043114185,
+      "learning_rate": 2.021749945784264e-06,
+      "loss": 6.488,
+      "step": 9618
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3335410767020154,
+      "learning_rate": 2.019213051664921e-06,
+      "loss": 0.6236,
+      "step": 9619
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3845602324498524,
+      "learning_rate": 2.0166776655724595e-06,
+      "loss": 0.6546,
+      "step": 9620
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.36768580995878913,
+      "learning_rate": 2.014143787719525e-06,
+      "loss": 0.6013,
+      "step": 9621
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3357146416864523,
+      "learning_rate": 2.011611418318631e-06,
+      "loss": 0.5694,
+      "step": 9622
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4242887937499874,
+      "learning_rate": 2.0090805575821637e-06,
+      "loss": 0.6794,
+      "step": 9623
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4443181662947698,
+      "learning_rate": 2.006551205722376e-06,
+      "loss": 0.6624,
+      "step": 9624
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.35468942385128727,
+      "learning_rate": 2.004023362951417e-06,
+      "loss": 0.5904,
+      "step": 9625
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4156444749967388,
+      "learning_rate": 2.0014970294812807e-06,
+      "loss": 0.6288,
+      "step": 9626
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.35309985946903344,
+      "learning_rate": 1.99897220552385e-06,
+      "loss": 0.5943,
+      "step": 9627
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3878744438084428,
+      "learning_rate": 1.996448891290885e-06,
+      "loss": 0.6687,
+      "step": 9628
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.41314089042012714,
+      "learning_rate": 1.9939270869940074e-06,
+      "loss": 0.667,
+      "step": 9629
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.39074982670452063,
+      "learning_rate": 1.991406792844717e-06,
+      "loss": 0.6722,
+      "step": 9630
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3568962100804789,
+      "learning_rate": 1.9888880090543907e-06,
+      "loss": 0.6402,
+      "step": 9631
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.42020368678226805,
+      "learning_rate": 1.9863707358342753e-06,
+      "loss": 0.6322,
+      "step": 9632
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.41678897480414306,
+      "learning_rate": 1.9838549733954913e-06,
+      "loss": 0.6319,
+      "step": 9633
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.41972341885189096,
+      "learning_rate": 1.9813407219490346e-06,
+      "loss": 0.6724,
+      "step": 9634
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4071051801480256,
+      "learning_rate": 1.978827981705769e-06,
+      "loss": 0.6522,
+      "step": 9635
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3557977283660353,
+      "learning_rate": 1.976316752876435e-06,
+      "loss": 0.6198,
+      "step": 9636
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38648894845501924,
+      "learning_rate": 1.9738070356716534e-06,
+      "loss": 0.6126,
+      "step": 9637
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4285272211252968,
+      "learning_rate": 1.9712988303018977e-06,
+      "loss": 0.5813,
+      "step": 9638
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.39184430238189083,
+      "learning_rate": 1.968792136977542e-06,
+      "loss": 0.6253,
+      "step": 9639
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4061836261200838,
+      "learning_rate": 1.9662869559088116e-06,
+      "loss": 0.6264,
+      "step": 9640
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.40547680502686134,
+      "learning_rate": 1.963783287305814e-06,
+      "loss": 0.6158,
+      "step": 9641
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4053308319297897,
+      "learning_rate": 1.961281131378534e-06,
+      "loss": 0.6515,
+      "step": 9642
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.42854490395473216,
+      "learning_rate": 1.9587804883368156e-06,
+      "loss": 0.6661,
+      "step": 9643
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4295568356950292,
+      "learning_rate": 1.9562813583903904e-06,
+      "loss": 0.6194,
+      "step": 9644
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.37367597175404343,
+      "learning_rate": 1.953783741748856e-06,
+      "loss": 0.6145,
+      "step": 9645
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.36689038616045055,
+      "learning_rate": 1.9512876386216838e-06,
+      "loss": 0.6235,
+      "step": 9646
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.43119009006195713,
+      "learning_rate": 1.9487930492182206e-06,
+      "loss": 0.6918,
+      "step": 9647
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.39337500182142104,
+      "learning_rate": 1.9462999737476897e-06,
+      "loss": 0.5822,
+      "step": 9648
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.37825499342215646,
+      "learning_rate": 1.9438084124191703e-06,
+      "loss": 0.6162,
+      "step": 9649
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3902256374251976,
+      "learning_rate": 1.9413183654416336e-06,
+      "loss": 0.6388,
+      "step": 9650
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3924717607734598,
+      "learning_rate": 1.9388298330239207e-06,
+      "loss": 0.5909,
+      "step": 9651
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.34702029023201847,
+      "learning_rate": 1.93634281537473e-06,
+      "loss": 0.5919,
+      "step": 9652
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.40635050533674555,
+      "learning_rate": 1.9338573127026583e-06,
+      "loss": 0.6026,
+      "step": 9653
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4301614873342409,
+      "learning_rate": 1.9313733252161526e-06,
+      "loss": 0.6237,
+      "step": 9654
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.35745426747983117,
+      "learning_rate": 1.928890853123546e-06,
+      "loss": 0.636,
+      "step": 9655
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.40460745303474854,
+      "learning_rate": 1.926409896633037e-06,
+      "loss": 0.6367,
+      "step": 9656
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3791250945599518,
+      "learning_rate": 1.923930455952707e-06,
+      "loss": 0.6288,
+      "step": 9657
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.41201330757050386,
+      "learning_rate": 1.921452531290495e-06,
+      "loss": 0.6425,
+      "step": 9658
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4405875480075388,
+      "learning_rate": 1.9189761228542282e-06,
+      "loss": 0.6709,
+      "step": 9659
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.39153345885737967,
+      "learning_rate": 1.9165012308515953e-06,
+      "loss": 0.6207,
+      "step": 9660
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3762955564942473,
+      "learning_rate": 1.914027855490166e-06,
+      "loss": 0.6603,
+      "step": 9661
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.40243077077990197,
+      "learning_rate": 1.9115559969773812e-06,
+      "loss": 0.6704,
+      "step": 9662
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.36799521977994876,
+      "learning_rate": 1.909085655520546e-06,
+      "loss": 0.6668,
+      "step": 9663
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.44715500682627135,
+      "learning_rate": 1.9066168313268485e-06,
+      "loss": 0.6505,
+      "step": 9664
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38016550550167744,
+      "learning_rate": 1.9041495246033492e-06,
+      "loss": 0.6306,
+      "step": 9665
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38517601418769865,
+      "learning_rate": 1.9016837355569696e-06,
+      "loss": 0.6259,
+      "step": 9666
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3971651244761935,
+      "learning_rate": 1.8992194643945194e-06,
+      "loss": 0.6326,
+      "step": 9667
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.39176766323024,
+      "learning_rate": 1.8967567113226782e-06,
+      "loss": 0.6422,
+      "step": 9668
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.39633664439257904,
+      "learning_rate": 1.8942954765479826e-06,
+      "loss": 0.6308,
+      "step": 9669
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38763244633471805,
+      "learning_rate": 1.8918357602768611e-06,
+      "loss": 0.6431,
+      "step": 9670
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38860287160762025,
+      "learning_rate": 1.8893775627156062e-06,
+      "loss": 0.6147,
+      "step": 9671
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4057268099143647,
+      "learning_rate": 1.8869208840703823e-06,
+      "loss": 0.64,
+      "step": 9672
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4098717203739361,
+      "learning_rate": 1.8844657245472287e-06,
+      "loss": 0.6467,
+      "step": 9673
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.357345569240905,
+      "learning_rate": 1.882012084352054e-06,
+      "loss": 0.5934,
+      "step": 9674
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4211158123531343,
+      "learning_rate": 1.879559963690647e-06,
+      "loss": 0.6607,
+      "step": 9675
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.39207577216501344,
+      "learning_rate": 1.8771093627686653e-06,
+      "loss": 0.6457,
+      "step": 9676
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4480822771646595,
+      "learning_rate": 1.8746602817916315e-06,
+      "loss": 0.6358,
+      "step": 9677
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.38249665335161026,
+      "learning_rate": 1.8722127209649476e-06,
+      "loss": 0.6338,
+      "step": 9678
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.36903463285982196,
+      "learning_rate": 1.869766680493892e-06,
+      "loss": 0.6553,
+      "step": 9679
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3784114284230691,
+      "learning_rate": 1.867322160583609e-06,
+      "loss": 0.6401,
+      "step": 9680
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.42132747664681625,
+      "learning_rate": 1.8648791614391172e-06,
+      "loss": 0.6604,
+      "step": 9681
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.40696801316998604,
+      "learning_rate": 1.8624376832653125e-06,
+      "loss": 0.6411,
+      "step": 9682
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.347287756771572,
+      "learning_rate": 1.859997726266951e-06,
+      "loss": 0.5839,
+      "step": 9683
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.38989312539182014,
+      "learning_rate": 1.8575592906486717e-06,
+      "loss": 0.6574,
+      "step": 9684
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.35480440656403794,
+      "learning_rate": 1.8551223766149884e-06,
+      "loss": 0.6518,
+      "step": 9685
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.37336675371627187,
+      "learning_rate": 1.8526869843702754e-06,
+      "loss": 0.6709,
+      "step": 9686
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3642804201817032,
+      "learning_rate": 1.8502531141187829e-06,
+      "loss": 0.5947,
+      "step": 9687
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.44078527822510805,
+      "learning_rate": 1.847820766064652e-06,
+      "loss": 0.6918,
+      "step": 9688
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.46310467897077795,
+      "learning_rate": 1.8453899404118658e-06,
+      "loss": 0.6618,
+      "step": 9689
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.41924787695455407,
+      "learning_rate": 1.8429606373642994e-06,
+      "loss": 0.6424,
+      "step": 9690
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4220060578635592,
+      "learning_rate": 1.8405328571257008e-06,
+      "loss": 0.6154,
+      "step": 9691
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4088573549163705,
+      "learning_rate": 1.8381065998996783e-06,
+      "loss": 0.6195,
+      "step": 9692
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4190277155116841,
+      "learning_rate": 1.8356818658897203e-06,
+      "loss": 0.6771,
+      "step": 9693
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3786329933894872,
+      "learning_rate": 1.833258655299186e-06,
+      "loss": 0.6416,
+      "step": 9694
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.43812147995734657,
+      "learning_rate": 1.8308369683313089e-06,
+      "loss": 0.6503,
+      "step": 9695
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3968765398465015,
+      "learning_rate": 1.828416805189197e-06,
+      "loss": 0.5987,
+      "step": 9696
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3808428024212573,
+      "learning_rate": 1.8259981660758176e-06,
+      "loss": 0.6666,
+      "step": 9697
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3731771318555426,
+      "learning_rate": 1.8235810511940232e-06,
+      "loss": 0.6259,
+      "step": 9698
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.40350997565202773,
+      "learning_rate": 1.821165460746539e-06,
+      "loss": 0.6464,
+      "step": 9699
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3744810255958858,
+      "learning_rate": 1.8187513949359493e-06,
+      "loss": 0.6552,
+      "step": 9700
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.39445258887841333,
+      "learning_rate": 1.8163388539647164e-06,
+      "loss": 0.6187,
+      "step": 9701
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.40817987737549594,
+      "learning_rate": 1.8139278380351943e-06,
+      "loss": 0.6537,
+      "step": 9702
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.44025773139694685,
+      "learning_rate": 1.8115183473495767e-06,
+      "loss": 0.6944,
+      "step": 9703
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4169320278653233,
+      "learning_rate": 1.8091103821099486e-06,
+      "loss": 0.6405,
+      "step": 9704
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3907508373628756,
+      "learning_rate": 1.8067039425182663e-06,
+      "loss": 0.6524,
+      "step": 9705
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3948446521713379,
+      "learning_rate": 1.8042990287763507e-06,
+      "loss": 0.6597,
+      "step": 9706
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4060845837495556,
+      "learning_rate": 1.8018956410859002e-06,
+      "loss": 0.6345,
+      "step": 9707
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.36353342521227866,
+      "learning_rate": 1.7994937796484847e-06,
+      "loss": 0.6419,
+      "step": 9708
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.40359728128608424,
+      "learning_rate": 1.7970934446655452e-06,
+      "loss": 0.6357,
+      "step": 9709
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.41815911322347726,
+      "learning_rate": 1.7946946363383987e-06,
+      "loss": 0.6533,
+      "step": 9710
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.32767706348085135,
+      "learning_rate": 1.792297354868222e-06,
+      "loss": 0.6191,
+      "step": 9711
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.43615422879296245,
+      "learning_rate": 1.7899016004560765e-06,
+      "loss": 0.6535,
+      "step": 9712
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4002958449956589,
+      "learning_rate": 1.7875073733028924e-06,
+      "loss": 0.617,
+      "step": 9713
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.35840017965490173,
+      "learning_rate": 1.7851146736094716e-06,
+      "loss": 0.6201,
+      "step": 9714
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4270304143388718,
+      "learning_rate": 1.782723501576482e-06,
+      "loss": 0.6535,
+      "step": 9715
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.34055465480188174,
+      "learning_rate": 1.780333857404477e-06,
+      "loss": 0.6098,
+      "step": 9716
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4047886292380682,
+      "learning_rate": 1.7779457412938628e-06,
+      "loss": 0.6397,
+      "step": 9717
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.35648132313695613,
+      "learning_rate": 1.7755591534449345e-06,
+      "loss": 0.6524,
+      "step": 9718
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4154520086426924,
+      "learning_rate": 1.7731740940578524e-06,
+      "loss": 0.6236,
+      "step": 9719
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.35921401905921946,
+      "learning_rate": 1.7707905633326428e-06,
+      "loss": 0.6411,
+      "step": 9720
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3972639529634764,
+      "learning_rate": 1.7684085614692104e-06,
+      "loss": 0.6377,
+      "step": 9721
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4278367623643684,
+      "learning_rate": 1.7660280886673397e-06,
+      "loss": 0.6467,
+      "step": 9722
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.35043776938387233,
+      "learning_rate": 1.7636491451266713e-06,
+      "loss": 0.6449,
+      "step": 9723
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3975594881847104,
+      "learning_rate": 1.7612717310467232e-06,
+      "loss": 0.6799,
+      "step": 9724
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 7.143808206551627,
+      "learning_rate": 1.7588958466268912e-06,
+      "loss": 6.5655,
+      "step": 9725
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3855186325831244,
+      "learning_rate": 1.7565214920664341e-06,
+      "loss": 0.6612,
+      "step": 9726
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.47349105044145673,
+      "learning_rate": 1.7541486675644836e-06,
+      "loss": 0.6151,
+      "step": 9727
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.37664310366754994,
+      "learning_rate": 1.7517773733200517e-06,
+      "loss": 0.6197,
+      "step": 9728
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4210092162922295,
+      "learning_rate": 1.74940760953201e-06,
+      "loss": 0.6472,
+      "step": 9729
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.39542625533315445,
+      "learning_rate": 1.7470393763991178e-06,
+      "loss": 0.6403,
+      "step": 9730
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3192999835119231,
+      "learning_rate": 1.744672674119985e-06,
+      "loss": 0.6101,
+      "step": 9731
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.358334407126079,
+      "learning_rate": 1.7423075028931057e-06,
+      "loss": 0.6145,
+      "step": 9732
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.41201308533687947,
+      "learning_rate": 1.739943862916853e-06,
+      "loss": 0.6445,
+      "step": 9733
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3850147786979074,
+      "learning_rate": 1.7375817543894503e-06,
+      "loss": 0.6243,
+      "step": 9734
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.40517884522154624,
+      "learning_rate": 1.7352211775090077e-06,
+      "loss": 0.6517,
+      "step": 9735
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.39362738691135907,
+      "learning_rate": 1.7328621324735139e-06,
+      "loss": 0.636,
+      "step": 9736
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4181406108981985,
+      "learning_rate": 1.7305046194808084e-06,
+      "loss": 0.6305,
+      "step": 9737
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3609941171308245,
+      "learning_rate": 1.728148638728615e-06,
+      "loss": 0.6521,
+      "step": 9738
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.35553272129310665,
+      "learning_rate": 1.7257941904145336e-06,
+      "loss": 0.6158,
+      "step": 9739
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.38036084760122907,
+      "learning_rate": 1.7234412747360196e-06,
+      "loss": 0.6151,
+      "step": 9740
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3998035860826181,
+      "learning_rate": 1.721089891890415e-06,
+      "loss": 0.6147,
+      "step": 9741
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3898360286549096,
+      "learning_rate": 1.7187400420749244e-06,
+      "loss": 0.6268,
+      "step": 9742
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.42366418928277155,
+      "learning_rate": 1.716391725486628e-06,
+      "loss": 0.6137,
+      "step": 9743
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.32386816098965593,
+      "learning_rate": 1.714044942322477e-06,
+      "loss": 0.5853,
+      "step": 9744
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.40489738543937337,
+      "learning_rate": 1.7116996927792961e-06,
+      "loss": 0.6235,
+      "step": 9745
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.38060137821332357,
+      "learning_rate": 1.7093559770537726e-06,
+      "loss": 0.6253,
+      "step": 9746
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3749812151216578,
+      "learning_rate": 1.7070137953424737e-06,
+      "loss": 0.6563,
+      "step": 9747
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3746016123829275,
+      "learning_rate": 1.7046731478418377e-06,
+      "loss": 0.6585,
+      "step": 9748
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3717945275881446,
+      "learning_rate": 1.7023340347481632e-06,
+      "loss": 0.6345,
+      "step": 9749
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3800189253673692,
+      "learning_rate": 1.6999964562576442e-06,
+      "loss": 0.6226,
+      "step": 9750
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4290934999505781,
+      "learning_rate": 1.6976604125663155e-06,
+      "loss": 0.6382,
+      "step": 9751
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3933969556170163,
+      "learning_rate": 1.6953259038701065e-06,
+      "loss": 0.6316,
+      "step": 9752
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3970916757872241,
+      "learning_rate": 1.6929929303648096e-06,
+      "loss": 0.6964,
+      "step": 9753
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3818364086529966,
+      "learning_rate": 1.6906614922460817e-06,
+      "loss": 0.6269,
+      "step": 9754
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.36913749049861233,
+      "learning_rate": 1.6883315897094643e-06,
+      "loss": 0.6203,
+      "step": 9755
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.34076536540439645,
+      "learning_rate": 1.6860032229503609e-06,
+      "loss": 0.6304,
+      "step": 9756
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4019683180464037,
+      "learning_rate": 1.6836763921640487e-06,
+      "loss": 0.6088,
+      "step": 9757
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3272948316654931,
+      "learning_rate": 1.6813510975456782e-06,
+      "loss": 0.6079,
+      "step": 9758
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4205890459440815,
+      "learning_rate": 1.679027339290271e-06,
+      "loss": 0.6378,
+      "step": 9759
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3896758989423687,
+      "learning_rate": 1.6767051175927117e-06,
+      "loss": 0.6422,
+      "step": 9760
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.47131904651161854,
+      "learning_rate": 1.6743844326477643e-06,
+      "loss": 0.6435,
+      "step": 9761
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4037224261852669,
+      "learning_rate": 1.6720652846500661e-06,
+      "loss": 0.6172,
+      "step": 9762
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3696186090615625,
+      "learning_rate": 1.6697476737941132e-06,
+      "loss": 0.6329,
+      "step": 9763
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.36094426862682505,
+      "learning_rate": 1.6674316002742896e-06,
+      "loss": 0.6448,
+      "step": 9764
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.43570252445401775,
+      "learning_rate": 1.6651170642848358e-06,
+      "loss": 0.6335,
+      "step": 9765
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3957782652734812,
+      "learning_rate": 1.6628040660198696e-06,
+      "loss": 0.6276,
+      "step": 9766
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3587243217459271,
+      "learning_rate": 1.660492605673385e-06,
+      "loss": 0.6454,
+      "step": 9767
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.39262140718651034,
+      "learning_rate": 1.6581826834392355e-06,
+      "loss": 0.6425,
+      "step": 9768
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4297534334091787,
+      "learning_rate": 1.655874299511151e-06,
+      "loss": 0.6279,
+      "step": 9769
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.42731877350230657,
+      "learning_rate": 1.653567454082734e-06,
+      "loss": 0.6432,
+      "step": 9770
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.36392208456326597,
+      "learning_rate": 1.6512621473474588e-06,
+      "loss": 0.5876,
+      "step": 9771
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.37669922842954373,
+      "learning_rate": 1.6489583794986685e-06,
+      "loss": 0.6234,
+      "step": 9772
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4264931984161977,
+      "learning_rate": 1.6466561507295774e-06,
+      "loss": 0.6526,
+      "step": 9773
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3732601504103136,
+      "learning_rate": 1.644355461233269e-06,
+      "loss": 0.6159,
+      "step": 9774
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3835196379327116,
+      "learning_rate": 1.6420563112026977e-06,
+      "loss": 0.6196,
+      "step": 9775
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.37353239728436055,
+      "learning_rate": 1.6397587008306982e-06,
+      "loss": 0.6271,
+      "step": 9776
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3396400061319634,
+      "learning_rate": 1.6374626303099562e-06,
+      "loss": 0.621,
+      "step": 9777
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3807681843333217,
+      "learning_rate": 1.6351680998330489e-06,
+      "loss": 0.6282,
+      "step": 9778
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3606303717242928,
+      "learning_rate": 1.632875109592418e-06,
+      "loss": 0.6252,
+      "step": 9779
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 8.430657149147352,
+      "learning_rate": 1.630583659780367e-06,
+      "loss": 6.4761,
+      "step": 9780
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3473797945758247,
+      "learning_rate": 1.6282937505890783e-06,
+      "loss": 0.619,
+      "step": 9781
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3819800744276431,
+      "learning_rate": 1.6260053822106093e-06,
+      "loss": 0.6187,
+      "step": 9782
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.38819072104498165,
+      "learning_rate": 1.6237185548368772e-06,
+      "loss": 0.6157,
+      "step": 9783
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.41554653902439415,
+      "learning_rate": 1.6214332686596756e-06,
+      "loss": 0.6645,
+      "step": 9784
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3651479737886421,
+      "learning_rate": 1.6191495238706688e-06,
+      "loss": 0.6421,
+      "step": 9785
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.35289759528357945,
+      "learning_rate": 1.616867320661395e-06,
+      "loss": 0.6409,
+      "step": 9786
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39514923770660987,
+      "learning_rate": 1.6145866592232607e-06,
+      "loss": 0.6482,
+      "step": 9787
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.43297455419462727,
+      "learning_rate": 1.612307539747535e-06,
+      "loss": 0.6674,
+      "step": 9788
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4092296791196425,
+      "learning_rate": 1.61002996242537e-06,
+      "loss": 0.6436,
+      "step": 9789
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.44889881527961784,
+      "learning_rate": 1.6077539274477816e-06,
+      "loss": 0.626,
+      "step": 9790
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3834906964749485,
+      "learning_rate": 1.6054794350056613e-06,
+      "loss": 0.6629,
+      "step": 9791
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39073209016957244,
+      "learning_rate": 1.6032064852897633e-06,
+      "loss": 0.6636,
+      "step": 9792
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.33239069548975037,
+      "learning_rate": 1.600935078490724e-06,
+      "loss": 0.5961,
+      "step": 9793
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4076837398861143,
+      "learning_rate": 1.5986652147990378e-06,
+      "loss": 0.6478,
+      "step": 9794
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.34541738471219324,
+      "learning_rate": 1.5963968944050745e-06,
+      "loss": 0.6383,
+      "step": 9795
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.42466725272953193,
+      "learning_rate": 1.594130117499082e-06,
+      "loss": 0.6819,
+      "step": 9796
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.36607776199630154,
+      "learning_rate": 1.5918648842711592e-06,
+      "loss": 0.6114,
+      "step": 9797
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.44416529928045995,
+      "learning_rate": 1.5896011949113078e-06,
+      "loss": 0.6564,
+      "step": 9798
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.43450858028589834,
+      "learning_rate": 1.5873390496093643e-06,
+      "loss": 0.6682,
+      "step": 9799
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3591712336995368,
+      "learning_rate": 1.5850784485550573e-06,
+      "loss": 0.6245,
+      "step": 9800
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3963850897324652,
+      "learning_rate": 1.5828193919379819e-06,
+      "loss": 0.633,
+      "step": 9801
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.40794387195556014,
+      "learning_rate": 1.5805618799476042e-06,
+      "loss": 0.623,
+      "step": 9802
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.34836961368856045,
+      "learning_rate": 1.5783059127732547e-06,
+      "loss": 0.5839,
+      "step": 9803
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.43660415126833413,
+      "learning_rate": 1.5760514906041401e-06,
+      "loss": 0.6431,
+      "step": 9804
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.36278192904253753,
+      "learning_rate": 1.573798613629336e-06,
+      "loss": 0.6233,
+      "step": 9805
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3316516053683752,
+      "learning_rate": 1.571547282037791e-06,
+      "loss": 0.6228,
+      "step": 9806
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3793299396759672,
+      "learning_rate": 1.5692974960183204e-06,
+      "loss": 0.6257,
+      "step": 9807
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3666324468846042,
+      "learning_rate": 1.5670492557596073e-06,
+      "loss": 0.6298,
+      "step": 9808
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4314067830047112,
+      "learning_rate": 1.5648025614502117e-06,
+      "loss": 0.6467,
+      "step": 9809
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4013377627920245,
+      "learning_rate": 1.5625574132785649e-06,
+      "loss": 0.6276,
+      "step": 9810
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4312304263867626,
+      "learning_rate": 1.560313811432954e-06,
+      "loss": 0.6657,
+      "step": 9811
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3768368420940759,
+      "learning_rate": 1.558071756101558e-06,
+      "loss": 0.6408,
+      "step": 9812
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4227887258700948,
+      "learning_rate": 1.555831247472417e-06,
+      "loss": 0.6479,
+      "step": 9813
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.38363757605857124,
+      "learning_rate": 1.5535922857334273e-06,
+      "loss": 0.633,
+      "step": 9814
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.37435289705663827,
+      "learning_rate": 1.551354871072377e-06,
+      "loss": 0.6083,
+      "step": 9815
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3901566374112452,
+      "learning_rate": 1.5491190036769154e-06,
+      "loss": 0.6226,
+      "step": 9816
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.36712881916563744,
+      "learning_rate": 1.5468846837345575e-06,
+      "loss": 0.6128,
+      "step": 9817
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39070279382648176,
+      "learning_rate": 1.5446519114326974e-06,
+      "loss": 0.6407,
+      "step": 9818
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.44843519025224077,
+      "learning_rate": 1.54242068695859e-06,
+      "loss": 0.663,
+      "step": 9819
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4587467994318788,
+      "learning_rate": 1.54019101049937e-06,
+      "loss": 0.6666,
+      "step": 9820
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.34029763148534775,
+      "learning_rate": 1.537962882242039e-06,
+      "loss": 0.5983,
+      "step": 9821
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.42156039385866323,
+      "learning_rate": 1.5357363023734606e-06,
+      "loss": 0.6258,
+      "step": 9822
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4264270102187568,
+      "learning_rate": 1.5335112710803812e-06,
+      "loss": 0.6831,
+      "step": 9823
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4330116323760847,
+      "learning_rate": 1.5312877885494114e-06,
+      "loss": 0.6906,
+      "step": 9824
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.42709847892164027,
+      "learning_rate": 1.5290658549670246e-06,
+      "loss": 0.664,
+      "step": 9825
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.37415322589294453,
+      "learning_rate": 1.52684547051958e-06,
+      "loss": 0.6378,
+      "step": 9826
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3617614435212861,
+      "learning_rate": 1.524626635393298e-06,
+      "loss": 0.6217,
+      "step": 9827
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4016331999717277,
+      "learning_rate": 1.5224093497742654e-06,
+      "loss": 0.645,
+      "step": 9828
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4217080069462149,
+      "learning_rate": 1.5201936138484463e-06,
+      "loss": 0.6678,
+      "step": 9829
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.38721177536093454,
+      "learning_rate": 1.5179794278016724e-06,
+      "loss": 0.6533,
+      "step": 9830
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4004777085559381,
+      "learning_rate": 1.5157667918196417e-06,
+      "loss": 0.6247,
+      "step": 9831
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39280362835761296,
+      "learning_rate": 1.5135557060879214e-06,
+      "loss": 0.6338,
+      "step": 9832
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.40066930480203533,
+      "learning_rate": 1.5113461707919652e-06,
+      "loss": 0.6407,
+      "step": 9833
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.642363032296358,
+      "learning_rate": 1.5091381861170762e-06,
+      "loss": 6.5071,
+      "step": 9834
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3908617740780603,
+      "learning_rate": 1.5069317522484328e-06,
+      "loss": 0.6821,
+      "step": 9835
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.37650292868239454,
+      "learning_rate": 1.5047268693710959e-06,
+      "loss": 0.6472,
+      "step": 9836
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.35757073950808976,
+      "learning_rate": 1.5025235376699732e-06,
+      "loss": 0.6125,
+      "step": 9837
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.36898136052863034,
+      "learning_rate": 1.5003217573298657e-06,
+      "loss": 0.6309,
+      "step": 9838
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4349503614101192,
+      "learning_rate": 1.4981215285354278e-06,
+      "loss": 0.6471,
+      "step": 9839
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39107404769555904,
+      "learning_rate": 1.4959228514711944e-06,
+      "loss": 0.6491,
+      "step": 9840
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.40316935922468533,
+      "learning_rate": 1.4937257263215643e-06,
+      "loss": 0.6513,
+      "step": 9841
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.41461975660503547,
+      "learning_rate": 1.4915301532708081e-06,
+      "loss": 0.6503,
+      "step": 9842
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3657273157554142,
+      "learning_rate": 1.489336132503063e-06,
+      "loss": 0.6352,
+      "step": 9843
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39118320416052355,
+      "learning_rate": 1.4871436642023463e-06,
+      "loss": 0.5672,
+      "step": 9844
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3625343931460874,
+      "learning_rate": 1.4849527485525284e-06,
+      "loss": 0.6664,
+      "step": 9845
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4359563175187919,
+      "learning_rate": 1.482763385737358e-06,
+      "loss": 0.6711,
+      "step": 9846
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.34888337006809356,
+      "learning_rate": 1.4805755759404684e-06,
+      "loss": 0.6231,
+      "step": 9847
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.42178223782755886,
+      "learning_rate": 1.4783893193453347e-06,
+      "loss": 0.6388,
+      "step": 9848
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4229291948145514,
+      "learning_rate": 1.4762046161353215e-06,
+      "loss": 0.5967,
+      "step": 9849
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3878339878876316,
+      "learning_rate": 1.4740214664936581e-06,
+      "loss": 0.633,
+      "step": 9850
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.35321989210873506,
+      "learning_rate": 1.4718398706034399e-06,
+      "loss": 0.6054,
+      "step": 9851
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4219449084187321,
+      "learning_rate": 1.4696598286476315e-06,
+      "loss": 0.6167,
+      "step": 9852
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3990857000487051,
+      "learning_rate": 1.467481340809076e-06,
+      "loss": 0.6136,
+      "step": 9853
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4029389146049189,
+      "learning_rate": 1.4653044072704802e-06,
+      "loss": 0.629,
+      "step": 9854
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4088379256446601,
+      "learning_rate": 1.4631290282144205e-06,
+      "loss": 0.621,
+      "step": 9855
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.38282484274885814,
+      "learning_rate": 1.4609552038233422e-06,
+      "loss": 0.6447,
+      "step": 9856
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3334367471174035,
+      "learning_rate": 1.458782934279559e-06,
+      "loss": 0.6081,
+      "step": 9857
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39842353685896303,
+      "learning_rate": 1.4566122197652588e-06,
+      "loss": 0.63,
+      "step": 9858
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3940725064892947,
+      "learning_rate": 1.4544430604625005e-06,
+      "loss": 0.6351,
+      "step": 9859
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.37134925450059636,
+      "learning_rate": 1.4522754565532004e-06,
+      "loss": 0.665,
+      "step": 9860
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3911395069473402,
+      "learning_rate": 1.4501094082191625e-06,
+      "loss": 0.6135,
+      "step": 9861
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3961213650538383,
+      "learning_rate": 1.4479449156420433e-06,
+      "loss": 0.6198,
+      "step": 9862
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.42854377057050547,
+      "learning_rate": 1.44578197900338e-06,
+      "loss": 0.6061,
+      "step": 9863
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3534201840550202,
+      "learning_rate": 1.4436205984845763e-06,
+      "loss": 0.6293,
+      "step": 9864
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.41819878777066,
+      "learning_rate": 1.4414607742668985e-06,
+      "loss": 0.6581,
+      "step": 9865
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.40309604694512585,
+      "learning_rate": 1.4393025065314947e-06,
+      "loss": 0.6473,
+      "step": 9866
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3874910766897578,
+      "learning_rate": 1.4371457954593716e-06,
+      "loss": 0.6368,
+      "step": 9867
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.36152460548593096,
+      "learning_rate": 1.434990641231413e-06,
+      "loss": 0.6216,
+      "step": 9868
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.38307366505434126,
+      "learning_rate": 1.43283704402837e-06,
+      "loss": 0.6211,
+      "step": 9869
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.40675135490402126,
+      "learning_rate": 1.4306850040308629e-06,
+      "loss": 0.6444,
+      "step": 9870
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3870358476532838,
+      "learning_rate": 1.4285345214193757e-06,
+      "loss": 0.5981,
+      "step": 9871
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3557484589787282,
+      "learning_rate": 1.426385596374269e-06,
+      "loss": 0.6044,
+      "step": 9872
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.388909903575861,
+      "learning_rate": 1.4242382290757717e-06,
+      "loss": 0.6695,
+      "step": 9873
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3977652574852269,
+      "learning_rate": 1.42209241970398e-06,
+      "loss": 0.6187,
+      "step": 9874
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4045401062107291,
+      "learning_rate": 1.4199481684388628e-06,
+      "loss": 0.6179,
+      "step": 9875
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.407997833977421,
+      "learning_rate": 1.41780547546025e-06,
+      "loss": 0.667,
+      "step": 9876
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.551830362156057,
+      "learning_rate": 1.4156643409478532e-06,
+      "loss": 6.3666,
+      "step": 9877
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4247277108929582,
+      "learning_rate": 1.4135247650812445e-06,
+      "loss": 0.6551,
+      "step": 9878
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3542047580939844,
+      "learning_rate": 1.4113867480398647e-06,
+      "loss": 0.6437,
+      "step": 9879
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39524973620207965,
+      "learning_rate": 1.409250290003026e-06,
+      "loss": 0.6492,
+      "step": 9880
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3924273126844862,
+      "learning_rate": 1.4071153911499203e-06,
+      "loss": 0.6259,
+      "step": 9881
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4231724984019242,
+      "learning_rate": 1.4049820516595891e-06,
+      "loss": 0.6429,
+      "step": 9882
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.38462993653482713,
+      "learning_rate": 1.402850271710956e-06,
+      "loss": 0.6565,
+      "step": 9883
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.34360069440550706,
+      "learning_rate": 1.4007200514828134e-06,
+      "loss": 0.6193,
+      "step": 9884
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4088484187040867,
+      "learning_rate": 1.398591391153814e-06,
+      "loss": 0.6342,
+      "step": 9885
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.41758015521171826,
+      "learning_rate": 1.3964642909024907e-06,
+      "loss": 0.6632,
+      "step": 9886
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.35493091143438327,
+      "learning_rate": 1.3943387509072404e-06,
+      "loss": 0.654,
+      "step": 9887
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39059356416186575,
+      "learning_rate": 1.3922147713463296e-06,
+      "loss": 0.6316,
+      "step": 9888
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4183607879727658,
+      "learning_rate": 1.3900923523978914e-06,
+      "loss": 0.64,
+      "step": 9889
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4088671508046192,
+      "learning_rate": 1.387971494239937e-06,
+      "loss": 0.6509,
+      "step": 9890
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3534961282928409,
+      "learning_rate": 1.3858521970503325e-06,
+      "loss": 0.6009,
+      "step": 9891
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4243234584125155,
+      "learning_rate": 1.383734461006825e-06,
+      "loss": 0.6614,
+      "step": 9892
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4314804650504709,
+      "learning_rate": 1.3816182862870276e-06,
+      "loss": 0.6631,
+      "step": 9893
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.41743503767434614,
+      "learning_rate": 1.3795036730684119e-06,
+      "loss": 0.6631,
+      "step": 9894
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3431432071928597,
+      "learning_rate": 1.3773906215283428e-06,
+      "loss": 0.6195,
+      "step": 9895
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4011605871408303,
+      "learning_rate": 1.3752791318440296e-06,
+      "loss": 0.6195,
+      "step": 9896
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39882069024251154,
+      "learning_rate": 1.3731692041925614e-06,
+      "loss": 0.6248,
+      "step": 9897
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3754158573375378,
+      "learning_rate": 1.3710608387508995e-06,
+      "loss": 0.6322,
+      "step": 9898
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3888138425686754,
+      "learning_rate": 1.3689540356958642e-06,
+      "loss": 0.6191,
+      "step": 9899
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4369190872802299,
+      "learning_rate": 1.3668487952041543e-06,
+      "loss": 0.6496,
+      "step": 9900
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.398392650961541,
+      "learning_rate": 1.3647451174523307e-06,
+      "loss": 0.6466,
+      "step": 9901
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3870323678773162,
+      "learning_rate": 1.3626430026168302e-06,
+      "loss": 0.6256,
+      "step": 9902
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3930177393867819,
+      "learning_rate": 1.3605424508739518e-06,
+      "loss": 0.6127,
+      "step": 9903
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3895588959743109,
+      "learning_rate": 1.358443462399872e-06,
+      "loss": 0.631,
+      "step": 9904
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.373433462690629,
+      "learning_rate": 1.3563460373706217e-06,
+      "loss": 0.6044,
+      "step": 9905
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3722337651398262,
+      "learning_rate": 1.3542501759621128e-06,
+      "loss": 0.6154,
+      "step": 9906
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3852563723359351,
+      "learning_rate": 1.3521558783501255e-06,
+      "loss": 0.6418,
+      "step": 9907
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4017528683480568,
+      "learning_rate": 1.3500631447102985e-06,
+      "loss": 0.6458,
+      "step": 9908
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3219044438383871,
+      "learning_rate": 1.3479719752181586e-06,
+      "loss": 0.5984,
+      "step": 9909
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3826081356234636,
+      "learning_rate": 1.3458823700490808e-06,
+      "loss": 0.6078,
+      "step": 9910
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.34588067020274,
+      "learning_rate": 1.3437943293783207e-06,
+      "loss": 0.5991,
+      "step": 9911
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3901726845587442,
+      "learning_rate": 1.341707853381e-06,
+      "loss": 0.6018,
+      "step": 9912
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3737397081796392,
+      "learning_rate": 1.3396229422321061e-06,
+      "loss": 0.6244,
+      "step": 9913
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4000003359794074,
+      "learning_rate": 1.3375395961065008e-06,
+      "loss": 0.6238,
+      "step": 9914
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.34255690329844934,
+      "learning_rate": 1.3354578151789087e-06,
+      "loss": 0.6206,
+      "step": 9915
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40762909124383073,
+      "learning_rate": 1.3333775996239306e-06,
+      "loss": 0.6584,
+      "step": 9916
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.36987379440512663,
+      "learning_rate": 1.3312989496160288e-06,
+      "loss": 0.6263,
+      "step": 9917
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.43193640214394485,
+      "learning_rate": 1.329221865329542e-06,
+      "loss": 0.682,
+      "step": 9918
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3504862229935347,
+      "learning_rate": 1.3271463469386658e-06,
+      "loss": 0.6378,
+      "step": 9919
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.34394840578174957,
+      "learning_rate": 1.3250723946174727e-06,
+      "loss": 0.6088,
+      "step": 9920
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3621603468182866,
+      "learning_rate": 1.3230000085399097e-06,
+      "loss": 0.6207,
+      "step": 9921
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4220125604530378,
+      "learning_rate": 1.3209291888797715e-06,
+      "loss": 0.6771,
+      "step": 9922
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.357359024844003,
+      "learning_rate": 1.318859935810748e-06,
+      "loss": 0.6237,
+      "step": 9923
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.397607737195783,
+      "learning_rate": 1.3167922495063823e-06,
+      "loss": 0.6264,
+      "step": 9924
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.36739213474916854,
+      "learning_rate": 1.3147261301400849e-06,
+      "loss": 0.6141,
+      "step": 9925
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.35731572416732493,
+      "learning_rate": 1.3126615778851414e-06,
+      "loss": 0.6095,
+      "step": 9926
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4709558195395916,
+      "learning_rate": 1.3105985929147047e-06,
+      "loss": 0.6222,
+      "step": 9927
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.42359994370468423,
+      "learning_rate": 1.3085371754017917e-06,
+      "loss": 0.6388,
+      "step": 9928
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.44081627640824267,
+      "learning_rate": 1.3064773255192908e-06,
+      "loss": 0.6652,
+      "step": 9929
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3456533821501081,
+      "learning_rate": 1.3044190434399596e-06,
+      "loss": 0.6414,
+      "step": 9930
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.39723232804672765,
+      "learning_rate": 1.3023623293364263e-06,
+      "loss": 0.6549,
+      "step": 9931
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4016698320017435,
+      "learning_rate": 1.3003071833811864e-06,
+      "loss": 0.62,
+      "step": 9932
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.38260189363615243,
+      "learning_rate": 1.2982536057465955e-06,
+      "loss": 0.6393,
+      "step": 9933
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.37796737420308085,
+      "learning_rate": 1.2962015966048891e-06,
+      "loss": 0.6194,
+      "step": 9934
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3907532590751704,
+      "learning_rate": 1.2941511561281695e-06,
+      "loss": 0.66,
+      "step": 9935
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3913085451980148,
+      "learning_rate": 1.2921022844883968e-06,
+      "loss": 0.6259,
+      "step": 9936
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.405893393037976,
+      "learning_rate": 1.2900549818574138e-06,
+      "loss": 0.6558,
+      "step": 9937
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3913528274673078,
+      "learning_rate": 1.288009248406925e-06,
+      "loss": 0.6369,
+      "step": 9938
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40504913668003056,
+      "learning_rate": 1.2859650843085026e-06,
+      "loss": 0.6259,
+      "step": 9939
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4134251309756753,
+      "learning_rate": 1.2839224897335844e-06,
+      "loss": 0.6425,
+      "step": 9940
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3851747406236995,
+      "learning_rate": 1.2818814648534895e-06,
+      "loss": 0.631,
+      "step": 9941
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3609391315289276,
+      "learning_rate": 1.279842009839385e-06,
+      "loss": 0.6121,
+      "step": 9942
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.33116175377686513,
+      "learning_rate": 1.2778041248623253e-06,
+      "loss": 0.6196,
+      "step": 9943
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40337953587793457,
+      "learning_rate": 1.2757678100932203e-06,
+      "loss": 0.6572,
+      "step": 9944
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.35631069262091863,
+      "learning_rate": 1.273733065702858e-06,
+      "loss": 0.6223,
+      "step": 9945
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4161626361784662,
+      "learning_rate": 1.271699891861886e-06,
+      "loss": 0.6912,
+      "step": 9946
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.39977116718119154,
+      "learning_rate": 1.2696682887408307e-06,
+      "loss": 0.6567,
+      "step": 9947
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.39754668617212535,
+      "learning_rate": 1.267638256510071e-06,
+      "loss": 0.6578,
+      "step": 9948
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.41470181969941616,
+      "learning_rate": 1.2656097953398706e-06,
+      "loss": 0.6731,
+      "step": 9949
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3629889409806882,
+      "learning_rate": 1.263582905400349e-06,
+      "loss": 0.6131,
+      "step": 9950
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.38540726699450667,
+      "learning_rate": 1.2615575868615015e-06,
+      "loss": 0.6707,
+      "step": 9951
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4108613657324931,
+      "learning_rate": 1.2595338398931922e-06,
+      "loss": 0.631,
+      "step": 9952
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3559844497651993,
+      "learning_rate": 1.257511664665143e-06,
+      "loss": 0.605,
+      "step": 9953
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3901988760849526,
+      "learning_rate": 1.2554910613469563e-06,
+      "loss": 0.6313,
+      "step": 9954
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40342599346767016,
+      "learning_rate": 1.2534720301080983e-06,
+      "loss": 0.6296,
+      "step": 9955
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3845312249442706,
+      "learning_rate": 1.251454571117896e-06,
+      "loss": 0.6192,
+      "step": 9956
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4090529387289829,
+      "learning_rate": 1.2494386845455608e-06,
+      "loss": 0.637,
+      "step": 9957
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.38298506537958604,
+      "learning_rate": 1.2474243705601596e-06,
+      "loss": 0.6431,
+      "step": 9958
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3437634891325096,
+      "learning_rate": 1.2454116293306263e-06,
+      "loss": 0.6068,
+      "step": 9959
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40021549564186853,
+      "learning_rate": 1.24340046102577e-06,
+      "loss": 0.6261,
+      "step": 9960
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3749412894902845,
+      "learning_rate": 1.241390865814267e-06,
+      "loss": 0.6512,
+      "step": 9961
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.39358001685515737,
+      "learning_rate": 1.239382843864656e-06,
+      "loss": 0.5947,
+      "step": 9962
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.38369487488065646,
+      "learning_rate": 1.2373763953453466e-06,
+      "loss": 0.6699,
+      "step": 9963
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4156730451717542,
+      "learning_rate": 1.235371520424622e-06,
+      "loss": 0.6702,
+      "step": 9964
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.36118642254851163,
+      "learning_rate": 1.2333682192706253e-06,
+      "loss": 0.6152,
+      "step": 9965
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.37095676266754074,
+      "learning_rate": 1.2313664920513736e-06,
+      "loss": 0.6169,
+      "step": 9966
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.39148444763610685,
+      "learning_rate": 1.2293663389347454e-06,
+      "loss": 0.6135,
+      "step": 9967
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.38681818549890296,
+      "learning_rate": 1.2273677600884915e-06,
+      "loss": 0.639,
+      "step": 9968
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3934084488089764,
+      "learning_rate": 1.2253707556802374e-06,
+      "loss": 0.6142,
+      "step": 9969
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.36745925966507487,
+      "learning_rate": 1.223375325877456e-06,
+      "loss": 0.6061,
+      "step": 9970
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3952040343232011,
+      "learning_rate": 1.2213814708475136e-06,
+      "loss": 0.642,
+      "step": 9971
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3479671737958768,
+      "learning_rate": 1.2193891907576317e-06,
+      "loss": 0.5991,
+      "step": 9972
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3749234147988364,
+      "learning_rate": 1.2173984857748944e-06,
+      "loss": 0.6269,
+      "step": 9973
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.424730765900598,
+      "learning_rate": 1.2154093560662616e-06,
+      "loss": 0.6626,
+      "step": 9974
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40158532593508994,
+      "learning_rate": 1.2134218017985622e-06,
+      "loss": 0.6239,
+      "step": 9975
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.38494851033232835,
+      "learning_rate": 1.2114358231384871e-06,
+      "loss": 0.6478,
+      "step": 9976
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40105673773062755,
+      "learning_rate": 1.2094514202525964e-06,
+      "loss": 0.6428,
+      "step": 9977
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3930660618772241,
+      "learning_rate": 1.2074685933073239e-06,
+      "loss": 0.6378,
+      "step": 9978
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.37017972139117356,
+      "learning_rate": 1.2054873424689627e-06,
+      "loss": 0.6575,
+      "step": 9979
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.37088642330342725,
+      "learning_rate": 1.2035076679036806e-06,
+      "loss": 0.6584,
+      "step": 9980
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3520645614147492,
+      "learning_rate": 1.2015295697775132e-06,
+      "loss": 0.6233,
+      "step": 9981
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4222244597488562,
+      "learning_rate": 1.1995530482563567e-06,
+      "loss": 0.5727,
+      "step": 9982
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4132788135215018,
+      "learning_rate": 1.1975781035059786e-06,
+      "loss": 0.6759,
+      "step": 9983
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.37137940532485614,
+      "learning_rate": 1.1956047356920197e-06,
+      "loss": 0.6126,
+      "step": 9984
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.41149372154509123,
+      "learning_rate": 1.1936329449799788e-06,
+      "loss": 0.6074,
+      "step": 9985
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.36755739307535557,
+      "learning_rate": 1.1916627315352348e-06,
+      "loss": 0.6455,
+      "step": 9986
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3921493896956426,
+      "learning_rate": 1.1896940955230196e-06,
+      "loss": 0.6489,
+      "step": 9987
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.42944938088940154,
+      "learning_rate": 1.1877270371084437e-06,
+      "loss": 0.6246,
+      "step": 9988
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40765186543514315,
+      "learning_rate": 1.1857615564564862e-06,
+      "loss": 0.6329,
+      "step": 9989
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.41914146018517595,
+      "learning_rate": 1.1837976537319795e-06,
+      "loss": 0.6538,
+      "step": 9990
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.35805320214954645,
+      "learning_rate": 1.1818353290996388e-06,
+      "loss": 0.6191,
+      "step": 9991
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.41409635556999325,
+      "learning_rate": 1.1798745827240476e-06,
+      "loss": 0.693,
+      "step": 9992
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.38577356777357386,
+      "learning_rate": 1.1779154147696414e-06,
+      "loss": 0.6626,
+      "step": 9993
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4279859749211337,
+      "learning_rate": 1.1759578254007398e-06,
+      "loss": 0.6468,
+      "step": 9994
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3600260238763218,
+      "learning_rate": 1.1740018147815246e-06,
+      "loss": 0.6217,
+      "step": 9995
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4161316957835854,
+      "learning_rate": 1.1720473830760381e-06,
+      "loss": 0.6448,
+      "step": 9996
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.39528442591268875,
+      "learning_rate": 1.170094530448198e-06,
+      "loss": 0.6457,
+      "step": 9997
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3632311694075283,
+      "learning_rate": 1.1681432570617913e-06,
+      "loss": 0.63,
+      "step": 9998
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.38171092045669797,
+      "learning_rate": 1.1661935630804643e-06,
+      "loss": 0.6244,
+      "step": 9999
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40407625158024096,
+      "learning_rate": 1.1642454486677425e-06,
+      "loss": 0.6505,
+      "step": 10000
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.43830434509338495,
+      "learning_rate": 1.1622989139870033e-06,
+      "loss": 0.6489,
+      "step": 10001
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.38948223347805333,
+      "learning_rate": 1.1603539592015033e-06,
+      "loss": 0.6361,
+      "step": 10002
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.43105549335550786,
+      "learning_rate": 1.1584105844743654e-06,
+      "loss": 0.6522,
+      "step": 10003
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.42781993385511047,
+      "learning_rate": 1.1564687899685788e-06,
+      "loss": 0.6568,
+      "step": 10004
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.37636258769010894,
+      "learning_rate": 1.1545285758469916e-06,
+      "loss": 0.6521,
+      "step": 10005
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.41292311857261776,
+      "learning_rate": 1.1525899422723396e-06,
+      "loss": 0.6512,
+      "step": 10006
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40876492019297234,
+      "learning_rate": 1.1506528894072045e-06,
+      "loss": 0.6274,
+      "step": 10007
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.41377990088769007,
+      "learning_rate": 1.148717417414047e-06,
+      "loss": 0.6385,
+      "step": 10008
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40733780124258434,
+      "learning_rate": 1.1467835264551952e-06,
+      "loss": 0.6642,
+      "step": 10009
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3299800018387916,
+      "learning_rate": 1.144851216692837e-06,
+      "loss": 0.6276,
+      "step": 10010
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.32554238350128073,
+      "learning_rate": 1.1429204882890389e-06,
+      "loss": 0.5828,
+      "step": 10011
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.36652293420962034,
+      "learning_rate": 1.140991341405724e-06,
+      "loss": 0.6244,
+      "step": 10012
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3719305253728938,
+      "learning_rate": 1.1390637762046875e-06,
+      "loss": 0.6136,
+      "step": 10013
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3400840209735348,
+      "learning_rate": 1.1371377928475958e-06,
+      "loss": 0.6398,
+      "step": 10014
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3515667498768049,
+      "learning_rate": 1.1352133914959794e-06,
+      "loss": 0.6102,
+      "step": 10015
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.43720113279286144,
+      "learning_rate": 1.133290572311232e-06,
+      "loss": 0.6385,
+      "step": 10016
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.35374030850958393,
+      "learning_rate": 1.1313693354546173e-06,
+      "loss": 0.6189,
+      "step": 10017
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.0828423517947554,
+      "learning_rate": 1.1294496810872713e-06,
+      "loss": 6.5631,
+      "step": 10018
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3792600176528777,
+      "learning_rate": 1.127531609370185e-06,
+      "loss": 0.6206,
+      "step": 10019
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3810854085175788,
+      "learning_rate": 1.1256151204642362e-06,
+      "loss": 0.612,
+      "step": 10020
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.42501327655777876,
+      "learning_rate": 1.1237002145301523e-06,
+      "loss": 0.6169,
+      "step": 10021
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3421606457601855,
+      "learning_rate": 1.1217868917285313e-06,
+      "loss": 0.6034,
+      "step": 10022
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.392676509387696,
+      "learning_rate": 1.1198751522198492e-06,
+      "loss": 0.6516,
+      "step": 10023
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.33605775263008353,
+      "learning_rate": 1.1179649961644335e-06,
+      "loss": 0.6403,
+      "step": 10024
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40342303610274677,
+      "learning_rate": 1.116056423722489e-06,
+      "loss": 0.669,
+      "step": 10025
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.39916320749774853,
+      "learning_rate": 1.1141494350540883e-06,
+      "loss": 0.6705,
+      "step": 10026
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3912234961067644,
+      "learning_rate": 1.1122440303191652e-06,
+      "loss": 0.6149,
+      "step": 10027
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38882769706526105,
+      "learning_rate": 1.1103402096775229e-06,
+      "loss": 0.6161,
+      "step": 10028
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4121240395543229,
+      "learning_rate": 1.1084379732888406e-06,
+      "loss": 0.6037,
+      "step": 10029
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.3971045877351886,
+      "learning_rate": 1.106537321312644e-06,
+      "loss": 6.5691,
+      "step": 10030
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4078447867010599,
+      "learning_rate": 1.1046382539083456e-06,
+      "loss": 0.6342,
+      "step": 10031
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38282713743875457,
+      "learning_rate": 1.1027407712352224e-06,
+      "loss": 0.6096,
+      "step": 10032
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3827099713745411,
+      "learning_rate": 1.1008448734524002e-06,
+      "loss": 0.6485,
+      "step": 10033
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4057693650797822,
+      "learning_rate": 1.098950560718901e-06,
+      "loss": 0.6442,
+      "step": 10034
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38979047202286166,
+      "learning_rate": 1.0970578331935867e-06,
+      "loss": 0.6172,
+      "step": 10035
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3126242850635281,
+      "learning_rate": 1.0951666910352055e-06,
+      "loss": 0.6127,
+      "step": 10036
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4465078349014708,
+      "learning_rate": 1.0932771344023619e-06,
+      "loss": 0.6467,
+      "step": 10037
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4357314149907904,
+      "learning_rate": 1.0913891634535334e-06,
+      "loss": 0.6997,
+      "step": 10038
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4346772629577251,
+      "learning_rate": 1.089502778347058e-06,
+      "loss": 0.6647,
+      "step": 10039
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.37314168421408683,
+      "learning_rate": 1.0876179792411468e-06,
+      "loss": 0.6162,
+      "step": 10040
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4019501221587872,
+      "learning_rate": 1.0857347662938734e-06,
+      "loss": 0.6106,
+      "step": 10041
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40254247801127685,
+      "learning_rate": 1.0838531396631845e-06,
+      "loss": 0.6388,
+      "step": 10042
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.382061969591007,
+      "learning_rate": 1.08197309950689e-06,
+      "loss": 0.6612,
+      "step": 10043
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3593290109667208,
+      "learning_rate": 1.0800946459826611e-06,
+      "loss": 0.6227,
+      "step": 10044
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3802567655849616,
+      "learning_rate": 1.0782177792480453e-06,
+      "loss": 0.6814,
+      "step": 10045
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.39123299508474824,
+      "learning_rate": 1.076342499460452e-06,
+      "loss": 0.6533,
+      "step": 10046
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.35234539092767725,
+      "learning_rate": 1.0744688067771603e-06,
+      "loss": 0.6519,
+      "step": 10047
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40197370579679903,
+      "learning_rate": 1.0725967013553129e-06,
+      "loss": 0.5851,
+      "step": 10048
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4323223921694183,
+      "learning_rate": 1.0707261833519244e-06,
+      "loss": 0.6669,
+      "step": 10049
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3842951976738702,
+      "learning_rate": 1.0688572529238694e-06,
+      "loss": 0.61,
+      "step": 10050
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4116017786751627,
+      "learning_rate": 1.0669899102278914e-06,
+      "loss": 0.6164,
+      "step": 10051
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.311213425239236,
+      "learning_rate": 1.0651241554206072e-06,
+      "loss": 6.4263,
+      "step": 10052
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38991095303623546,
+      "learning_rate": 1.0632599886584872e-06,
+      "loss": 0.6237,
+      "step": 10053
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.34437020839396376,
+      "learning_rate": 1.0613974100978885e-06,
+      "loss": 0.6112,
+      "step": 10054
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3804779271233301,
+      "learning_rate": 1.0595364198950132e-06,
+      "loss": 0.6591,
+      "step": 10055
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4073107054760933,
+      "learning_rate": 1.0576770182059447e-06,
+      "loss": 0.6281,
+      "step": 10056
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3906386530087661,
+      "learning_rate": 1.05581920518663e-06,
+      "loss": 0.6443,
+      "step": 10057
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3757578893769233,
+      "learning_rate": 1.0539629809928776e-06,
+      "loss": 0.6175,
+      "step": 10058
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3598028409947895,
+      "learning_rate": 1.0521083457803693e-06,
+      "loss": 0.6116,
+      "step": 10059
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.36820442301448225,
+      "learning_rate": 1.05025529970465e-06,
+      "loss": 0.6108,
+      "step": 10060
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.37794449224433374,
+      "learning_rate": 1.048403842921133e-06,
+      "loss": 0.6403,
+      "step": 10061
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.39510004186527586,
+      "learning_rate": 1.0465539755851006e-06,
+      "loss": 0.6853,
+      "step": 10062
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40674413307901824,
+      "learning_rate": 1.0447056978516956e-06,
+      "loss": 0.6241,
+      "step": 10063
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4065569539620178,
+      "learning_rate": 1.0428590098759317e-06,
+      "loss": 0.6284,
+      "step": 10064
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3711089032865731,
+      "learning_rate": 1.041013911812685e-06,
+      "loss": 0.612,
+      "step": 10065
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3817314667199769,
+      "learning_rate": 1.0391704038167115e-06,
+      "loss": 0.6282,
+      "step": 10066
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4477034019509935,
+      "learning_rate": 1.037328486042608e-06,
+      "loss": 0.6388,
+      "step": 10067
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3920263103856876,
+      "learning_rate": 1.0354881586448683e-06,
+      "loss": 0.6805,
+      "step": 10068
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4116936290999652,
+      "learning_rate": 1.0336494217778358e-06,
+      "loss": 0.6112,
+      "step": 10069
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3956435892122169,
+      "learning_rate": 1.0318122755957182e-06,
+      "loss": 0.6331,
+      "step": 10070
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3770877756213979,
+      "learning_rate": 1.0299767202525968e-06,
+      "loss": 0.6351,
+      "step": 10071
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38840238050464004,
+      "learning_rate": 1.0281427559024214e-06,
+      "loss": 0.6391,
+      "step": 10072
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.39517438025372786,
+      "learning_rate": 1.0263103826989961e-06,
+      "loss": 0.6299,
+      "step": 10073
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.37179704404814296,
+      "learning_rate": 1.0244796007960066e-06,
+      "loss": 0.615,
+      "step": 10074
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.385324725941045,
+      "learning_rate": 1.0226504103469947e-06,
+      "loss": 0.6278,
+      "step": 10075
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.42297517699532733,
+      "learning_rate": 1.0208228115053753e-06,
+      "loss": 0.6656,
+      "step": 10076
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.44574078264090794,
+      "learning_rate": 1.018996804424428e-06,
+      "loss": 0.6136,
+      "step": 10077
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40390085792486224,
+      "learning_rate": 1.0171723892572927e-06,
+      "loss": 0.657,
+      "step": 10078
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40363593534888953,
+      "learning_rate": 1.0153495661569845e-06,
+      "loss": 0.6044,
+      "step": 10079
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.41101235668756775,
+      "learning_rate": 1.0135283352763836e-06,
+      "loss": 0.6563,
+      "step": 10080
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4093941692022909,
+      "learning_rate": 1.0117086967682233e-06,
+      "loss": 0.6314,
+      "step": 10081
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.34775952962510515,
+      "learning_rate": 1.009890650785128e-06,
+      "loss": 0.6218,
+      "step": 10082
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.37893048308227073,
+      "learning_rate": 1.0080741974795717e-06,
+      "loss": 0.6215,
+      "step": 10083
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.41394347778346874,
+      "learning_rate": 1.006259337003892e-06,
+      "loss": 0.6456,
+      "step": 10084
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.39895254442353756,
+      "learning_rate": 1.004446069510303e-06,
+      "loss": 0.6067,
+      "step": 10085
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3763051369484957,
+      "learning_rate": 1.0026343951508855e-06,
+      "loss": 0.6112,
+      "step": 10086
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.33703514693894054,
+      "learning_rate": 1.0008243140775752e-06,
+      "loss": 0.6191,
+      "step": 10087
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3507774182555757,
+      "learning_rate": 9.990158264421846e-07,
+      "loss": 0.6175,
+      "step": 10088
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38779955125336857,
+      "learning_rate": 9.972089323963896e-07,
+      "loss": 0.6358,
+      "step": 10089
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4164707977047797,
+      "learning_rate": 9.954036320917293e-07,
+      "loss": 0.6461,
+      "step": 10090
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.396137805660721,
+      "learning_rate": 9.935999256796202e-07,
+      "loss": 0.6452,
+      "step": 10091
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.44836847791069323,
+      "learning_rate": 9.917978133113282e-07,
+      "loss": 0.6589,
+      "step": 10092
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4051493486236226,
+      "learning_rate": 9.899972951379966e-07,
+      "loss": 0.6452,
+      "step": 10093
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.44602628440954833,
+      "learning_rate": 9.881983713106335e-07,
+      "loss": 0.6556,
+      "step": 10094
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3861724475167864,
+      "learning_rate": 9.86401041980114e-07,
+      "loss": 0.639,
+      "step": 10095
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.0989817013503942,
+      "learning_rate": 9.84605307297175e-07,
+      "loss": 6.4516,
+      "step": 10096
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3961274015994455,
+      "learning_rate": 9.828111674124274e-07,
+      "loss": 0.6224,
+      "step": 10097
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.33808483844001513,
+      "learning_rate": 9.810186224763375e-07,
+      "loss": 0.6275,
+      "step": 10098
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.37242889779771804,
+      "learning_rate": 9.79227672639247e-07,
+      "loss": 0.6578,
+      "step": 10099
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38166178660450345,
+      "learning_rate": 9.774383180513624e-07,
+      "loss": 0.6672,
+      "step": 10100
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.37861402871739197,
+      "learning_rate": 9.756505588627507e-07,
+      "loss": 0.6297,
+      "step": 10101
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38419574242483634,
+      "learning_rate": 9.738643952233474e-07,
+      "loss": 0.6399,
+      "step": 10102
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4172897696101974,
+      "learning_rate": 9.720798272829657e-07,
+      "loss": 0.6338,
+      "step": 10103
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3774020938871532,
+      "learning_rate": 9.702968551912662e-07,
+      "loss": 0.6076,
+      "step": 10104
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4021894008419039,
+      "learning_rate": 9.685154790977868e-07,
+      "loss": 0.6182,
+      "step": 10105
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40211160486485187,
+      "learning_rate": 9.667356991519327e-07,
+      "loss": 0.6098,
+      "step": 10106
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4087120913916888,
+      "learning_rate": 9.649575155029667e-07,
+      "loss": 0.6303,
+      "step": 10107
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3835935167789014,
+      "learning_rate": 9.631809283000248e-07,
+      "loss": 0.5915,
+      "step": 10108
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.42600873509140647,
+      "learning_rate": 9.614059376921081e-07,
+      "loss": 0.6149,
+      "step": 10109
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3635703201511786,
+      "learning_rate": 9.59632543828084e-07,
+      "loss": 0.619,
+      "step": 10110
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.41482467274924556,
+      "learning_rate": 9.578607468566848e-07,
+      "loss": 0.6183,
+      "step": 10111
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3719199523245953,
+      "learning_rate": 9.560905469265048e-07,
+      "loss": 0.6497,
+      "step": 10112
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4209803709673983,
+      "learning_rate": 9.54321944186012e-07,
+      "loss": 0.6025,
+      "step": 10113
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4535833562971813,
+      "learning_rate": 9.525549387835386e-07,
+      "loss": 0.6754,
+      "step": 10114
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3455951731816933,
+      "learning_rate": 9.507895308672777e-07,
+      "loss": 0.6137,
+      "step": 10115
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3845017938643843,
+      "learning_rate": 9.490257205852882e-07,
+      "loss": 0.6016,
+      "step": 10116
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40539582167749794,
+      "learning_rate": 9.472635080855119e-07,
+      "loss": 0.6527,
+      "step": 10117
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.30027267409716746,
+      "learning_rate": 9.455028935157306e-07,
+      "loss": 0.6097,
+      "step": 10118
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.42117748887697337,
+      "learning_rate": 9.437438770236107e-07,
+      "loss": 0.6689,
+      "step": 10119
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.43143099100186605,
+      "learning_rate": 9.419864587566785e-07,
+      "loss": 0.5732,
+      "step": 10120
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3813216765838099,
+      "learning_rate": 9.40230638862325e-07,
+      "loss": 0.6353,
+      "step": 10121
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.37898651896931546,
+      "learning_rate": 9.384764174878081e-07,
+      "loss": 0.6114,
+      "step": 10122
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4000095458641561,
+      "learning_rate": 9.367237947802543e-07,
+      "loss": 0.683,
+      "step": 10123
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.35839300877923064,
+      "learning_rate": 9.34972770886653e-07,
+      "loss": 0.6584,
+      "step": 10124
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4214546969251535,
+      "learning_rate": 9.33223345953862e-07,
+      "loss": 0.6028,
+      "step": 10125
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3776647585658098,
+      "learning_rate": 9.314755201286041e-07,
+      "loss": 0.6579,
+      "step": 10126
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3776226304368366,
+      "learning_rate": 9.29729293557462e-07,
+      "loss": 0.6258,
+      "step": 10127
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.35377489894738506,
+      "learning_rate": 9.279846663868941e-07,
+      "loss": 0.6066,
+      "step": 10128
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4131190116952077,
+      "learning_rate": 9.262416387632167e-07,
+      "loss": 0.5996,
+      "step": 10129
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3882729127332048,
+      "learning_rate": 9.245002108326195e-07,
+      "loss": 0.6188,
+      "step": 10130
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.41152363103101003,
+      "learning_rate": 9.227603827411546e-07,
+      "loss": 0.627,
+      "step": 10131
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.42720523404370686,
+      "learning_rate": 9.210221546347365e-07,
+      "loss": 0.6233,
+      "step": 10132
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.40343643771016224,
+      "learning_rate": 9.19285526659146e-07,
+      "loss": 0.6081,
+      "step": 10133
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.36077781798981523,
+      "learning_rate": 9.175504989600382e-07,
+      "loss": 0.6076,
+      "step": 10134
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.365358241129841,
+      "learning_rate": 9.158170716829207e-07,
+      "loss": 0.6304,
+      "step": 10135
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38181705616236983,
+      "learning_rate": 9.140852449731752e-07,
+      "loss": 0.6355,
+      "step": 10136
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4075558782122992,
+      "learning_rate": 9.123550189760566e-07,
+      "loss": 0.6363,
+      "step": 10137
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4244603016965118,
+      "learning_rate": 9.106263938366666e-07,
+      "loss": 0.6218,
+      "step": 10138
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4168196332684923,
+      "learning_rate": 9.088993696999892e-07,
+      "loss": 0.6423,
+      "step": 10139
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38209521571520927,
+      "learning_rate": 9.071739467108665e-07,
+      "loss": 0.5998,
+      "step": 10140
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.33096710325058015,
+      "learning_rate": 9.054501250140024e-07,
+      "loss": 0.6252,
+      "step": 10141
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3757923230804419,
+      "learning_rate": 9.037279047539793e-07,
+      "loss": 0.6118,
+      "step": 10142
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.42761791432122387,
+      "learning_rate": 9.020072860752305e-07,
+      "loss": 0.6727,
+      "step": 10143
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3913677348701402,
+      "learning_rate": 9.002882691220693e-07,
+      "loss": 0.6346,
+      "step": 10144
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4002555235738563,
+      "learning_rate": 8.98570854038665e-07,
+      "loss": 0.6525,
+      "step": 10145
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.2369118129037053,
+      "learning_rate": 8.968550409690536e-07,
+      "loss": 6.4276,
+      "step": 10146
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4125049869439548,
+      "learning_rate": 8.951408300571373e-07,
+      "loss": 0.656,
+      "step": 10147
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3562758589330267,
+      "learning_rate": 8.934282214466905e-07,
+      "loss": 0.5779,
+      "step": 10148
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4369824599413742,
+      "learning_rate": 8.917172152813402e-07,
+      "loss": 0.6701,
+      "step": 10149
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.39716356105011696,
+      "learning_rate": 8.900078117045897e-07,
+      "loss": 0.6179,
+      "step": 10150
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3531647131634872,
+      "learning_rate": 8.883000108598083e-07,
+      "loss": 0.6151,
+      "step": 10151
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4053851672195615,
+      "learning_rate": 8.865938128902218e-07,
+      "loss": 0.611,
+      "step": 10152
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4295487942180297,
+      "learning_rate": 8.84889217938929e-07,
+      "loss": 0.6753,
+      "step": 10153
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.35806417461413836,
+      "learning_rate": 8.831862261488955e-07,
+      "loss": 0.6287,
+      "step": 10154
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.42525835635616815,
+      "learning_rate": 8.814848376629426e-07,
+      "loss": 0.5809,
+      "step": 10155
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.36570538280568504,
+      "learning_rate": 8.797850526237673e-07,
+      "loss": 0.5913,
+      "step": 10156
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.37159653192571107,
+      "learning_rate": 8.780868711739287e-07,
+      "loss": 0.6346,
+      "step": 10157
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38166343756625964,
+      "learning_rate": 8.763902934558488e-07,
+      "loss": 0.6166,
+      "step": 10158
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.44501117871321033,
+      "learning_rate": 8.746953196118224e-07,
+      "loss": 0.6177,
+      "step": 10159
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.41524352561742484,
+      "learning_rate": 8.730019497840025e-07,
+      "loss": 0.6332,
+      "step": 10160
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3895812857128292,
+      "learning_rate": 8.71310184114409e-07,
+      "loss": 0.6665,
+      "step": 10161
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4597830482440918,
+      "learning_rate": 8.696200227449259e-07,
+      "loss": 0.6939,
+      "step": 10162
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.35279990043981463,
+      "learning_rate": 8.679314658173133e-07,
+      "loss": 0.6132,
+      "step": 10163
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38868619657707143,
+      "learning_rate": 8.662445134731757e-07,
+      "loss": 0.6009,
+      "step": 10164
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3879605421825949,
+      "learning_rate": 8.645591658540086e-07,
+      "loss": 0.6258,
+      "step": 10165
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.37563818553819517,
+      "learning_rate": 8.628754231011527e-07,
+      "loss": 0.63,
+      "step": 10166
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.37063717853721234,
+      "learning_rate": 8.611932853558236e-07,
+      "loss": 0.6235,
+      "step": 10167
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.40911026683609036,
+      "learning_rate": 8.595127527591018e-07,
+      "loss": 0.6491,
+      "step": 10168
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4099872743573709,
+      "learning_rate": 8.578338254519258e-07,
+      "loss": 0.6146,
+      "step": 10169
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.374461523089038,
+      "learning_rate": 8.561565035751096e-07,
+      "loss": 0.6195,
+      "step": 10170
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.41159542057875176,
+      "learning_rate": 8.544807872693295e-07,
+      "loss": 0.6302,
+      "step": 10171
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.372198896419156,
+      "learning_rate": 8.52806676675122e-07,
+      "loss": 0.6266,
+      "step": 10172
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.313297654284193,
+      "learning_rate": 8.511341719328947e-07,
+      "loss": 6.3721,
+      "step": 10173
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3965738332756004,
+      "learning_rate": 8.4946327318292e-07,
+      "loss": 0.6321,
+      "step": 10174
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.398455845416617,
+      "learning_rate": 8.477939805653324e-07,
+      "loss": 0.6543,
+      "step": 10175
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.40880510711004436,
+      "learning_rate": 8.46126294220131e-07,
+      "loss": 0.601,
+      "step": 10176
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4008800624519407,
+      "learning_rate": 8.444602142871905e-07,
+      "loss": 0.6399,
+      "step": 10177
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.37901721457734855,
+      "learning_rate": 8.427957409062304e-07,
+      "loss": 0.682,
+      "step": 10178
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3701139604938117,
+      "learning_rate": 8.411328742168612e-07,
+      "loss": 0.619,
+      "step": 10179
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3487799232626175,
+      "learning_rate": 8.394716143585379e-07,
+      "loss": 0.6366,
+      "step": 10180
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3933135314404828,
+      "learning_rate": 8.378119614705915e-07,
+      "loss": 0.6318,
+      "step": 10181
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3923371355452029,
+      "learning_rate": 8.361539156922127e-07,
+      "loss": 0.6586,
+      "step": 10182
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38484982231584175,
+      "learning_rate": 8.344974771624614e-07,
+      "loss": 0.6337,
+      "step": 10183
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.43701457907716307,
+      "learning_rate": 8.328426460202599e-07,
+      "loss": 0.6443,
+      "step": 10184
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4024608218499838,
+      "learning_rate": 8.311894224043971e-07,
+      "loss": 0.6001,
+      "step": 10185
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4105228716302048,
+      "learning_rate": 8.295378064535286e-07,
+      "loss": 0.6661,
+      "step": 10186
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3743516261088227,
+      "learning_rate": 8.278877983061706e-07,
+      "loss": 0.6181,
+      "step": 10187
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4410150739866084,
+      "learning_rate": 8.262393981007122e-07,
+      "loss": 0.6592,
+      "step": 10188
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3741392918536046,
+      "learning_rate": 8.245926059753983e-07,
+      "loss": 0.6331,
+      "step": 10189
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4114149227766348,
+      "learning_rate": 8.229474220683431e-07,
+      "loss": 0.6332,
+      "step": 10190
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.43293074843495577,
+      "learning_rate": 8.213038465175294e-07,
+      "loss": 0.6756,
+      "step": 10191
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.39313038445958853,
+      "learning_rate": 8.196618794607958e-07,
+      "loss": 0.5959,
+      "step": 10192
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3539619550548981,
+      "learning_rate": 8.180215210358589e-07,
+      "loss": 0.6586,
+      "step": 10193
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.35960856099435945,
+      "learning_rate": 8.163827713802952e-07,
+      "loss": 0.6493,
+      "step": 10194
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3997332307175675,
+      "learning_rate": 8.147456306315349e-07,
+      "loss": 0.6437,
+      "step": 10195
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3926892421408394,
+      "learning_rate": 8.131100989268903e-07,
+      "loss": 0.6017,
+      "step": 10196
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.41425764613794613,
+      "learning_rate": 8.114761764035317e-07,
+      "loss": 0.6234,
+      "step": 10197
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.39445079445203335,
+      "learning_rate": 8.098438631984917e-07,
+      "loss": 0.6373,
+      "step": 10198
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4221000344789147,
+      "learning_rate": 8.082131594486698e-07,
+      "loss": 0.6587,
+      "step": 10199
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.34364873551485464,
+      "learning_rate": 8.065840652908319e-07,
+      "loss": 0.5971,
+      "step": 10200
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3976633276163423,
+      "learning_rate": 8.049565808616089e-07,
+      "loss": 0.6282,
+      "step": 10201
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.36229792495082036,
+      "learning_rate": 8.033307062974982e-07,
+      "loss": 0.6083,
+      "step": 10202
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3860331696124334,
+      "learning_rate": 8.017064417348552e-07,
+      "loss": 0.6141,
+      "step": 10203
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7376206669252492,
+      "learning_rate": 8.000837873099065e-07,
+      "loss": 0.637,
+      "step": 10204
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.3919090768465374,
+      "learning_rate": 7.984627431587455e-07,
+      "loss": 6.4532,
+      "step": 10205
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3848844850728495,
+      "learning_rate": 7.968433094173211e-07,
+      "loss": 0.616,
+      "step": 10206
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3745939452990824,
+      "learning_rate": 7.952254862214581e-07,
+      "loss": 0.6063,
+      "step": 10207
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3541901525418179,
+      "learning_rate": 7.936092737068435e-07,
+      "loss": 0.6136,
+      "step": 10208
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.44431375452529376,
+      "learning_rate": 7.919946720090199e-07,
+      "loss": 0.6647,
+      "step": 10209
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4127514140618112,
+      "learning_rate": 7.90381681263408e-07,
+      "loss": 0.6212,
+      "step": 10210
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4000502303372294,
+      "learning_rate": 7.887703016052862e-07,
+      "loss": 0.6598,
+      "step": 10211
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4233845156114536,
+      "learning_rate": 7.871605331697907e-07,
+      "loss": 0.6213,
+      "step": 10212
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4288417548945822,
+      "learning_rate": 7.85552376091947e-07,
+      "loss": 0.6336,
+      "step": 10213
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38254994932908315,
+      "learning_rate": 7.839458305066161e-07,
+      "loss": 0.617,
+      "step": 10214
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3872375490972777,
+      "learning_rate": 7.823408965485413e-07,
+      "loss": 0.6582,
+      "step": 10215
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3813091621106285,
+      "learning_rate": 7.80737574352326e-07,
+      "loss": 0.6516,
+      "step": 10216
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.389880391236278,
+      "learning_rate": 7.791358640524405e-07,
+      "loss": 0.6546,
+      "step": 10217
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38106507526505945,
+      "learning_rate": 7.775357657832172e-07,
+      "loss": 0.6318,
+      "step": 10218
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.410824247643555,
+      "learning_rate": 7.759372796788534e-07,
+      "loss": 0.6973,
+      "step": 10219
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.43038991892609796,
+      "learning_rate": 7.743404058734127e-07,
+      "loss": 0.5935,
+      "step": 10220
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3968296589372148,
+      "learning_rate": 7.727451445008238e-07,
+      "loss": 0.681,
+      "step": 10221
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.37539402740239797,
+      "learning_rate": 7.711514956948818e-07,
+      "loss": 0.6246,
+      "step": 10222
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38950937185656187,
+      "learning_rate": 7.695594595892398e-07,
+      "loss": 0.5686,
+      "step": 10223
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7163028077095035,
+      "learning_rate": 7.679690363174197e-07,
+      "loss": 6.5536,
+      "step": 10224
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3913787915582634,
+      "learning_rate": 7.663802260128128e-07,
+      "loss": 0.6385,
+      "step": 10225
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.33758735916716337,
+      "learning_rate": 7.647930288086636e-07,
+      "loss": 0.6288,
+      "step": 10226
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.41763549205717754,
+      "learning_rate": 7.632074448380966e-07,
+      "loss": 0.6484,
+      "step": 10227
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3756901684400432,
+      "learning_rate": 7.6162347423409e-07,
+      "loss": 0.6193,
+      "step": 10228
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3738476724570728,
+      "learning_rate": 7.600411171294841e-07,
+      "loss": 0.6116,
+      "step": 10229
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3668020454472716,
+      "learning_rate": 7.584603736569974e-07,
+      "loss": 0.6198,
+      "step": 10230
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.405245186928302,
+      "learning_rate": 7.568812439492013e-07,
+      "loss": 0.6223,
+      "step": 10231
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.37890377670778236,
+      "learning_rate": 7.553037281385344e-07,
+      "loss": 0.6296,
+      "step": 10232
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4025644084179944,
+      "learning_rate": 7.537278263573022e-07,
+      "loss": 0.6249,
+      "step": 10233
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.41255905729713704,
+      "learning_rate": 7.521535387376722e-07,
+      "loss": 0.6297,
+      "step": 10234
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.42123280766451376,
+      "learning_rate": 7.505808654116809e-07,
+      "loss": 0.6869,
+      "step": 10235
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3717689783486814,
+      "learning_rate": 7.49009806511225e-07,
+      "loss": 0.6519,
+      "step": 10236
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3893887531127034,
+      "learning_rate": 7.474403621680682e-07,
+      "loss": 0.6557,
+      "step": 10237
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.37718100893697576,
+      "learning_rate": 7.458725325138338e-07,
+      "loss": 0.6285,
+      "step": 10238
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3639653956984706,
+      "learning_rate": 7.443063176800192e-07,
+      "loss": 0.6571,
+      "step": 10239
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.080555444275317,
+      "learning_rate": 7.42741717797979e-07,
+      "loss": 6.3148,
+      "step": 10240
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.38957928413945964,
+      "learning_rate": 7.411787329989329e-07,
+      "loss": 0.6426,
+      "step": 10241
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3714949047394426,
+      "learning_rate": 7.396173634139692e-07,
+      "loss": 0.6382,
+      "step": 10242
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.422928502888512,
+      "learning_rate": 7.380576091740366e-07,
+      "loss": 0.6322,
+      "step": 10243
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.36055705970919244,
+      "learning_rate": 7.364994704099482e-07,
+      "loss": 0.6269,
+      "step": 10244
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3882949474872563,
+      "learning_rate": 7.349429472523862e-07,
+      "loss": 0.6285,
+      "step": 10245
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3773581730815767,
+      "learning_rate": 7.333880398318926e-07,
+      "loss": 0.6214,
+      "step": 10246
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.37434503905403116,
+      "learning_rate": 7.318347482788723e-07,
+      "loss": 0.6062,
+      "step": 10247
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.34806230151277573,
+      "learning_rate": 7.302830727236077e-07,
+      "loss": 0.637,
+      "step": 10248
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.34949463057277774,
+      "learning_rate": 7.287330132962255e-07,
+      "loss": 0.6165,
+      "step": 10249
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4002052009496352,
+      "learning_rate": 7.271845701267333e-07,
+      "loss": 0.6147,
+      "step": 10250
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4005251297320391,
+      "learning_rate": 7.256377433449979e-07,
+      "loss": 0.6341,
+      "step": 10251
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.384893046027381,
+      "learning_rate": 7.240925330807447e-07,
+      "loss": 0.6417,
+      "step": 10252
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.42074048954460225,
+      "learning_rate": 7.225489394635698e-07,
+      "loss": 0.6554,
+      "step": 10253
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4059020498089369,
+      "learning_rate": 7.210069626229365e-07,
+      "loss": 0.6271,
+      "step": 10254
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.38874807481022977,
+      "learning_rate": 7.194666026881658e-07,
+      "loss": 0.6458,
+      "step": 10255
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.38053467794560963,
+      "learning_rate": 7.179278597884476e-07,
+      "loss": 0.6148,
+      "step": 10256
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.36710269676509344,
+      "learning_rate": 7.163907340528297e-07,
+      "loss": 0.6336,
+      "step": 10257
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.36468766009870507,
+      "learning_rate": 7.148552256102337e-07,
+      "loss": 0.6383,
+      "step": 10258
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4104246380899912,
+      "learning_rate": 7.133213345894407e-07,
+      "loss": 0.6282,
+      "step": 10259
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3858815056974222,
+      "learning_rate": 7.117890611190948e-07,
+      "loss": 0.6253,
+      "step": 10260
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.37356692636397815,
+      "learning_rate": 7.102584053277017e-07,
+      "loss": 0.6067,
+      "step": 10261
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3668462058845492,
+      "learning_rate": 7.087293673436457e-07,
+      "loss": 0.6376,
+      "step": 10262
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.40407403064686487,
+      "learning_rate": 7.072019472951575e-07,
+      "loss": 0.6254,
+      "step": 10263
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4210952500982653,
+      "learning_rate": 7.056761453103412e-07,
+      "loss": 0.6637,
+      "step": 10264
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3807319992376673,
+      "learning_rate": 7.041519615171677e-07,
+      "loss": 0.6112,
+      "step": 10265
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3787703654087501,
+      "learning_rate": 7.026293960434615e-07,
+      "loss": 0.6285,
+      "step": 10266
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4494397241228169,
+      "learning_rate": 7.011084490169251e-07,
+      "loss": 0.6982,
+      "step": 10267
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3961978421164319,
+      "learning_rate": 6.995891205651162e-07,
+      "loss": 0.6114,
+      "step": 10268
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.39587236120341734,
+      "learning_rate": 6.980714108154573e-07,
+      "loss": 0.6542,
+      "step": 10269
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.37930007133089355,
+      "learning_rate": 6.965553198952402e-07,
+      "loss": 0.6542,
+      "step": 10270
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.41757664804535044,
+      "learning_rate": 6.950408479316162e-07,
+      "loss": 0.6201,
+      "step": 10271
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.41894677827982635,
+      "learning_rate": 6.935279950515994e-07,
+      "loss": 0.6492,
+      "step": 10272
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3764161923255525,
+      "learning_rate": 6.920167613820728e-07,
+      "loss": 0.6316,
+      "step": 10273
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.41491600454410715,
+      "learning_rate": 6.905071470497859e-07,
+      "loss": 0.6447,
+      "step": 10274
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3720186872728736,
+      "learning_rate": 6.8899915218134e-07,
+      "loss": 0.643,
+      "step": 10275
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4207784016556494,
+      "learning_rate": 6.874927769032181e-07,
+      "loss": 0.6398,
+      "step": 10276
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.421358918765337,
+      "learning_rate": 6.859880213417502e-07,
+      "loss": 0.6444,
+      "step": 10277
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.40315883999283086,
+      "learning_rate": 6.844848856231423e-07,
+      "loss": 0.6406,
+      "step": 10278
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.38173487734866607,
+      "learning_rate": 6.829833698734623e-07,
+      "loss": 0.6535,
+      "step": 10279
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.40617765371156883,
+      "learning_rate": 6.814834742186361e-07,
+      "loss": 0.5854,
+      "step": 10280
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4372998127287107,
+      "learning_rate": 6.799851987844585e-07,
+      "loss": 0.6192,
+      "step": 10281
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.4867747385807886,
+      "learning_rate": 6.784885436965915e-07,
+      "loss": 6.5241,
+      "step": 10282
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3423150432021085,
+      "learning_rate": 6.769935090805546e-07,
+      "loss": 0.5715,
+      "step": 10283
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4089938070737715,
+      "learning_rate": 6.755000950617363e-07,
+      "loss": 0.6207,
+      "step": 10284
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4131527715459369,
+      "learning_rate": 6.740083017653875e-07,
+      "loss": 0.5572,
+      "step": 10285
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.43346399754110504,
+      "learning_rate": 6.725181293166217e-07,
+      "loss": 0.6182,
+      "step": 10286
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.36753174281322476,
+      "learning_rate": 6.710295778404185e-07,
+      "loss": 0.6191,
+      "step": 10287
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.41427519742061925,
+      "learning_rate": 6.695426474616251e-07,
+      "loss": 0.6613,
+      "step": 10288
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.419979510237073,
+      "learning_rate": 6.680573383049371e-07,
+      "loss": 0.663,
+      "step": 10289
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.39754936324504403,
+      "learning_rate": 6.665736504949394e-07,
+      "loss": 0.6345,
+      "step": 10290
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.36934392050492815,
+      "learning_rate": 6.65091584156059e-07,
+      "loss": 0.6256,
+      "step": 10291
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3876399406187358,
+      "learning_rate": 6.636111394125966e-07,
+      "loss": 0.6279,
+      "step": 10292
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.35469940561087554,
+      "learning_rate": 6.621323163887172e-07,
+      "loss": 0.6132,
+      "step": 10293
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.39169437373366117,
+      "learning_rate": 6.606551152084439e-07,
+      "loss": 0.6184,
+      "step": 10294
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5136666463960259,
+      "learning_rate": 6.591795359956665e-07,
+      "loss": 0.7087,
+      "step": 10295
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.38380364935664335,
+      "learning_rate": 6.577055788741505e-07,
+      "loss": 0.5897,
+      "step": 10296
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4181196468880247,
+      "learning_rate": 6.562332439675034e-07,
+      "loss": 0.64,
+      "step": 10297
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3867533672914167,
+      "learning_rate": 6.547625313992134e-07,
+      "loss": 0.6373,
+      "step": 10298
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.38497263110559055,
+      "learning_rate": 6.532934412926306e-07,
+      "loss": 0.6426,
+      "step": 10299
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.35758285156836744,
+      "learning_rate": 6.518259737709587e-07,
+      "loss": 0.5985,
+      "step": 10300
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.38571735817183495,
+      "learning_rate": 6.50360128957277e-07,
+      "loss": 0.6215,
+      "step": 10301
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4007006262774587,
+      "learning_rate": 6.488959069745204e-07,
+      "loss": 0.6681,
+      "step": 10302
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.439748372716922,
+      "learning_rate": 6.474333079454953e-07,
+      "loss": 0.6479,
+      "step": 10303
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3808268299065139,
+      "learning_rate": 6.459723319928657e-07,
+      "loss": 0.6672,
+      "step": 10304
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4082132649451248,
+      "learning_rate": 6.445129792391668e-07,
+      "loss": 0.6324,
+      "step": 10305
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4085850595164411,
+      "learning_rate": 6.430552498067854e-07,
+      "loss": 0.6363,
+      "step": 10306
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3318804683534536,
+      "learning_rate": 6.415991438179836e-07,
+      "loss": 0.5888,
+      "step": 10307
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3552471170481964,
+      "learning_rate": 6.401446613948836e-07,
+      "loss": 0.6026,
+      "step": 10308
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4306361833100345,
+      "learning_rate": 6.386918026594657e-07,
+      "loss": 0.6659,
+      "step": 10309
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3970829162265682,
+      "learning_rate": 6.372405677335902e-07,
+      "loss": 0.6527,
+      "step": 10310
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.396614763444236,
+      "learning_rate": 6.357909567389597e-07,
+      "loss": 0.6405,
+      "step": 10311
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3527006390151098,
+      "learning_rate": 6.34342969797157e-07,
+      "loss": 0.6382,
+      "step": 10312
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.40637965993351327,
+      "learning_rate": 6.328966070296227e-07,
+      "loss": 0.6314,
+      "step": 10313
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4121973333744949,
+      "learning_rate": 6.314518685576598e-07,
+      "loss": 0.6616,
+      "step": 10314
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4063035044337926,
+      "learning_rate": 6.300087545024358e-07,
+      "loss": 0.6512,
+      "step": 10315
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3469271701408051,
+      "learning_rate": 6.28567264984985e-07,
+      "loss": 0.6182,
+      "step": 10316
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3912761780361173,
+      "learning_rate": 6.27127400126204e-07,
+      "loss": 0.6286,
+      "step": 10317
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3507619050765508,
+      "learning_rate": 6.256891600468496e-07,
+      "loss": 0.6067,
+      "step": 10318
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.37496444253030886,
+      "learning_rate": 6.242525448675496e-07,
+      "loss": 0.6627,
+      "step": 10319
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.364043435566177,
+      "learning_rate": 6.228175547087878e-07,
+      "loss": 0.592,
+      "step": 10320
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.44297776545278394,
+      "learning_rate": 6.213841896909145e-07,
+      "loss": 0.7045,
+      "step": 10321
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4110437584006407,
+      "learning_rate": 6.199524499341492e-07,
+      "loss": 0.634,
+      "step": 10322
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4106772566659597,
+      "learning_rate": 6.185223355585623e-07,
+      "loss": 0.621,
+      "step": 10323
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4243451028396332,
+      "learning_rate": 6.170938466841026e-07,
+      "loss": 0.6547,
+      "step": 10324
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3977227995965551,
+      "learning_rate": 6.156669834305739e-07,
+      "loss": 0.6183,
+      "step": 10325
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4119883540330458,
+      "learning_rate": 6.142417459176431e-07,
+      "loss": 0.6373,
+      "step": 10326
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.41363267465681197,
+      "learning_rate": 6.128181342648476e-07,
+      "loss": 0.6615,
+      "step": 10327
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.44461727512362237,
+      "learning_rate": 6.113961485915809e-07,
+      "loss": 0.6846,
+      "step": 10328
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.36173096536030147,
+      "learning_rate": 6.09975789017101e-07,
+      "loss": 0.5957,
+      "step": 10329
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.5355697787618197,
+      "learning_rate": 6.085570556605369e-07,
+      "loss": 6.3904,
+      "step": 10330
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.42558257988475734,
+      "learning_rate": 6.071399486408713e-07,
+      "loss": 0.6731,
+      "step": 10331
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4754632527576475,
+      "learning_rate": 6.057244680769603e-07,
+      "loss": 0.6904,
+      "step": 10332
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.37823796379640606,
+      "learning_rate": 6.043106140875155e-07,
+      "loss": 0.6094,
+      "step": 10333
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3631480355933463,
+      "learning_rate": 6.028983867911132e-07,
+      "loss": 0.5962,
+      "step": 10334
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.40095202171007627,
+      "learning_rate": 6.014877863061985e-07,
+      "loss": 0.6151,
+      "step": 10335
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.37332595434782434,
+      "learning_rate": 6.000788127510792e-07,
+      "loss": 0.6402,
+      "step": 10336
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.40939464368543704,
+      "learning_rate": 5.986714662439142e-07,
+      "loss": 0.6307,
+      "step": 10337
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.35795841725809785,
+      "learning_rate": 5.972657469027443e-07,
+      "loss": 0.6347,
+      "step": 10338
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4084241047989836,
+      "learning_rate": 5.958616548454688e-07,
+      "loss": 0.6436,
+      "step": 10339
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.39064610074472095,
+      "learning_rate": 5.944591901898377e-07,
+      "loss": 0.6324,
+      "step": 10340
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.38752665776976786,
+      "learning_rate": 5.930583530534772e-07,
+      "loss": 0.6295,
+      "step": 10341
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.39268932918092025,
+      "learning_rate": 5.916591435538798e-07,
+      "loss": 0.6397,
+      "step": 10342
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.35978141654101903,
+      "learning_rate": 5.902615618083874e-07,
+      "loss": 0.6337,
+      "step": 10343
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.45426845081993195,
+      "learning_rate": 5.888656079342169e-07,
+      "loss": 0.6809,
+      "step": 10344
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4076747043977626,
+      "learning_rate": 5.87471282048444e-07,
+      "loss": 0.5733,
+      "step": 10345
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4214168571875972,
+      "learning_rate": 5.860785842680106e-07,
+      "loss": 0.6809,
+      "step": 10346
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3581351972389268,
+      "learning_rate": 5.846875147097231e-07,
+      "loss": 0.6438,
+      "step": 10347
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40337908292158864,
+      "learning_rate": 5.832980734902438e-07,
+      "loss": 0.6687,
+      "step": 10348
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3651537143775743,
+      "learning_rate": 5.819102607261062e-07,
+      "loss": 0.6352,
+      "step": 10349
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3936199236583647,
+      "learning_rate": 5.805240765337039e-07,
+      "loss": 0.645,
+      "step": 10350
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4177975442034799,
+      "learning_rate": 5.791395210292927e-07,
+      "loss": 0.6243,
+      "step": 10351
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40182285188149913,
+      "learning_rate": 5.777565943289975e-07,
+      "loss": 0.6402,
+      "step": 10352
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.37251679494949436,
+      "learning_rate": 5.763752965488012e-07,
+      "loss": 0.6673,
+      "step": 10353
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.45978998316847025,
+      "learning_rate": 5.749956278045488e-07,
+      "loss": 0.6087,
+      "step": 10354
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.37357932573423963,
+      "learning_rate": 5.736175882119544e-07,
+      "loss": 0.654,
+      "step": 10355
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3919953526352825,
+      "learning_rate": 5.722411778865943e-07,
+      "loss": 0.6397,
+      "step": 10356
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.396916765650969,
+      "learning_rate": 5.70866396943901e-07,
+      "loss": 0.6633,
+      "step": 10357
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.42063237820096244,
+      "learning_rate": 5.694932454991798e-07,
+      "loss": 0.6439,
+      "step": 10358
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3831919062432463,
+      "learning_rate": 5.681217236675918e-07,
+      "loss": 0.639,
+      "step": 10359
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3869110220549615,
+      "learning_rate": 5.667518315641673e-07,
+      "loss": 0.6444,
+      "step": 10360
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3944904525351927,
+      "learning_rate": 5.65383569303799e-07,
+      "loss": 0.6425,
+      "step": 10361
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3785508690436136,
+      "learning_rate": 5.640169370012416e-07,
+      "loss": 0.5926,
+      "step": 10362
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4161496971292592,
+      "learning_rate": 5.626519347711079e-07,
+      "loss": 0.6192,
+      "step": 10363
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4204545790127703,
+      "learning_rate": 5.612885627278819e-07,
+      "loss": 0.6761,
+      "step": 10364
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.6743694526289112,
+      "learning_rate": 5.599268209859077e-07,
+      "loss": 6.407,
+      "step": 10365
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4018426561487563,
+      "learning_rate": 5.585667096593961e-07,
+      "loss": 0.6387,
+      "step": 10366
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3789228378198327,
+      "learning_rate": 5.572082288624136e-07,
+      "loss": 0.6179,
+      "step": 10367
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.37761672954818915,
+      "learning_rate": 5.558513787088959e-07,
+      "loss": 0.6392,
+      "step": 10368
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4140110380133274,
+      "learning_rate": 5.544961593126408e-07,
+      "loss": 0.593,
+      "step": 10369
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3965076023668181,
+      "learning_rate": 5.531425707873106e-07,
+      "loss": 0.6523,
+      "step": 10370
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.38768724249405456,
+      "learning_rate": 5.517906132464257e-07,
+      "loss": 0.5948,
+      "step": 10371
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.37833263345783563,
+      "learning_rate": 5.504402868033709e-07,
+      "loss": 0.6046,
+      "step": 10372
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.35961241399147836,
+      "learning_rate": 5.490915915714046e-07,
+      "loss": 0.6261,
+      "step": 10373
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.334147332814688,
+      "learning_rate": 5.477445276636339e-07,
+      "loss": 6.3904,
+      "step": 10374
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3930835925329336,
+      "learning_rate": 5.463990951930376e-07,
+      "loss": 0.6306,
+      "step": 10375
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39559231732388345,
+      "learning_rate": 5.450552942724563e-07,
+      "loss": 0.6325,
+      "step": 10376
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40085648708649285,
+      "learning_rate": 5.43713125014591e-07,
+      "loss": 0.6219,
+      "step": 10377
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.43640709599855093,
+      "learning_rate": 5.423725875320074e-07,
+      "loss": 0.6366,
+      "step": 10378
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.41190521894362375,
+      "learning_rate": 5.410336819371375e-07,
+      "loss": 0.6484,
+      "step": 10379
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39225202766579675,
+      "learning_rate": 5.396964083422718e-07,
+      "loss": 0.6566,
+      "step": 10380
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.35402660971870226,
+      "learning_rate": 5.383607668595669e-07,
+      "loss": 0.6101,
+      "step": 10381
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4045009098737618,
+      "learning_rate": 5.370267576010402e-07,
+      "loss": 0.6541,
+      "step": 10382
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3859561182667666,
+      "learning_rate": 5.356943806785731e-07,
+      "loss": 0.6415,
+      "step": 10383
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.393697088683778,
+      "learning_rate": 5.34363636203914e-07,
+      "loss": 0.6478,
+      "step": 10384
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39590934319469473,
+      "learning_rate": 5.330345242886648e-07,
+      "loss": 0.6305,
+      "step": 10385
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.43372416096263805,
+      "learning_rate": 5.317070450443007e-07,
+      "loss": 0.6084,
+      "step": 10386
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3825353973466847,
+      "learning_rate": 5.303811985821572e-07,
+      "loss": 0.6587,
+      "step": 10387
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4229229003446389,
+      "learning_rate": 5.290569850134275e-07,
+      "loss": 0.6373,
+      "step": 10388
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39853164736109803,
+      "learning_rate": 5.277344044491739e-07,
+      "loss": 0.6409,
+      "step": 10389
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39077589178450284,
+      "learning_rate": 5.26413457000321e-07,
+      "loss": 0.5866,
+      "step": 10390
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.43919628408725253,
+      "learning_rate": 5.250941427776513e-07,
+      "loss": 0.6465,
+      "step": 10391
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3877307357832067,
+      "learning_rate": 5.237764618918162e-07,
+      "loss": 0.667,
+      "step": 10392
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39313527412786226,
+      "learning_rate": 5.224604144533274e-07,
+      "loss": 0.6299,
+      "step": 10393
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40821047763802754,
+      "learning_rate": 5.21146000572561e-07,
+      "loss": 0.623,
+      "step": 10394
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3779836300004648,
+      "learning_rate": 5.198332203597555e-07,
+      "loss": 0.6069,
+      "step": 10395
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.34651627750368563,
+      "learning_rate": 5.185220739250141e-07,
+      "loss": 0.6189,
+      "step": 10396
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3769430015358391,
+      "learning_rate": 5.172125613782952e-07,
+      "loss": 0.6106,
+      "step": 10397
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4116000095484222,
+      "learning_rate": 5.15904682829429e-07,
+      "loss": 0.6471,
+      "step": 10398
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39118835815217257,
+      "learning_rate": 5.145984383881075e-07,
+      "loss": 0.6152,
+      "step": 10399
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.42875541369175185,
+      "learning_rate": 5.132938281638833e-07,
+      "loss": 0.6472,
+      "step": 10400
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.41166775362825797,
+      "learning_rate": 5.119908522661732e-07,
+      "loss": 0.6131,
+      "step": 10401
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.37577767412964075,
+      "learning_rate": 5.10689510804252e-07,
+      "loss": 0.6241,
+      "step": 10402
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40373510048827943,
+      "learning_rate": 5.093898038872658e-07,
+      "loss": 0.6552,
+      "step": 10403
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3738888425560235,
+      "learning_rate": 5.080917316242184e-07,
+      "loss": 0.6373,
+      "step": 10404
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3934862755963068,
+      "learning_rate": 5.067952941239762e-07,
+      "loss": 0.6365,
+      "step": 10405
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39000269435849333,
+      "learning_rate": 5.0550049149527e-07,
+      "loss": 0.6452,
+      "step": 10406
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4085372561739424,
+      "learning_rate": 5.042073238466949e-07,
+      "loss": 0.6457,
+      "step": 10407
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4339045706408991,
+      "learning_rate": 5.029157912867067e-07,
+      "loss": 0.6509,
+      "step": 10408
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5756469064489215,
+      "learning_rate": 5.016258939236252e-07,
+      "loss": 0.6351,
+      "step": 10409
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.36350028759692316,
+      "learning_rate": 5.003376318656328e-07,
+      "loss": 0.6052,
+      "step": 10410
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3994395619451603,
+      "learning_rate": 4.99051005220772e-07,
+      "loss": 0.6072,
+      "step": 10411
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3754511026047546,
+      "learning_rate": 4.97766014096952e-07,
+      "loss": 0.6088,
+      "step": 10412
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40040512257682376,
+      "learning_rate": 4.964826586019444e-07,
+      "loss": 0.6318,
+      "step": 10413
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.38107828276727107,
+      "learning_rate": 4.952009388433809e-07,
+      "loss": 0.6248,
+      "step": 10414
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.39805487341317647,
+      "learning_rate": 4.939208549287622e-07,
+      "loss": 0.6673,
+      "step": 10415
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3902727235341658,
+      "learning_rate": 4.9264240696544e-07,
+      "loss": 0.6459,
+      "step": 10416
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3999607838698995,
+      "learning_rate": 4.913655950606422e-07,
+      "loss": 0.6514,
+      "step": 10417
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3775215445070567,
+      "learning_rate": 4.900904193214495e-07,
+      "loss": 0.6372,
+      "step": 10418
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40596182435677847,
+      "learning_rate": 4.888168798548143e-07,
+      "loss": 0.6427,
+      "step": 10419
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.398370004565016,
+      "learning_rate": 4.875449767675399e-07,
+      "loss": 0.5968,
+      "step": 10420
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3604024454011457,
+      "learning_rate": 4.862747101663079e-07,
+      "loss": 0.6383,
+      "step": 10421
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40320208072919955,
+      "learning_rate": 4.850060801576461e-07,
+      "loss": 0.6329,
+      "step": 10422
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.36846428775651563,
+      "learning_rate": 4.837390868479564e-07,
+      "loss": 0.611,
+      "step": 10423
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3976946954480273,
+      "learning_rate": 4.824737303435023e-07,
+      "loss": 0.6099,
+      "step": 10424
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3802822707511558,
+      "learning_rate": 4.812100107504037e-07,
+      "loss": 0.6308,
+      "step": 10425
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3496146304912795,
+      "learning_rate": 4.799479281746488e-07,
+      "loss": 0.6076,
+      "step": 10426
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.36736441541557335,
+      "learning_rate": 4.786874827220844e-07,
+      "loss": 0.6434,
+      "step": 10427
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3594984355746421,
+      "learning_rate": 4.77428674498428e-07,
+      "loss": 0.5934,
+      "step": 10428
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.43518648325069603,
+      "learning_rate": 4.761715036092507e-07,
+      "loss": 0.6722,
+      "step": 10429
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.541430397407628,
+      "learning_rate": 4.7491597015999257e-07,
+      "loss": 6.399,
+      "step": 10430
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3659179192498137,
+      "learning_rate": 4.7366207425594945e-07,
+      "loss": 0.6395,
+      "step": 10431
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4107172011768415,
+      "learning_rate": 4.724098160022861e-07,
+      "loss": 0.632,
+      "step": 10432
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3640266963161465,
+      "learning_rate": 4.7115919550402953e-07,
+      "loss": 0.6374,
+      "step": 10433
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.42115739677415404,
+      "learning_rate": 4.6991021286606263e-07,
+      "loss": 0.6171,
+      "step": 10434
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.36461087469170944,
+      "learning_rate": 4.686628681931438e-07,
+      "loss": 0.615,
+      "step": 10435
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.2509811796601245,
+      "learning_rate": 4.674171615898826e-07,
+      "loss": 6.4892,
+      "step": 10436
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.38663516078409793,
+      "learning_rate": 4.661730931607511e-07,
+      "loss": 0.6224,
+      "step": 10437
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3577843831486039,
+      "learning_rate": 4.649306630100947e-07,
+      "loss": 0.5847,
+      "step": 10438
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.37003410423125827,
+      "learning_rate": 4.636898712421101e-07,
+      "loss": 0.6214,
+      "step": 10439
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40464924215302894,
+      "learning_rate": 4.624507179608628e-07,
+      "loss": 0.6129,
+      "step": 10440
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4166607744141586,
+      "learning_rate": 4.612132032702765e-07,
+      "loss": 0.6605,
+      "step": 10441
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4127034304440051,
+      "learning_rate": 4.599773272741437e-07,
+      "loss": 0.6135,
+      "step": 10442
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3551265441557955,
+      "learning_rate": 4.587430900761125e-07,
+      "loss": 0.6632,
+      "step": 10443
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3784178698797626,
+      "learning_rate": 4.575104917797002e-07,
+      "loss": 0.6389,
+      "step": 10444
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3715998091211121,
+      "learning_rate": 4.5627953248828183e-07,
+      "loss": 0.5945,
+      "step": 10445
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4771719923240168,
+      "learning_rate": 4.550502123050948e-07,
+      "loss": 0.6715,
+      "step": 10446
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.410815605309743,
+      "learning_rate": 4.538225313332434e-07,
+      "loss": 0.653,
+      "step": 10447
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.42261505386842235,
+      "learning_rate": 4.525964896756874e-07,
+      "loss": 0.6593,
+      "step": 10448
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4496322831337118,
+      "learning_rate": 4.5137208743526005e-07,
+      "loss": 0.6562,
+      "step": 10449
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3440780058313418,
+      "learning_rate": 4.50149324714646e-07,
+      "loss": 0.6003,
+      "step": 10450
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3848257030870218,
+      "learning_rate": 4.489282016163965e-07,
+      "loss": 0.6285,
+      "step": 10451
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3621558579560785,
+      "learning_rate": 4.477087182429274e-07,
+      "loss": 0.6322,
+      "step": 10452
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4021008022175723,
+      "learning_rate": 4.4649087469651685e-07,
+      "loss": 0.6445,
+      "step": 10453
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.33158340614543835,
+      "learning_rate": 4.45274671079301e-07,
+      "loss": 0.5831,
+      "step": 10454
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.7101793486467063,
+      "learning_rate": 4.4406010749328265e-07,
+      "loss": 6.458,
+      "step": 10455
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.34898700714641384,
+      "learning_rate": 4.428471840403248e-07,
+      "loss": 0.5993,
+      "step": 10456
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3856969853353922,
+      "learning_rate": 4.416359008221527e-07,
+      "loss": 0.6058,
+      "step": 10457
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.40042060766054416,
+      "learning_rate": 4.4042625794036063e-07,
+      "loss": 0.6697,
+      "step": 10458
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3476804237546603,
+      "learning_rate": 4.392182554963942e-07,
+      "loss": 0.6032,
+      "step": 10459
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39958033471891546,
+      "learning_rate": 4.3801189359157006e-07,
+      "loss": 0.6413,
+      "step": 10460
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3818401209688763,
+      "learning_rate": 4.3680717232706506e-07,
+      "loss": 0.6043,
+      "step": 10461
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3698405400324938,
+      "learning_rate": 4.3560409180391174e-07,
+      "loss": 0.6381,
+      "step": 10462
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4156313264784705,
+      "learning_rate": 4.3440265212301826e-07,
+      "loss": 0.6531,
+      "step": 10463
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3528593228604324,
+      "learning_rate": 4.332028533851462e-07,
+      "loss": 0.6021,
+      "step": 10464
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3729929531426921,
+      "learning_rate": 4.320046956909174e-07,
+      "loss": 0.6358,
+      "step": 10465
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.42642682777877156,
+      "learning_rate": 4.308081791408247e-07,
+      "loss": 0.6458,
+      "step": 10466
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.454781870666098,
+      "learning_rate": 4.296133038352168e-07,
+      "loss": 0.644,
+      "step": 10467
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4113135038309758,
+      "learning_rate": 4.284200698743024e-07,
+      "loss": 0.6444,
+      "step": 10468
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.7093469085385027,
+      "learning_rate": 4.2722847735816585e-07,
+      "loss": 6.434,
+      "step": 10469
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4304423141438785,
+      "learning_rate": 4.2603852638673393e-07,
+      "loss": 0.6566,
+      "step": 10470
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3508103797144575,
+      "learning_rate": 4.2485021705981343e-07,
+      "loss": 0.6311,
+      "step": 10471
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.42225357944481057,
+      "learning_rate": 4.236635494770669e-07,
+      "loss": 0.654,
+      "step": 10472
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37684585285653843,
+      "learning_rate": 4.2247852373801245e-07,
+      "loss": 0.6163,
+      "step": 10473
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3972598665247479,
+      "learning_rate": 4.2129513994204176e-07,
+      "loss": 0.6358,
+      "step": 10474
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3598475616205287,
+      "learning_rate": 4.20113398188402e-07,
+      "loss": 0.6343,
+      "step": 10475
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.43149622423504286,
+      "learning_rate": 4.1893329857620735e-07,
+      "loss": 0.6554,
+      "step": 10476
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37085109970463676,
+      "learning_rate": 4.1775484120442524e-07,
+      "loss": 0.6315,
+      "step": 10477
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4254954944999017,
+      "learning_rate": 4.16578026171901e-07,
+      "loss": 0.6358,
+      "step": 10478
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.35335855361170476,
+      "learning_rate": 4.1540285357732246e-07,
+      "loss": 0.6208,
+      "step": 10479
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3708622744869388,
+      "learning_rate": 4.142293235192552e-07,
+      "loss": 0.6251,
+      "step": 10480
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3898369543957563,
+      "learning_rate": 4.130574360961226e-07,
+      "loss": 0.6611,
+      "step": 10481
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3835393884570476,
+      "learning_rate": 4.118871914062039e-07,
+      "loss": 0.6608,
+      "step": 10482
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.35892912537409105,
+      "learning_rate": 4.107185895476518e-07,
+      "loss": 0.6238,
+      "step": 10483
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4289668385229235,
+      "learning_rate": 4.0955163061847434e-07,
+      "loss": 0.6235,
+      "step": 10484
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3536697195038224,
+      "learning_rate": 4.083863147165423e-07,
+      "loss": 0.6243,
+      "step": 10485
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.41273341556216037,
+      "learning_rate": 4.072226419395886e-07,
+      "loss": 0.6436,
+      "step": 10486
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3797421405912309,
+      "learning_rate": 4.0606061238521287e-07,
+      "loss": 0.6398,
+      "step": 10487
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.406284269786802,
+      "learning_rate": 4.049002261508683e-07,
+      "loss": 0.6149,
+      "step": 10488
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4275196914662052,
+      "learning_rate": 4.0374148333387484e-07,
+      "loss": 0.6207,
+      "step": 10489
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.400079478525124,
+      "learning_rate": 4.025843840314192e-07,
+      "loss": 0.6426,
+      "step": 10490
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39908221663300214,
+      "learning_rate": 4.014289283405415e-07,
+      "loss": 0.6477,
+      "step": 10491
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4437030949920965,
+      "learning_rate": 4.0027511635815306e-07,
+      "loss": 0.6814,
+      "step": 10492
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37680050958456635,
+      "learning_rate": 3.991229481810188e-07,
+      "loss": 0.6148,
+      "step": 10493
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.36624723937182435,
+      "learning_rate": 3.979724239057703e-07,
+      "loss": 0.6195,
+      "step": 10494
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39875958553843954,
+      "learning_rate": 3.968235436289036e-07,
+      "loss": 0.6153,
+      "step": 10495
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4084125876009756,
+      "learning_rate": 3.9567630744676843e-07,
+      "loss": 0.6448,
+      "step": 10496
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3877451653110974,
+      "learning_rate": 3.9453071545558776e-07,
+      "loss": 0.6026,
+      "step": 10497
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37467296414975143,
+      "learning_rate": 3.93386767751438e-07,
+      "loss": 0.6187,
+      "step": 10498
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.40138378267865765,
+      "learning_rate": 3.922444644302625e-07,
+      "loss": 0.6462,
+      "step": 10499
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3677519780520226,
+      "learning_rate": 3.9110380558786023e-07,
+      "loss": 0.6387,
+      "step": 10500
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39608330164843825,
+      "learning_rate": 3.899647913199056e-07,
+      "loss": 0.6463,
+      "step": 10501
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3957605821987103,
+      "learning_rate": 3.888274217219157e-07,
+      "loss": 0.6582,
+      "step": 10502
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4122824877800954,
+      "learning_rate": 3.876916968892874e-07,
+      "loss": 0.6679,
+      "step": 10503
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37891235838295867,
+      "learning_rate": 3.865576169172713e-07,
+      "loss": 0.6419,
+      "step": 10504
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.36469333170778234,
+      "learning_rate": 3.8542518190098023e-07,
+      "loss": 0.6443,
+      "step": 10505
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3799947675570023,
+      "learning_rate": 3.842943919353914e-07,
+      "loss": 0.5953,
+      "step": 10506
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3533970906901774,
+      "learning_rate": 3.831652471153424e-07,
+      "loss": 0.5901,
+      "step": 10507
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3917714593452391,
+      "learning_rate": 3.82037747535533e-07,
+      "loss": 0.6191,
+      "step": 10508
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3883717077670183,
+      "learning_rate": 3.809118932905231e-07,
+      "loss": 0.6308,
+      "step": 10509
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.42705004420924075,
+      "learning_rate": 3.7978768447473947e-07,
+      "loss": 0.6332,
+      "step": 10510
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.40005424590078453,
+      "learning_rate": 3.7866512118246876e-07,
+      "loss": 0.649,
+      "step": 10511
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3812904085314776,
+      "learning_rate": 3.77544203507858e-07,
+      "loss": 0.6438,
+      "step": 10512
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.32456265952730007,
+      "learning_rate": 3.7642493154491644e-07,
+      "loss": 0.6066,
+      "step": 10513
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4128838859828824,
+      "learning_rate": 3.753073053875178e-07,
+      "loss": 0.6675,
+      "step": 10514
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37126265479596543,
+      "learning_rate": 3.7419132512939384e-07,
+      "loss": 0.5855,
+      "step": 10515
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3741922110590528,
+      "learning_rate": 3.730769908641407e-07,
+      "loss": 0.6206,
+      "step": 10516
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.41059230415121495,
+      "learning_rate": 3.7196430268521486e-07,
+      "loss": 0.6015,
+      "step": 10517
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.43119301212146816,
+      "learning_rate": 3.7085326068594383e-07,
+      "loss": 0.613,
+      "step": 10518
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39367917465381086,
+      "learning_rate": 3.6974386495949977e-07,
+      "loss": 0.6521,
+      "step": 10519
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3534929712140865,
+      "learning_rate": 3.686361155989304e-07,
+      "loss": 0.6396,
+      "step": 10520
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.40460017052091063,
+      "learning_rate": 3.6753001269714597e-07,
+      "loss": 0.6101,
+      "step": 10521
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39120932363148925,
+      "learning_rate": 3.6642555634690547e-07,
+      "loss": 0.6333,
+      "step": 10522
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39982658754054845,
+      "learning_rate": 3.6532274664084157e-07,
+      "loss": 0.6505,
+      "step": 10523
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.40457880191519147,
+      "learning_rate": 3.6422158367144687e-07,
+      "loss": 0.6411,
+      "step": 10524
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4253674805208507,
+      "learning_rate": 3.6312206753107427e-07,
+      "loss": 0.6442,
+      "step": 10525
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3590683023725771,
+      "learning_rate": 3.620241983119388e-07,
+      "loss": 0.628,
+      "step": 10526
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3837050477160448,
+      "learning_rate": 3.6092797610611794e-07,
+      "loss": 0.6278,
+      "step": 10527
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.364684258659451,
+      "learning_rate": 3.598334010055471e-07,
+      "loss": 0.6481,
+      "step": 10528
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37786587546964656,
+      "learning_rate": 3.5874047310203276e-07,
+      "loss": 0.6494,
+      "step": 10529
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.35473178033123953,
+      "learning_rate": 3.5764919248723274e-07,
+      "loss": 0.6283,
+      "step": 10530
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3891309703749942,
+      "learning_rate": 3.565595592526716e-07,
+      "loss": 0.662,
+      "step": 10531
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3490486198285421,
+      "learning_rate": 3.554715734897385e-07,
+      "loss": 0.6036,
+      "step": 10532
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4209708109139575,
+      "learning_rate": 3.543852352896804e-07,
+      "loss": 0.6298,
+      "step": 10533
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39720126713357656,
+      "learning_rate": 3.5330054474360666e-07,
+      "loss": 0.6482,
+      "step": 10534
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3887048922502393,
+      "learning_rate": 3.5221750194248896e-07,
+      "loss": 0.5988,
+      "step": 10535
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4221870289639071,
+      "learning_rate": 3.5113610697716126e-07,
+      "loss": 0.6924,
+      "step": 10536
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.38266135858355693,
+      "learning_rate": 3.5005635993831554e-07,
+      "loss": 0.6891,
+      "step": 10537
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.42658846572876463,
+      "learning_rate": 3.489782609165149e-07,
+      "loss": 0.6404,
+      "step": 10538
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3617864047827957,
+      "learning_rate": 3.479018100021736e-07,
+      "loss": 0.5914,
+      "step": 10539
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.41525193350095335,
+      "learning_rate": 3.4682700728557286e-07,
+      "loss": 0.6011,
+      "step": 10540
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.36396925523663903,
+      "learning_rate": 3.4575385285685824e-07,
+      "loss": 0.6425,
+      "step": 10541
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3389758737291848,
+      "learning_rate": 3.446823468060312e-07,
+      "loss": 0.6267,
+      "step": 10542
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39777402657911576,
+      "learning_rate": 3.436124892229575e-07,
+      "loss": 0.6624,
+      "step": 10543
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3744002397900172,
+      "learning_rate": 3.4254428019736774e-07,
+      "loss": 0.6263,
+      "step": 10544
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.40350017051373094,
+      "learning_rate": 3.414777198188435e-07,
+      "loss": 0.6129,
+      "step": 10545
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.35161948894820244,
+      "learning_rate": 3.4041280817684654e-07,
+      "loss": 0.6329,
+      "step": 10546
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3867437787961342,
+      "learning_rate": 3.393495453606832e-07,
+      "loss": 0.6847,
+      "step": 10547
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.413772066222783,
+      "learning_rate": 3.382879314595289e-07,
+      "loss": 0.6589,
+      "step": 10548
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3644626481906091,
+      "learning_rate": 3.372279665624212e-07,
+      "loss": 0.6279,
+      "step": 10549
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3885898954872711,
+      "learning_rate": 3.3616965075825794e-07,
+      "loss": 0.6239,
+      "step": 10550
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3621997935691625,
+      "learning_rate": 3.3511298413579475e-07,
+      "loss": 0.6075,
+      "step": 10551
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.38614274895160694,
+      "learning_rate": 3.340579667836585e-07,
+      "loss": 0.644,
+      "step": 10552
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3672900075599242,
+      "learning_rate": 3.3300459879033185e-07,
+      "loss": 0.6666,
+      "step": 10553
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.33828014017741825,
+      "learning_rate": 3.3195288024415515e-07,
+      "loss": 0.5894,
+      "step": 10554
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3339117387605445,
+      "learning_rate": 3.309028112333401e-07,
+      "loss": 0.638,
+      "step": 10555
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37634262805071106,
+      "learning_rate": 3.298543918459496e-07,
+      "loss": 0.6244,
+      "step": 10556
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.33843400909268934,
+      "learning_rate": 3.2880762216991544e-07,
+      "loss": 0.651,
+      "step": 10557
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.38407044355356573,
+      "learning_rate": 3.2776250229302977e-07,
+      "loss": 0.6416,
+      "step": 10558
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37448825475772846,
+      "learning_rate": 3.2671903230294677e-07,
+      "loss": 0.6202,
+      "step": 10559
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3952079167209763,
+      "learning_rate": 3.256772122871787e-07,
+      "loss": 0.6132,
+      "step": 10560
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37454173398513607,
+      "learning_rate": 3.2463704233310025e-07,
+      "loss": 0.6594,
+      "step": 10561
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.9636703297880493,
+      "learning_rate": 3.2359852252795256e-07,
+      "loss": 6.6572,
+      "step": 10562
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.42804619819005685,
+      "learning_rate": 3.225616529588349e-07,
+      "loss": 0.6243,
+      "step": 10563
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3706555690703258,
+      "learning_rate": 3.215264337127066e-07,
+      "loss": 0.6252,
+      "step": 10564
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4316394765587597,
+      "learning_rate": 3.204928648763872e-07,
+      "loss": 0.6005,
+      "step": 10565
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.36644140182393076,
+      "learning_rate": 3.1946094653656945e-07,
+      "loss": 0.6048,
+      "step": 10566
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3899575994426499,
+      "learning_rate": 3.184306787797908e-07,
+      "loss": 0.6545,
+      "step": 10567
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.38774627649467436,
+      "learning_rate": 3.1740206169246223e-07,
+      "loss": 0.6119,
+      "step": 10568
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4079889594882756,
+      "learning_rate": 3.1637509536085464e-07,
+      "loss": 0.6445,
+      "step": 10569
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.40104094739494384,
+      "learning_rate": 3.1534977987109474e-07,
+      "loss": 0.6503,
+      "step": 10570
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3857285657915117,
+      "learning_rate": 3.1432611530917367e-07,
+      "loss": 0.641,
+      "step": 10571
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.35003640936797553,
+      "learning_rate": 3.133041017609495e-07,
+      "loss": 0.6196,
+      "step": 10572
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.375951441815167,
+      "learning_rate": 3.1228373931213365e-07,
+      "loss": 0.6337,
+      "step": 10573
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4122817627926913,
+      "learning_rate": 3.112650280483021e-07,
+      "loss": 0.6345,
+      "step": 10574
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.40331382770064284,
+      "learning_rate": 3.1024796805489753e-07,
+      "loss": 0.6455,
+      "step": 10575
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3716793088376285,
+      "learning_rate": 3.092325594172163e-07,
+      "loss": 0.5971,
+      "step": 10576
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.41797412419765567,
+      "learning_rate": 3.0821880222041913e-07,
+      "loss": 0.6148,
+      "step": 10577
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3654054545240594,
+      "learning_rate": 3.072066965495313e-07,
+      "loss": 0.6279,
+      "step": 10578
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.35542523523959146,
+      "learning_rate": 3.061962424894316e-07,
+      "loss": 0.6313,
+      "step": 10579
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4021301484528834,
+      "learning_rate": 3.0518744012487e-07,
+      "loss": 0.6088,
+      "step": 10580
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4272211460540573,
+      "learning_rate": 3.041802895404544e-07,
+      "loss": 0.647,
+      "step": 10581
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.45221777888831294,
+      "learning_rate": 3.031747908206484e-07,
+      "loss": 0.68,
+      "step": 10582
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3927795217042324,
+      "learning_rate": 3.0217094404978886e-07,
+      "loss": 0.6609,
+      "step": 10583
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.40416633369032035,
+      "learning_rate": 3.0116874931205965e-07,
+      "loss": 0.6571,
+      "step": 10584
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3505716793652102,
+      "learning_rate": 3.0016820669152016e-07,
+      "loss": 0.6247,
+      "step": 10585
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4095088060448185,
+      "learning_rate": 2.991693162720788e-07,
+      "loss": 0.6287,
+      "step": 10586
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4264181913521355,
+      "learning_rate": 2.9817207813751527e-07,
+      "loss": 0.6522,
+      "step": 10587
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.42908104881476006,
+      "learning_rate": 2.9717649237146704e-07,
+      "loss": 0.67,
+      "step": 10588
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.425251501787044,
+      "learning_rate": 2.9618255905743187e-07,
+      "loss": 6.4716,
+      "step": 10589
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3615860000133043,
+      "learning_rate": 2.951902782787674e-07,
+      "loss": 0.6164,
+      "step": 10590
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3674231144221192,
+      "learning_rate": 2.9419965011869613e-07,
+      "loss": 0.641,
+      "step": 10591
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.42618587517116674,
+      "learning_rate": 2.932106746603047e-07,
+      "loss": 0.6433,
+      "step": 10592
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.37354164779673765,
+      "learning_rate": 2.922233519865314e-07,
+      "loss": 0.5955,
+      "step": 10593
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.39209482454911976,
+      "learning_rate": 2.9123768218018764e-07,
+      "loss": 0.6249,
+      "step": 10594
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4039015455372435,
+      "learning_rate": 2.9025366532393404e-07,
+      "loss": 0.6449,
+      "step": 10595
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.7054725504268204,
+      "learning_rate": 2.8927130150030457e-07,
+      "loss": 6.5095,
+      "step": 10596
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3889993782535679,
+      "learning_rate": 2.8829059079168445e-07,
+      "loss": 0.6396,
+      "step": 10597
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4206601010187513,
+      "learning_rate": 2.8731153328033e-07,
+      "loss": 0.6086,
+      "step": 10598
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.44039020924940314,
+      "learning_rate": 2.863341290483468e-07,
+      "loss": 0.6577,
+      "step": 10599
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4125957912854783,
+      "learning_rate": 2.853583781777136e-07,
+      "loss": 0.6449,
+      "step": 10600
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3829836132662457,
+      "learning_rate": 2.843842807502628e-07,
+      "loss": 0.6316,
+      "step": 10601
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.39407459778152476,
+      "learning_rate": 2.834118368476935e-07,
+      "loss": 0.66,
+      "step": 10602
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4508863575285839,
+      "learning_rate": 2.8244104655156257e-07,
+      "loss": 0.6465,
+      "step": 10603
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.47982655358326537,
+      "learning_rate": 2.814719099432872e-07,
+      "loss": 0.6419,
+      "step": 10604
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3700842493095291,
+      "learning_rate": 2.8050442710414685e-07,
+      "loss": 0.6055,
+      "step": 10605
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3650319755061093,
+      "learning_rate": 2.7953859811528984e-07,
+      "loss": 0.6128,
+      "step": 10606
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.318242913724516,
+      "learning_rate": 2.785744230577092e-07,
+      "loss": 0.6239,
+      "step": 10607
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.38296592056728507,
+      "learning_rate": 2.776119020122758e-07,
+      "loss": 0.6147,
+      "step": 10608
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.41831761793059125,
+      "learning_rate": 2.766510350597162e-07,
+      "loss": 0.6446,
+      "step": 10609
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.34564351084524436,
+      "learning_rate": 2.756918222806104e-07,
+      "loss": 0.6144,
+      "step": 10610
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.41189519246816597,
+      "learning_rate": 2.7473426375541403e-07,
+      "loss": 0.6612,
+      "step": 10611
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3296578982692426,
+      "learning_rate": 2.7377835956443166e-07,
+      "loss": 0.6237,
+      "step": 10612
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.37958331818259233,
+      "learning_rate": 2.7282410978783483e-07,
+      "loss": 0.6322,
+      "step": 10613
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3396912338815792,
+      "learning_rate": 2.7187151450565495e-07,
+      "loss": 0.615,
+      "step": 10614
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3871511492667203,
+      "learning_rate": 2.7092057379778604e-07,
+      "loss": 0.6284,
+      "step": 10615
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4001542277346474,
+      "learning_rate": 2.6997128774398195e-07,
+      "loss": 0.6099,
+      "step": 10616
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4083764543795899,
+      "learning_rate": 2.690236564238591e-07,
+      "loss": 0.6294,
+      "step": 10617
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.48826505360961464,
+      "learning_rate": 2.680776799168916e-07,
+      "loss": 0.6134,
+      "step": 10618
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3745659114638386,
+      "learning_rate": 2.671333583024205e-07,
+      "loss": 0.6033,
+      "step": 10619
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.38148102438212633,
+      "learning_rate": 2.661906916596424e-07,
+      "loss": 0.6176,
+      "step": 10620
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.41538687774375394,
+      "learning_rate": 2.652496800676185e-07,
+      "loss": 0.6221,
+      "step": 10621
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3641349861915255,
+      "learning_rate": 2.643103236052702e-07,
+      "loss": 0.65,
+      "step": 10622
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.428365116764232,
+      "learning_rate": 2.633726223513811e-07,
+      "loss": 0.6361,
+      "step": 10623
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3996558877038266,
+      "learning_rate": 2.6243657638459487e-07,
+      "loss": 0.6408,
+      "step": 10624
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.40692192963094065,
+      "learning_rate": 2.6150218578341547e-07,
+      "loss": 0.6252,
+      "step": 10625
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3968824790035688,
+      "learning_rate": 2.605694506262091e-07,
+      "loss": 0.6243,
+      "step": 10626
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.42084400219937934,
+      "learning_rate": 2.5963837099120423e-07,
+      "loss": 0.6056,
+      "step": 10627
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3988701631498605,
+      "learning_rate": 2.587089469564874e-07,
+      "loss": 0.6626,
+      "step": 10628
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.39317122553429795,
+      "learning_rate": 2.5778117860001173e-07,
+      "loss": 0.619,
+      "step": 10629
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3702848853016354,
+      "learning_rate": 2.568550659995861e-07,
+      "loss": 0.6507,
+      "step": 10630
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.5858750638583259,
+      "learning_rate": 2.5593060923288173e-07,
+      "loss": 6.4773,
+      "step": 10631
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.35877224615732234,
+      "learning_rate": 2.550078083774321e-07,
+      "loss": 0.6121,
+      "step": 10632
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3966528908270933,
+      "learning_rate": 2.5408666351063094e-07,
+      "loss": 0.6023,
+      "step": 10633
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.417231976582843,
+      "learning_rate": 2.5316717470973637e-07,
+      "loss": 0.6575,
+      "step": 10634
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.37700420643068516,
+      "learning_rate": 2.522493420518601e-07,
+      "loss": 0.6586,
+      "step": 10635
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5478711952242085,
+      "learning_rate": 2.51333165613985e-07,
+      "loss": 0.6392,
+      "step": 10636
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.37725105394738023,
+      "learning_rate": 2.5041864547294513e-07,
+      "loss": 0.6219,
+      "step": 10637
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.37503289222784714,
+      "learning_rate": 2.495057817054436e-07,
+      "loss": 0.6256,
+      "step": 10638
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3575026234004002,
+      "learning_rate": 2.485945743880369e-07,
+      "loss": 0.6106,
+      "step": 10639
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.34459981374801674,
+      "learning_rate": 2.4768502359715284e-07,
+      "loss": 0.6565,
+      "step": 10640
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.42733111205113505,
+      "learning_rate": 2.4677712940906816e-07,
+      "loss": 0.6275,
+      "step": 10641
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3820514994418619,
+      "learning_rate": 2.4587089189993086e-07,
+      "loss": 0.6225,
+      "step": 10642
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3969783636049279,
+      "learning_rate": 2.4496631114574675e-07,
+      "loss": 0.62,
+      "step": 10643
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.36547064958905406,
+      "learning_rate": 2.440633872223774e-07,
+      "loss": 0.6439,
+      "step": 10644
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.37314217699897745,
+      "learning_rate": 2.4316212020555343e-07,
+      "loss": 0.6325,
+      "step": 10645
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3740005121017704,
+      "learning_rate": 2.4226251017086534e-07,
+      "loss": 0.6278,
+      "step": 10646
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.41050537533554166,
+      "learning_rate": 2.413645571937551e-07,
+      "loss": 0.6452,
+      "step": 10647
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.36417155069222845,
+      "learning_rate": 2.404682613495379e-07,
+      "loss": 0.6244,
+      "step": 10648
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.39842510031859746,
+      "learning_rate": 2.395736227133849e-07,
+      "loss": 0.6504,
+      "step": 10649
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.36424662814890346,
+      "learning_rate": 2.386806413603271e-07,
+      "loss": 0.6245,
+      "step": 10650
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3766471271635552,
+      "learning_rate": 2.377893173652601e-07,
+      "loss": 0.6423,
+      "step": 10651
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.37929348234818344,
+      "learning_rate": 2.368996508029331e-07,
+      "loss": 0.6304,
+      "step": 10652
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3846789298121536,
+      "learning_rate": 2.360116417479663e-07,
+      "loss": 0.6214,
+      "step": 10653
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3718003635428145,
+      "learning_rate": 2.3512529027483355e-07,
+      "loss": 0.6292,
+      "step": 10654
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.35765191240784877,
+      "learning_rate": 2.342405964578731e-07,
+      "loss": 0.6037,
+      "step": 10655
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.39607203980442446,
+      "learning_rate": 2.3335756037128343e-07,
+      "loss": 0.612,
+      "step": 10656
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5098986269255781,
+      "learning_rate": 2.3247618208912525e-07,
+      "loss": 0.6082,
+      "step": 10657
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4220379777579146,
+      "learning_rate": 2.315964616853128e-07,
+      "loss": 0.6655,
+      "step": 10658
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4088850601376685,
+      "learning_rate": 2.307183992336315e-07,
+      "loss": 0.6119,
+      "step": 10659
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3581703061472351,
+      "learning_rate": 2.2984199480772462e-07,
+      "loss": 0.6405,
+      "step": 10660
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3919101095006124,
+      "learning_rate": 2.2896724848109342e-07,
+      "loss": 0.6451,
+      "step": 10661
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.41658005068818443,
+      "learning_rate": 2.280941603270992e-07,
+      "loss": 0.6465,
+      "step": 10662
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.40762150663400076,
+      "learning_rate": 2.2722273041897002e-07,
+      "loss": 0.6405,
+      "step": 10663
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4080292700778098,
+      "learning_rate": 2.263529588297919e-07,
+      "loss": 0.6674,
+      "step": 10664
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4160036099617296,
+      "learning_rate": 2.2548484563251095e-07,
+      "loss": 0.6488,
+      "step": 10665
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.40948002048816323,
+      "learning_rate": 2.2461839089993555e-07,
+      "loss": 0.646,
+      "step": 10666
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3739700077516803,
+      "learning_rate": 2.237535947047298e-07,
+      "loss": 0.641,
+      "step": 10667
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3950822633824022,
+      "learning_rate": 2.2289045711942902e-07,
+      "loss": 0.6265,
+      "step": 10668
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3575595781195641,
+      "learning_rate": 2.2202897821641979e-07,
+      "loss": 0.6132,
+      "step": 10669
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.4533937688622547,
+      "learning_rate": 2.2116915806795314e-07,
+      "loss": 6.3754,
+      "step": 10670
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4002808392903457,
+      "learning_rate": 2.2031099674614476e-07,
+      "loss": 0.6244,
+      "step": 10671
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3883612231173123,
+      "learning_rate": 2.1945449432296373e-07,
+      "loss": 0.6526,
+      "step": 10672
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3780054807193886,
+      "learning_rate": 2.1859965087024592e-07,
+      "loss": 0.6511,
+      "step": 10673
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.42998187146846745,
+      "learning_rate": 2.1774646645968734e-07,
+      "loss": 0.6696,
+      "step": 10674
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.39683804187430094,
+      "learning_rate": 2.168949411628396e-07,
+      "loss": 0.6209,
+      "step": 10675
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4232200351408774,
+      "learning_rate": 2.1604507505112115e-07,
+      "loss": 0.6446,
+      "step": 10676
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.38518630099256196,
+      "learning_rate": 2.1519686819581053e-07,
+      "loss": 0.6098,
+      "step": 10677
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4181392845207775,
+      "learning_rate": 2.1435032066804195e-07,
+      "loss": 0.6289,
+      "step": 10678
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3656800601885039,
+      "learning_rate": 2.135054325388186e-07,
+      "loss": 0.6306,
+      "step": 10679
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3643277538186973,
+      "learning_rate": 2.126622038790016e-07,
+      "loss": 0.6062,
+      "step": 10680
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3665377757501576,
+      "learning_rate": 2.1182063475930548e-07,
+      "loss": 0.6305,
+      "step": 10681
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3885541775419142,
+      "learning_rate": 2.1098072525031376e-07,
+      "loss": 0.645,
+      "step": 10682
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3510743161234239,
+      "learning_rate": 2.1014247542247013e-07,
+      "loss": 0.6261,
+      "step": 10683
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.46135929659126507,
+      "learning_rate": 2.0930588534607833e-07,
+      "loss": 0.6615,
+      "step": 10684
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.35671281313536013,
+      "learning_rate": 2.0847095509129777e-07,
+      "loss": 0.6313,
+      "step": 10685
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.41577407971338076,
+      "learning_rate": 2.0763768472815914e-07,
+      "loss": 0.6389,
+      "step": 10686
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.37052121479583233,
+      "learning_rate": 2.068060743265443e-07,
+      "loss": 0.6292,
+      "step": 10687
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.41862348163354857,
+      "learning_rate": 2.059761239561997e-07,
+      "loss": 0.6222,
+      "step": 10688
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.35727001793303226,
+      "learning_rate": 2.051478336867341e-07,
+      "loss": 0.6003,
+      "step": 10689
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.40700865266255687,
+      "learning_rate": 2.0432120358760964e-07,
+      "loss": 0.6564,
+      "step": 10690
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.33799358247488387,
+      "learning_rate": 2.0349623372816207e-07,
+      "loss": 0.6374,
+      "step": 10691
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.45204961273644656,
+      "learning_rate": 2.02672924177576e-07,
+      "loss": 0.6818,
+      "step": 10692
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39013904725978177,
+      "learning_rate": 2.0185127500490066e-07,
+      "loss": 0.6479,
+      "step": 10693
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3987424226007503,
+      "learning_rate": 2.01031286279052e-07,
+      "loss": 0.6684,
+      "step": 10694
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4122018392221005,
+      "learning_rate": 2.0021295806879726e-07,
+      "loss": 0.6717,
+      "step": 10695
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38923202646911675,
+      "learning_rate": 1.9939629044277042e-07,
+      "loss": 0.6185,
+      "step": 10696
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4387211625658543,
+      "learning_rate": 1.9858128346946115e-07,
+      "loss": 0.6389,
+      "step": 10697
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3922073883646672,
+      "learning_rate": 1.977679372172281e-07,
+      "loss": 0.6391,
+      "step": 10698
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3680952955500908,
+      "learning_rate": 1.9695625175428114e-07,
+      "loss": 0.5702,
+      "step": 10699
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3752294703121898,
+      "learning_rate": 1.9614622714869914e-07,
+      "loss": 0.6138,
+      "step": 10700
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38398812303090174,
+      "learning_rate": 1.9533786346841444e-07,
+      "loss": 0.644,
+      "step": 10701
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.33257509357793996,
+      "learning_rate": 1.9453116078122615e-07,
+      "loss": 0.6312,
+      "step": 10702
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.44194701631985944,
+      "learning_rate": 1.9372611915479123e-07,
+      "loss": 0.6556,
+      "step": 10703
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.43249877854592694,
+      "learning_rate": 1.929227386566246e-07,
+      "loss": 0.674,
+      "step": 10704
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3797913792734567,
+      "learning_rate": 1.9212101935410786e-07,
+      "loss": 0.6562,
+      "step": 10705
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.37304306489628253,
+      "learning_rate": 1.9132096131448064e-07,
+      "loss": 0.6356,
+      "step": 10706
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39700237307422265,
+      "learning_rate": 1.905225646048403e-07,
+      "loss": 0.6292,
+      "step": 10707
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3620712101947402,
+      "learning_rate": 1.897258292921489e-07,
+      "loss": 0.6261,
+      "step": 10708
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3349785796392628,
+      "learning_rate": 1.8893075544322846e-07,
+      "loss": 0.6103,
+      "step": 10709
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38282966930082507,
+      "learning_rate": 1.8813734312475685e-07,
+      "loss": 0.6625,
+      "step": 10710
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4019186709891631,
+      "learning_rate": 1.87345592403283e-07,
+      "loss": 0.656,
+      "step": 10711
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.398341131748708,
+      "learning_rate": 1.8655550334520488e-07,
+      "loss": 0.6224,
+      "step": 10712
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39275979599341104,
+      "learning_rate": 1.8576707601678733e-07,
+      "loss": 0.6416,
+      "step": 10713
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39100174882120664,
+      "learning_rate": 1.8498031048415743e-07,
+      "loss": 0.6221,
+      "step": 10714
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3947476839639442,
+      "learning_rate": 1.8419520681329795e-07,
+      "loss": 0.6273,
+      "step": 10715
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.41992523006364146,
+      "learning_rate": 1.83411765070054e-07,
+      "loss": 0.6793,
+      "step": 10716
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3691025691800207,
+      "learning_rate": 1.8262998532013522e-07,
+      "loss": 0.618,
+      "step": 10717
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3712017642543309,
+      "learning_rate": 1.8184986762910473e-07,
+      "loss": 0.6262,
+      "step": 10718
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3726287062249199,
+      "learning_rate": 1.810714120623902e-07,
+      "loss": 0.6325,
+      "step": 10719
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38924908096160615,
+      "learning_rate": 1.8029461868528386e-07,
+      "loss": 0.6481,
+      "step": 10720
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4030665961773799,
+      "learning_rate": 1.7951948756292915e-07,
+      "loss": 0.6627,
+      "step": 10721
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.41893601277852244,
+      "learning_rate": 1.787460187603407e-07,
+      "loss": 0.6261,
+      "step": 10722
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3759002359548194,
+      "learning_rate": 1.7797421234238443e-07,
+      "loss": 0.6113,
+      "step": 10723
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4177647436280539,
+      "learning_rate": 1.7720406837379078e-07,
+      "loss": 0.6515,
+      "step": 10724
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.37451090583065505,
+      "learning_rate": 1.7643558691915252e-07,
+      "loss": 0.6357,
+      "step": 10725
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3899227384588738,
+      "learning_rate": 1.7566876804292254e-07,
+      "loss": 0.5851,
+      "step": 10726
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3946572993488272,
+      "learning_rate": 1.749036118094094e-07,
+      "loss": 0.6589,
+      "step": 10727
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3967347110970643,
+      "learning_rate": 1.7414011828278844e-07,
+      "loss": 0.6406,
+      "step": 10728
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38161285101049636,
+      "learning_rate": 1.7337828752709284e-07,
+      "loss": 0.6376,
+      "step": 10729
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.43553748890801913,
+      "learning_rate": 1.7261811960621378e-07,
+      "loss": 0.637,
+      "step": 10730
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.45523031057116514,
+      "learning_rate": 1.718596145839091e-07,
+      "loss": 0.6889,
+      "step": 10731
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39423971833951466,
+      "learning_rate": 1.7110277252379238e-07,
+      "loss": 0.6421,
+      "step": 10732
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3636775355305938,
+      "learning_rate": 1.7034759348933728e-07,
+      "loss": 0.5922,
+      "step": 10733
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39051686316810746,
+      "learning_rate": 1.6959407754388424e-07,
+      "loss": 0.616,
+      "step": 10734
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3551720036247442,
+      "learning_rate": 1.6884222475062494e-07,
+      "loss": 0.6315,
+      "step": 10735
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4101603501084584,
+      "learning_rate": 1.6809203517261784e-07,
+      "loss": 0.6466,
+      "step": 10736
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3927823939357106,
+      "learning_rate": 1.673435088727815e-07,
+      "loss": 0.6814,
+      "step": 10737
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38732913510871225,
+      "learning_rate": 1.6659664591389236e-07,
+      "loss": 0.6061,
+      "step": 10738
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3996202324912599,
+      "learning_rate": 1.658514463585914e-07,
+      "loss": 0.6312,
+      "step": 10739
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3950548694347203,
+      "learning_rate": 1.6510791026937534e-07,
+      "loss": 0.6429,
+      "step": 10740
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38083797606027947,
+      "learning_rate": 1.6436603770860538e-07,
+      "loss": 0.6309,
+      "step": 10741
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.381596430928946,
+      "learning_rate": 1.636258287384984e-07,
+      "loss": 0.6223,
+      "step": 10742
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3894084363534938,
+      "learning_rate": 1.6288728342113813e-07,
+      "loss": 0.6245,
+      "step": 10743
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4376580923626985,
+      "learning_rate": 1.6215040181846385e-07,
+      "loss": 0.6469,
+      "step": 10744
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.35241224890148687,
+      "learning_rate": 1.6141518399227728e-07,
+      "loss": 0.6275,
+      "step": 10745
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.40343535504563904,
+      "learning_rate": 1.6068163000424018e-07,
+      "loss": 0.618,
+      "step": 10746
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.9874655896170963,
+      "learning_rate": 1.599497399158767e-07,
+      "loss": 6.3965,
+      "step": 10747
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3900553367792027,
+      "learning_rate": 1.592195137885666e-07,
+      "loss": 0.6342,
+      "step": 10748
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3927921405064935,
+      "learning_rate": 1.584909516835542e-07,
+      "loss": 0.6392,
+      "step": 10749
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.36603022133649654,
+      "learning_rate": 1.577640536619418e-07,
+      "loss": 0.5977,
+      "step": 10750
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39166072567802834,
+      "learning_rate": 1.5703881978469838e-07,
+      "loss": 0.6179,
+      "step": 10751
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3906322136198259,
+      "learning_rate": 1.5631525011264192e-07,
+      "loss": 0.6294,
+      "step": 10752
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39229211499225863,
+      "learning_rate": 1.5559334470645948e-07,
+      "loss": 0.6225,
+      "step": 10753
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39013357911827334,
+      "learning_rate": 1.5487310362670037e-07,
+      "loss": 0.6625,
+      "step": 10754
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.33907332615900104,
+      "learning_rate": 1.541545269337652e-07,
+      "loss": 0.6147,
+      "step": 10755
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38602721977734555,
+      "learning_rate": 1.534376146879235e-07,
+      "loss": 0.6644,
+      "step": 10756
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.41800831237673475,
+      "learning_rate": 1.5272236694930054e-07,
+      "loss": 0.6509,
+      "step": 10757
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4443573140277173,
+      "learning_rate": 1.5200878377788163e-07,
+      "loss": 0.6487,
+      "step": 10758
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38600700844343305,
+      "learning_rate": 1.512968652335167e-07,
+      "loss": 0.631,
+      "step": 10759
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.42743238847455245,
+      "learning_rate": 1.5058661137591136e-07,
+      "loss": 0.6668,
+      "step": 10760
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38876512452422435,
+      "learning_rate": 1.4987802226463566e-07,
+      "loss": 0.6842,
+      "step": 10761
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3617054355723812,
+      "learning_rate": 1.4917109795911765e-07,
+      "loss": 0.6066,
+      "step": 10762
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3891561492622192,
+      "learning_rate": 1.484658385186455e-07,
+      "loss": 0.6275,
+      "step": 10763
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.424330537226233,
+      "learning_rate": 1.477622440023696e-07,
+      "loss": 0.6525,
+      "step": 10764
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3623750825142924,
+      "learning_rate": 1.4706031446929836e-07,
+      "loss": 0.6349,
+      "step": 10765
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3876907074828916,
+      "learning_rate": 1.4636004997830245e-07,
+      "loss": 0.6078,
+      "step": 10766
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.35041647516376834,
+      "learning_rate": 1.4566145058811042e-07,
+      "loss": 0.6331,
+      "step": 10767
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4134479307284524,
+      "learning_rate": 1.4496451635731767e-07,
+      "loss": 0.6198,
+      "step": 10768
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3595695444816515,
+      "learning_rate": 1.4426924734437075e-07,
+      "loss": 0.5942,
+      "step": 10769
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.34077041057284696,
+      "learning_rate": 1.435756436075808e-07,
+      "loss": 0.6205,
+      "step": 10770
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4027044296395187,
+      "learning_rate": 1.4288370520512352e-07,
+      "loss": 0.6214,
+      "step": 10771
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.41392241198005064,
+      "learning_rate": 1.4219343219502803e-07,
+      "loss": 0.6358,
+      "step": 10772
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.39994758830317606,
+      "learning_rate": 1.415048246351858e-07,
+      "loss": 0.6283,
+      "step": 10773
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.40539039437778723,
+      "learning_rate": 1.4081788258335281e-07,
+      "loss": 0.6463,
+      "step": 10774
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3330049455273338,
+      "learning_rate": 1.4013260609713864e-07,
+      "loss": 0.6316,
+      "step": 10775
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4069385573158518,
+      "learning_rate": 1.3944899523401723e-07,
+      "loss": 0.5599,
+      "step": 10776
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.41717088218364096,
+      "learning_rate": 1.387670500513272e-07,
+      "loss": 0.6728,
+      "step": 10777
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3788830329541129,
+      "learning_rate": 1.3808677060625385e-07,
+      "loss": 0.668,
+      "step": 10778
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3862675149275053,
+      "learning_rate": 1.374081569558583e-07,
+      "loss": 0.6297,
+      "step": 10779
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3784282172732695,
+      "learning_rate": 1.3673120915705275e-07,
+      "loss": 0.5909,
+      "step": 10780
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4316098564897954,
+      "learning_rate": 1.3605592726661176e-07,
+      "loss": 0.6301,
+      "step": 10781
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4320470431546389,
+      "learning_rate": 1.3538231134117008e-07,
+      "loss": 0.6243,
+      "step": 10782
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.34611719795094714,
+      "learning_rate": 1.347103614372247e-07,
+      "loss": 0.5997,
+      "step": 10783
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.341283217048528,
+      "learning_rate": 1.3404007761112836e-07,
+      "loss": 0.6191,
+      "step": 10784
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3624418449751481,
+      "learning_rate": 1.333714599191005e-07,
+      "loss": 0.6025,
+      "step": 10785
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.400406509965093,
+      "learning_rate": 1.3270450841721406e-07,
+      "loss": 0.6124,
+      "step": 10786
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4536782662574686,
+      "learning_rate": 1.320392231614065e-07,
+      "loss": 0.628,
+      "step": 10787
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4260436582655983,
+      "learning_rate": 1.3137560420747542e-07,
+      "loss": 0.6207,
+      "step": 10788
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3673222145435964,
+      "learning_rate": 1.3071365161107852e-07,
+      "loss": 0.6445,
+      "step": 10789
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.34553792288639534,
+      "learning_rate": 1.3005336542772917e-07,
+      "loss": 0.6172,
+      "step": 10790
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3870990678586634,
+      "learning_rate": 1.2939474571280975e-07,
+      "loss": 0.602,
+      "step": 10791
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.40194225166633996,
+      "learning_rate": 1.2873779252155382e-07,
+      "loss": 0.6856,
+      "step": 10792
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.38977156056421963,
+      "learning_rate": 1.280825059090618e-07,
+      "loss": 0.6486,
+      "step": 10793
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.47275448863529407,
+      "learning_rate": 1.2742888593028967e-07,
+      "loss": 0.6329,
+      "step": 10794
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.35535575109818457,
+      "learning_rate": 1.2677693264005808e-07,
+      "loss": 0.6368,
+      "step": 10795
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.38019498670159196,
+      "learning_rate": 1.2612664609304547e-07,
+      "loss": 0.6243,
+      "step": 10796
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.40907869515320633,
+      "learning_rate": 1.2547802634378603e-07,
+      "loss": 0.6315,
+      "step": 10797
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.345674651382928,
+      "learning_rate": 1.2483107344668512e-07,
+      "loss": 0.6453,
+      "step": 10798
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.37541860287133394,
+      "learning_rate": 1.2418578745599708e-07,
+      "loss": 0.6475,
+      "step": 10799
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3703168534870402,
+      "learning_rate": 1.2354216842584532e-07,
+      "loss": 0.6044,
+      "step": 10800
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.36664647876832684,
+      "learning_rate": 1.2290021641020444e-07,
+      "loss": 0.6554,
+      "step": 10801
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3754097826939186,
+      "learning_rate": 1.2225993146291804e-07,
+      "loss": 0.6187,
+      "step": 10802
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.38902260468231314,
+      "learning_rate": 1.2162131363768536e-07,
+      "loss": 0.628,
+      "step": 10803
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3878906774534563,
+      "learning_rate": 1.2098436298806582e-07,
+      "loss": 0.6371,
+      "step": 10804
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39468852311578223,
+      "learning_rate": 1.203490795674811e-07,
+      "loss": 0.656,
+      "step": 10805
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.41575533267060066,
+      "learning_rate": 1.1971546342920858e-07,
+      "loss": 0.6378,
+      "step": 10806
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.35007095371953345,
+      "learning_rate": 1.1908351462639245e-07,
+      "loss": 0.5987,
+      "step": 10807
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4334509679299794,
+      "learning_rate": 1.1845323321203029e-07,
+      "loss": 0.6586,
+      "step": 10808
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4157915909333719,
+      "learning_rate": 1.1782461923898647e-07,
+      "loss": 0.6485,
+      "step": 10809
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3842598262276994,
+      "learning_rate": 1.1719767275997885e-07,
+      "loss": 0.6443,
+      "step": 10810
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3778638977149289,
+      "learning_rate": 1.16572393827592e-07,
+      "loss": 0.6181,
+      "step": 10811
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3373820062760664,
+      "learning_rate": 1.15948782494264e-07,
+      "loss": 0.6068,
+      "step": 10812
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39115720628881284,
+      "learning_rate": 1.1532683881229744e-07,
+      "loss": 0.6596,
+      "step": 10813
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3368571446897459,
+      "learning_rate": 1.1470656283385285e-07,
+      "loss": 0.6079,
+      "step": 10814
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3689586460065943,
+      "learning_rate": 1.1408795461095523e-07,
+      "loss": 0.6317,
+      "step": 10815
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3578709870031204,
+      "learning_rate": 1.1347101419548534e-07,
+      "loss": 0.6291,
+      "step": 10816
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.43664399802109516,
+      "learning_rate": 1.1285574163918401e-07,
+      "loss": 0.6467,
+      "step": 10817
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.41699906985529495,
+      "learning_rate": 1.1224213699365439e-07,
+      "loss": 0.6514,
+      "step": 10818
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.37903989265184546,
+      "learning_rate": 1.1163020031035977e-07,
+      "loss": 0.5772,
+      "step": 10819
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.37717608785576295,
+      "learning_rate": 1.1101993164061908e-07,
+      "loss": 0.5895,
+      "step": 10820
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.397417718694834,
+      "learning_rate": 1.1041133103561585e-07,
+      "loss": 0.6252,
+      "step": 10821
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3678368067388458,
+      "learning_rate": 1.0980439854639591e-07,
+      "loss": 0.5946,
+      "step": 10822
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3888456453270636,
+      "learning_rate": 1.0919913422385853e-07,
+      "loss": 0.6373,
+      "step": 10823
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3460443855951544,
+      "learning_rate": 1.0859553811876978e-07,
+      "loss": 0.618,
+      "step": 10824
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3969296080481651,
+      "learning_rate": 1.0799361028174915e-07,
+      "loss": 0.6405,
+      "step": 10825
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.38626661118714545,
+      "learning_rate": 1.0739335076328072e-07,
+      "loss": 0.6317,
+      "step": 10826
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39976734828980065,
+      "learning_rate": 1.0679475961370867e-07,
+      "loss": 0.6021,
+      "step": 10827
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.37970162483880143,
+      "learning_rate": 1.0619783688323282e-07,
+      "loss": 0.6632,
+      "step": 10828
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39087683128532585,
+      "learning_rate": 1.0560258262191981e-07,
+      "loss": 0.6004,
+      "step": 10829
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.36667117001320465,
+      "learning_rate": 1.0500899687969412e-07,
+      "loss": 0.6543,
+      "step": 10830
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.36585729071053374,
+      "learning_rate": 1.0441707970633597e-07,
+      "loss": 0.629,
+      "step": 10831
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3938629104505181,
+      "learning_rate": 1.0382683115148784e-07,
+      "loss": 0.6281,
+      "step": 10832
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4186989594365788,
+      "learning_rate": 1.0323825126465458e-07,
+      "loss": 0.6174,
+      "step": 10833
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3777154393468777,
+      "learning_rate": 1.0265134009520339e-07,
+      "loss": 0.6467,
+      "step": 10834
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4023047697902933,
+      "learning_rate": 1.0206609769235043e-07,
+      "loss": 0.623,
+      "step": 10835
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3822428017926037,
+      "learning_rate": 1.0148252410518534e-07,
+      "loss": 0.6276,
+      "step": 10836
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4158849745652118,
+      "learning_rate": 1.0090061938264894e-07,
+      "loss": 0.5946,
+      "step": 10837
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.34869199635694875,
+      "learning_rate": 1.0032038357354668e-07,
+      "loss": 0.5512,
+      "step": 10838
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3729242172927673,
+      "learning_rate": 9.974181672654182e-08,
+      "loss": 0.6188,
+      "step": 10839
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.37490944871936094,
+      "learning_rate": 9.916491889015556e-08,
+      "loss": 0.6202,
+      "step": 10840
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3623868663103693,
+      "learning_rate": 9.858969011277363e-08,
+      "loss": 0.6134,
+      "step": 10841
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4423037293961836,
+      "learning_rate": 9.801613044263969e-08,
+      "loss": 0.6866,
+      "step": 10842
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.40963309485071514,
+      "learning_rate": 9.744423992785746e-08,
+      "loss": 0.652,
+      "step": 10843
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.42881521962569163,
+      "learning_rate": 9.687401861639078e-08,
+      "loss": 0.6312,
+      "step": 10844
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4217169016493629,
+      "learning_rate": 9.630546655606365e-08,
+      "loss": 0.6319,
+      "step": 10845
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.427931541199305,
+      "learning_rate": 9.573858379455791e-08,
+      "loss": 0.5986,
+      "step": 10846
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39679565911370196,
+      "learning_rate": 9.517337037941998e-08,
+      "loss": 0.6596,
+      "step": 10847
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.43263882695289774,
+      "learning_rate": 9.460982635805193e-08,
+      "loss": 0.6055,
+      "step": 10848
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3655830841253834,
+      "learning_rate": 9.40479517777182e-08,
+      "loss": 0.6008,
+      "step": 10849
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.40350488543985474,
+      "learning_rate": 9.348774668554329e-08,
+      "loss": 0.6413,
+      "step": 10850
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4213738107657056,
+      "learning_rate": 9.292921112850961e-08,
+      "loss": 0.6576,
+      "step": 10851
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3767753798306356,
+      "learning_rate": 9.237234515346194e-08,
+      "loss": 0.6457,
+      "step": 10852
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.38316102760250215,
+      "learning_rate": 9.18171488071029e-08,
+      "loss": 0.6602,
+      "step": 10853
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3948178781860868,
+      "learning_rate": 9.126362213599749e-08,
+      "loss": 0.6449,
+      "step": 10854
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39925601752367046,
+      "learning_rate": 9.071176518656632e-08,
+      "loss": 0.6208,
+      "step": 10855
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.397663419656758,
+      "learning_rate": 9.016157800509684e-08,
+      "loss": 0.6326,
+      "step": 10856
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.38854022868448646,
+      "learning_rate": 8.96130606377299e-08,
+      "loss": 0.6419,
+      "step": 10857
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.41106800258652126,
+      "learning_rate": 8.90662131304687e-08,
+      "loss": 0.5935,
+      "step": 10858
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3773887287259898,
+      "learning_rate": 8.852103552918101e-08,
+      "loss": 0.6489,
+      "step": 10859
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3667225384193779,
+      "learning_rate": 8.79775278795858e-08,
+      "loss": 0.625,
+      "step": 10860
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.46496282815509177,
+      "learning_rate": 8.74356902272666e-08,
+      "loss": 0.6318,
+      "step": 10861
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.406563458358104,
+      "learning_rate": 8.689552261766932e-08,
+      "loss": 0.6576,
+      "step": 10862
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3948145834835401,
+      "learning_rate": 8.635702509609322e-08,
+      "loss": 0.6092,
+      "step": 10863
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39868591863167757,
+      "learning_rate": 8.582019770770667e-08,
+      "loss": 0.6856,
+      "step": 10864
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4082285112507124,
+      "learning_rate": 8.528504049752917e-08,
+      "loss": 0.6357,
+      "step": 10865
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.38896126815452214,
+      "learning_rate": 8.475155351044262e-08,
+      "loss": 0.6333,
+      "step": 10866
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3650205866855372,
+      "learning_rate": 8.421973679119344e-08,
+      "loss": 0.6156,
+      "step": 10867
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.40605961962605797,
+      "learning_rate": 8.368959038438374e-08,
+      "loss": 0.6528,
+      "step": 10868
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.38899174593694597,
+      "learning_rate": 8.316111433447349e-08,
+      "loss": 0.6198,
+      "step": 10869
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4259780236372657,
+      "learning_rate": 8.263430868578726e-08,
+      "loss": 0.6516,
+      "step": 10870
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.35609059152897654,
+      "learning_rate": 8.210917348250746e-08,
+      "loss": 0.6254,
+      "step": 10871
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.5241101925091995,
+      "learning_rate": 8.158570876867667e-08,
+      "loss": 6.3693,
+      "step": 10872
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3489815203763033,
+      "learning_rate": 8.106391458819752e-08,
+      "loss": 0.6275,
+      "step": 10873
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3635149392844488,
+      "learning_rate": 8.054379098483278e-08,
+      "loss": 0.6434,
+      "step": 10874
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39672570130195567,
+      "learning_rate": 8.002533800220091e-08,
+      "loss": 0.6596,
+      "step": 10875
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.41650128329752345,
+      "learning_rate": 7.950855568378934e-08,
+      "loss": 0.6508,
+      "step": 10876
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.31655809528913187,
+      "learning_rate": 7.899344407293674e-08,
+      "loss": 0.6211,
+      "step": 10877
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4115183892365752,
+      "learning_rate": 7.84800032128441e-08,
+      "loss": 0.6323,
+      "step": 10878
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.32721848431363704,
+      "learning_rate": 7.796823314657697e-08,
+      "loss": 0.6087,
+      "step": 10879
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4128995834802421,
+      "learning_rate": 7.745813391705214e-08,
+      "loss": 0.6239,
+      "step": 10880
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3525693886917249,
+      "learning_rate": 7.694970556705317e-08,
+      "loss": 0.6306,
+      "step": 10881
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.36584329998257886,
+      "learning_rate": 7.644294813922371e-08,
+      "loss": 0.6572,
+      "step": 10882
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.34306117512635387,
+      "learning_rate": 7.593786167606087e-08,
+      "loss": 0.6069,
+      "step": 10883
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4239295880785797,
+      "learning_rate": 7.543444621992857e-08,
+      "loss": 0.6553,
+      "step": 10884
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.40708929148142775,
+      "learning_rate": 7.493270181304412e-08,
+      "loss": 0.6514,
+      "step": 10885
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4086456965430752,
+      "learning_rate": 7.443262849749389e-08,
+      "loss": 0.6156,
+      "step": 10886
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.1406353256862432,
+      "learning_rate": 7.393422631521319e-08,
+      "loss": 6.3918,
+      "step": 10887
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4005023578469892,
+      "learning_rate": 7.343749530800415e-08,
+      "loss": 0.642,
+      "step": 10888
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3432970867943441,
+      "learning_rate": 7.294243551752678e-08,
+      "loss": 0.5908,
+      "step": 10889
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3607920527068944,
+      "learning_rate": 7.244904698530342e-08,
+      "loss": 0.5877,
+      "step": 10890
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3557995398343183,
+      "learning_rate": 7.195732975270986e-08,
+      "loss": 0.5639,
+      "step": 10891
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3974914882716564,
+      "learning_rate": 7.146728386098867e-08,
+      "loss": 0.6683,
+      "step": 10892
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.36961531474989423,
+      "learning_rate": 7.09789093512403e-08,
+      "loss": 0.6427,
+      "step": 10893
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39385388772648194,
+      "learning_rate": 7.04922062644231e-08,
+      "loss": 0.6662,
+      "step": 10894
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.347435410162392,
+      "learning_rate": 7.000717464135332e-08,
+      "loss": 0.6362,
+      "step": 10895
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.33967491430766483,
+      "learning_rate": 6.952381452271395e-08,
+      "loss": 0.6252,
+      "step": 10896
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3842620588740611,
+      "learning_rate": 6.904212594904148e-08,
+      "loss": 0.6212,
+      "step": 10897
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3698457050220147,
+      "learning_rate": 6.85621089607369e-08,
+      "loss": 0.5897,
+      "step": 10898
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.32482245288628925,
+      "learning_rate": 6.80837635980569e-08,
+      "loss": 0.5767,
+      "step": 10899
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4382088938890151,
+      "learning_rate": 6.76070899011183e-08,
+      "loss": 0.6028,
+      "step": 10900
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4156893466031939,
+      "learning_rate": 6.713208790990244e-08,
+      "loss": 0.6673,
+      "step": 10901
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3676259955214203,
+      "learning_rate": 6.665875766424412e-08,
+      "loss": 0.6172,
+      "step": 10902
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.37277979549282125,
+      "learning_rate": 6.61870992038427e-08,
+      "loss": 0.6128,
+      "step": 10903
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3685650465338923,
+      "learning_rate": 6.571711256825541e-08,
+      "loss": 0.6077,
+      "step": 10904
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37951722843450486,
+      "learning_rate": 6.524879779689963e-08,
+      "loss": 0.5836,
+      "step": 10905
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4545634824238874,
+      "learning_rate": 6.478215492905282e-08,
+      "loss": 0.6556,
+      "step": 10906
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4041392752135622,
+      "learning_rate": 6.431718400385035e-08,
+      "loss": 0.6567,
+      "step": 10907
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3551368483822259,
+      "learning_rate": 6.385388506028989e-08,
+      "loss": 0.6096,
+      "step": 10908
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.35632501902034946,
+      "learning_rate": 6.339225813722705e-08,
+      "loss": 0.614,
+      "step": 10909
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3739336380834769,
+      "learning_rate": 6.293230327337974e-08,
+      "loss": 0.6448,
+      "step": 10910
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.40532976820282535,
+      "learning_rate": 6.247402050731932e-08,
+      "loss": 0.6512,
+      "step": 10911
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.36391614662179655,
+      "learning_rate": 6.201740987748617e-08,
+      "loss": 0.626,
+      "step": 10912
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3927707812179303,
+      "learning_rate": 6.156247142217409e-08,
+      "loss": 0.5912,
+      "step": 10913
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4056709668453997,
+      "learning_rate": 6.1109205179537e-08,
+      "loss": 0.6287,
+      "step": 10914
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3821935346604362,
+      "learning_rate": 6.065761118759117e-08,
+      "loss": 0.6243,
+      "step": 10915
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.39182838338158094,
+      "learning_rate": 6.020768948421074e-08,
+      "loss": 0.6357,
+      "step": 10916
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.403414120433593,
+      "learning_rate": 5.975944010712997e-08,
+      "loss": 0.6241,
+      "step": 10917
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4065562653914822,
+      "learning_rate": 5.931286309394324e-08,
+      "loss": 0.6437,
+      "step": 10918
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.875087637195422,
+      "learning_rate": 5.886795848210502e-08,
+      "loss": 6.3376,
+      "step": 10919
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37809039079271645,
+      "learning_rate": 5.8424726308927707e-08,
+      "loss": 0.5948,
+      "step": 10920
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3605666116032633,
+      "learning_rate": 5.798316661158598e-08,
+      "loss": 0.583,
+      "step": 10921
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3941022318706249,
+      "learning_rate": 5.754327942711247e-08,
+      "loss": 0.6397,
+      "step": 10922
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3925767964749898,
+      "learning_rate": 5.7105064792397636e-08,
+      "loss": 0.6478,
+      "step": 10923
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.39689782298682336,
+      "learning_rate": 5.666852274419876e-08,
+      "loss": 0.6566,
+      "step": 10924
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3978312339930528,
+      "learning_rate": 5.623365331912434e-08,
+      "loss": 0.6603,
+      "step": 10925
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.38367285497531023,
+      "learning_rate": 5.5800456553647406e-08,
+      "loss": 0.6743,
+      "step": 10926
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4251966327598592,
+      "learning_rate": 5.536893248409891e-08,
+      "loss": 0.6131,
+      "step": 10927
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.45503672654140537,
+      "learning_rate": 5.493908114667212e-08,
+      "loss": 0.6867,
+      "step": 10928
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4239119385339685,
+      "learning_rate": 5.451090257741598e-08,
+      "loss": 0.6103,
+      "step": 10929
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4101551511352138,
+      "learning_rate": 5.408439681224398e-08,
+      "loss": 0.637,
+      "step": 10930
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.39806997966495816,
+      "learning_rate": 5.365956388692528e-08,
+      "loss": 0.6255,
+      "step": 10931
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37024427449227837,
+      "learning_rate": 5.323640383708695e-08,
+      "loss": 0.6471,
+      "step": 10932
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.45276734748112785,
+      "learning_rate": 5.281491669822503e-08,
+      "loss": 0.6798,
+      "step": 10933
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4265156263314558,
+      "learning_rate": 5.239510250568458e-08,
+      "loss": 0.6326,
+      "step": 10934
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.42882342302075555,
+      "learning_rate": 5.197696129467522e-08,
+      "loss": 0.656,
+      "step": 10935
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.44348114045068127,
+      "learning_rate": 5.1560493100268896e-08,
+      "loss": 0.6195,
+      "step": 10936
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3819869842408269,
+      "learning_rate": 5.1145697957391e-08,
+      "loss": 0.6723,
+      "step": 10937
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3784198483698531,
+      "learning_rate": 5.073257590083147e-08,
+      "loss": 0.6199,
+      "step": 10938
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4148554737553993,
+      "learning_rate": 5.032112696523817e-08,
+      "loss": 0.6648,
+      "step": 10939
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3824724417864458,
+      "learning_rate": 4.9911351185121246e-08,
+      "loss": 0.6412,
+      "step": 10940
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4565923172299564,
+      "learning_rate": 4.9503248594842125e-08,
+      "loss": 0.6498,
+      "step": 10941
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4203359806736928,
+      "learning_rate": 4.909681922863341e-08,
+      "loss": 0.66,
+      "step": 10942
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4170067342516517,
+      "learning_rate": 4.8692063120578945e-08,
+      "loss": 0.6577,
+      "step": 10943
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.39972280935163174,
+      "learning_rate": 4.8288980304627144e-08,
+      "loss": 0.6091,
+      "step": 10944
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3873539877872512,
+      "learning_rate": 4.788757081458428e-08,
+      "loss": 0.6433,
+      "step": 10945
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4208166493512853,
+      "learning_rate": 4.7487834684112335e-08,
+      "loss": 0.6498,
+      "step": 10946
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3803555661250889,
+      "learning_rate": 4.7089771946740024e-08,
+      "loss": 0.6268,
+      "step": 10947
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37999920892086614,
+      "learning_rate": 4.669338263584955e-08,
+      "loss": 0.6329,
+      "step": 10948
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37330296419717707,
+      "learning_rate": 4.629866678468986e-08,
+      "loss": 0.6398,
+      "step": 10949
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3848892205310072,
+      "learning_rate": 4.590562442636337e-08,
+      "loss": 0.6464,
+      "step": 10950
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.38649392568363267,
+      "learning_rate": 4.55142555938326e-08,
+      "loss": 0.6274,
+      "step": 10951
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3416608386834745,
+      "learning_rate": 4.5124560319922406e-08,
+      "loss": 0.5527,
+      "step": 10952
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 3.4055392324825577,
+      "learning_rate": 4.473653863731553e-08,
+      "loss": 6.3579,
+      "step": 10953
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.38754063931410426,
+      "learning_rate": 4.435019057855705e-08,
+      "loss": 0.6524,
+      "step": 10954
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.8549976807792792,
+      "learning_rate": 4.39655161760455e-08,
+      "loss": 6.3719,
+      "step": 10955
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4238838776314916,
+      "learning_rate": 4.3582515462048394e-08,
+      "loss": 0.6497,
+      "step": 10956
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.34671553703377983,
+      "learning_rate": 4.320118846868227e-08,
+      "loss": 0.6349,
+      "step": 10957
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3520259992800872,
+      "learning_rate": 4.2821535227930424e-08,
+      "loss": 0.634,
+      "step": 10958
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3545587008479222,
+      "learning_rate": 4.244355577163406e-08,
+      "loss": 0.6148,
+      "step": 10959
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3677564383960457,
+      "learning_rate": 4.206725013149449e-08,
+      "loss": 0.6126,
+      "step": 10960
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.44297430052709974,
+      "learning_rate": 4.1692618339073124e-08,
+      "loss": 0.6713,
+      "step": 10961
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.42221042037432227,
+      "learning_rate": 4.131966042578706e-08,
+      "loss": 0.6383,
+      "step": 10962
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4174275831219546,
+      "learning_rate": 4.0948376422915713e-08,
+      "loss": 0.6248,
+      "step": 10963
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.39557208457660975,
+      "learning_rate": 4.057876636160307e-08,
+      "loss": 0.5847,
+      "step": 10964
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4183924959040814,
+      "learning_rate": 4.021083027284212e-08,
+      "loss": 0.6439,
+      "step": 10965
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.408986655136401,
+      "learning_rate": 3.984456818749261e-08,
+      "loss": 0.5967,
+      "step": 10966
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3795827318828233,
+      "learning_rate": 3.9479980136274406e-08,
+      "loss": 0.6269,
+      "step": 10967
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.40702002155114336,
+      "learning_rate": 3.911706614976529e-08,
+      "loss": 0.6005,
+      "step": 10968
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.42885609500005345,
+      "learning_rate": 3.875582625839869e-08,
+      "loss": 0.6408,
+      "step": 10969
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4161685814949421,
+      "learning_rate": 3.8396260492477024e-08,
+      "loss": 0.6563,
+      "step": 10970
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3739402983716061,
+      "learning_rate": 3.803836888215173e-08,
+      "loss": 0.6077,
+      "step": 10971
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.35322949245068325,
+      "learning_rate": 3.768215145743881e-08,
+      "loss": 0.6333,
+      "step": 10972
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.43887007020211843,
+      "learning_rate": 3.7327608248216576e-08,
+      "loss": 0.6323,
+      "step": 10973
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37216454330459675,
+      "learning_rate": 3.697473928422124e-08,
+      "loss": 0.6303,
+      "step": 10974
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3299814497637736,
+      "learning_rate": 3.662354459504247e-08,
+      "loss": 0.6792,
+      "step": 10975
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3781988279971231,
+      "learning_rate": 3.627402421013893e-08,
+      "loss": 0.6358,
+      "step": 10976
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.39624206594687494,
+      "learning_rate": 3.592617815882271e-08,
+      "loss": 0.6006,
+      "step": 10977
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3863803824079895,
+      "learning_rate": 3.558000647026827e-08,
+      "loss": 0.6157,
+      "step": 10978
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.43855080640922417,
+      "learning_rate": 3.523550917350571e-08,
+      "loss": 0.6367,
+      "step": 10979
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.39116421046798333,
+      "learning_rate": 3.48926862974297e-08,
+      "loss": 0.5986,
+      "step": 10980
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.5071162023338416,
+      "learning_rate": 3.455153787079501e-08,
+      "loss": 6.4185,
+      "step": 10981
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.35424706100974884,
+      "learning_rate": 3.421206392220766e-08,
+      "loss": 0.6404,
+      "step": 10982
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4070549559081413,
+      "learning_rate": 3.3874264480144856e-08,
+      "loss": 0.6425,
+      "step": 10983
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4270479123218154,
+      "learning_rate": 3.3538139572932834e-08,
+      "loss": 0.6683,
+      "step": 10984
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4149476690586157,
+      "learning_rate": 3.32036892287646e-08,
+      "loss": 0.6902,
+      "step": 10985
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.44174230377449386,
+      "learning_rate": 3.287091347568661e-08,
+      "loss": 0.6655,
+      "step": 10986
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4517316846966959,
+      "learning_rate": 3.2539812341614296e-08,
+      "loss": 0.6566,
+      "step": 10987
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.35563932749528676,
+      "learning_rate": 3.221038585430991e-08,
+      "loss": 0.6173,
+      "step": 10988
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.38953345314048904,
+      "learning_rate": 3.18826340414069e-08,
+      "loss": 0.6399,
+      "step": 10989
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3845644101411575,
+      "learning_rate": 3.155655693039439e-08,
+      "loss": 0.6351,
+      "step": 10990
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3485840220518135,
+      "learning_rate": 3.123215454861495e-08,
+      "loss": 0.597,
+      "step": 10991
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3932614353502084,
+      "learning_rate": 3.090942692328014e-08,
+      "loss": 0.6193,
+      "step": 10992
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.40195374708199927,
+      "learning_rate": 3.058837408145277e-08,
+      "loss": 0.6815,
+      "step": 10993
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37765358290328005,
+      "learning_rate": 3.02689960500624e-08,
+      "loss": 0.6619,
+      "step": 10994
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.33748047584123875,
+      "learning_rate": 2.995129285589649e-08,
+      "loss": 0.6302,
+      "step": 10995
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.403009336839424,
+      "learning_rate": 2.9635264525595953e-08,
+      "loss": 0.6478,
+      "step": 10996
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3320492464224679,
+      "learning_rate": 2.932091108566626e-08,
+      "loss": 0.6041,
+      "step": 10997
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.40272617866856103,
+      "learning_rate": 2.9008232562475202e-08,
+      "loss": 0.5958,
+      "step": 10998
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4219034355001011,
+      "learning_rate": 2.869722898224625e-08,
+      "loss": 0.6494,
+      "step": 10999
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.40332658821998013,
+      "learning_rate": 2.8387900371058542e-08,
+      "loss": 0.62,
+      "step": 11000
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4726895968487947,
+      "learning_rate": 2.8080246754857988e-08,
+      "loss": 6.4125,
+      "step": 11001
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.36826271878174305,
+      "learning_rate": 2.777426815944839e-08,
+      "loss": 0.6332,
+      "step": 11002
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.40881632661638434,
+      "learning_rate": 2.7469964610491452e-08,
+      "loss": 0.6262,
+      "step": 11003
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.5216858920914058,
+      "learning_rate": 2.716733613350675e-08,
+      "loss": 6.4396,
+      "step": 11004
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37974384072955353,
+      "learning_rate": 2.6866382753876207e-08,
+      "loss": 0.6532,
+      "step": 11005
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.38222343167361317,
+      "learning_rate": 2.6567104496841855e-08,
+      "loss": 0.6253,
+      "step": 11006
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3707980202433784,
+      "learning_rate": 2.6269501387503617e-08,
+      "loss": 0.6489,
+      "step": 11007
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.38816946436599514,
+      "learning_rate": 2.597357345081708e-08,
+      "loss": 0.6358,
+      "step": 11008
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.42040972034458624,
+      "learning_rate": 2.5679320711606835e-08,
+      "loss": 0.618,
+      "step": 11009
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4041382017847448,
+      "learning_rate": 2.5386743194546483e-08,
+      "loss": 0.6534,
+      "step": 11010
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.39447229396281835,
+      "learning_rate": 2.5095840924178604e-08,
+      "loss": 0.612,
+      "step": 11011
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.395727822772332,
+      "learning_rate": 2.4806613924899248e-08,
+      "loss": 0.6443,
+      "step": 11012
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.362071660936786,
+      "learning_rate": 2.451906222096456e-08,
+      "loss": 0.6146,
+      "step": 11013
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.36763131057037246,
+      "learning_rate": 2.4233185836493034e-08,
+      "loss": 0.6096,
+      "step": 11014
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.37855558953998447,
+      "learning_rate": 2.3948984795458818e-08,
+      "loss": 0.6307,
+      "step": 11015
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4521258960994786,
+      "learning_rate": 2.3666459121696184e-08,
+      "loss": 0.6374,
+      "step": 11016
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3884397690011655,
+      "learning_rate": 2.338560883890395e-08,
+      "loss": 0.652,
+      "step": 11017
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.38844409235228416,
+      "learning_rate": 2.310643397063661e-08,
+      "loss": 0.6096,
+      "step": 11018
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.40778268638936854,
+      "learning_rate": 2.2828934540304325e-08,
+      "loss": 0.6403,
+      "step": 11019
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3860485734213309,
+      "learning_rate": 2.2553110571184034e-08,
+      "loss": 0.6235,
+      "step": 11020
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.40146423023940847,
+      "learning_rate": 2.227896208640834e-08,
+      "loss": 0.6183,
+      "step": 11021
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3311768213759181,
+      "learning_rate": 2.2006489108967744e-08,
+      "loss": 0.6241,
+      "step": 11022
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.43826428825837094,
+      "learning_rate": 2.1735691661717297e-08,
+      "loss": 0.6736,
+      "step": 11023
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.34225810857565303,
+      "learning_rate": 2.1466569767365496e-08,
+      "loss": 0.6069,
+      "step": 11024
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.38335807313493875,
+      "learning_rate": 2.1199123448485403e-08,
+      "loss": 0.6652,
+      "step": 11025
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39620281646717787,
+      "learning_rate": 2.0933352727505742e-08,
+      "loss": 0.6413,
+      "step": 11026
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.4605533785177898,
+      "learning_rate": 2.0669257626717564e-08,
+      "loss": 6.3865,
+      "step": 11027
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.38611934123463587,
+      "learning_rate": 2.0406838168269828e-08,
+      "loss": 0.6164,
+      "step": 11028
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3804492892093342,
+      "learning_rate": 2.0146094374171587e-08,
+      "loss": 0.6272,
+      "step": 11029
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3644171007870826,
+      "learning_rate": 1.98870262662898e-08,
+      "loss": 0.6092,
+      "step": 11030
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.32407951248560707,
+      "learning_rate": 1.962963386635375e-08,
+      "loss": 0.5953,
+      "step": 11031
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39318558171863416,
+      "learning_rate": 1.937391719595061e-08,
+      "loss": 0.6507,
+      "step": 11032
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3853982319763574,
+      "learning_rate": 1.9119876276527672e-08,
+      "loss": 0.618,
+      "step": 11033
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.41951583515141033,
+      "learning_rate": 1.8867511129387893e-08,
+      "loss": 0.6294,
+      "step": 11034
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4214903000875841,
+      "learning_rate": 1.861682177569879e-08,
+      "loss": 0.6546,
+      "step": 11035
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.363461278000444,
+      "learning_rate": 1.8367808236485763e-08,
+      "loss": 0.5791,
+      "step": 11036
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.40058386243888155,
+      "learning_rate": 1.8120470532634325e-08,
+      "loss": 0.6233,
+      "step": 11037
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.35300384995887124,
+      "learning_rate": 1.787480868488567e-08,
+      "loss": 0.6359,
+      "step": 11038
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.38872079269282184,
+      "learning_rate": 1.763082271384553e-08,
+      "loss": 0.6359,
+      "step": 11039
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3820196727776456,
+      "learning_rate": 1.7388512639975317e-08,
+      "loss": 0.6156,
+      "step": 11040
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4270798220126327,
+      "learning_rate": 1.7147878483596557e-08,
+      "loss": 0.6342,
+      "step": 11041
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3513842208317666,
+      "learning_rate": 1.6908920264893104e-08,
+      "loss": 0.641,
+      "step": 11042
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3785665367039119,
+      "learning_rate": 1.6671638003904477e-08,
+      "loss": 0.6658,
+      "step": 11043
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3932205067853669,
+      "learning_rate": 1.6436031720532543e-08,
+      "loss": 0.6063,
+      "step": 11044
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4186690614442782,
+      "learning_rate": 1.620210143453482e-08,
+      "loss": 0.6655,
+      "step": 11045
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39250694861932084,
+      "learning_rate": 1.5969847165535623e-08,
+      "loss": 0.6401,
+      "step": 11046
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3593411078732928,
+      "learning_rate": 1.5739268933008256e-08,
+      "loss": 0.6183,
+      "step": 11047
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.37389599826884395,
+      "learning_rate": 1.5510366756295025e-08,
+      "loss": 0.666,
+      "step": 11048
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3840275284332479,
+      "learning_rate": 1.5283140654591687e-08,
+      "loss": 0.614,
+      "step": 11049
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39033619391328156,
+      "learning_rate": 1.5057590646954113e-08,
+      "loss": 0.6356,
+      "step": 11050
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3935425722401356,
+      "learning_rate": 1.4833716752302718e-08,
+      "loss": 0.6286,
+      "step": 11051
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.389278035368114,
+      "learning_rate": 1.4611518989411378e-08,
+      "loss": 0.6257,
+      "step": 11052
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3714147244448717,
+      "learning_rate": 1.4390997376916293e-08,
+      "loss": 0.6586,
+      "step": 11053
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.37638363751535053,
+      "learning_rate": 1.4172151933309342e-08,
+      "loss": 0.5943,
+      "step": 11054
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4088388116460171,
+      "learning_rate": 1.3954982676949169e-08,
+      "loss": 0.6619,
+      "step": 11055
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3838408446097167,
+      "learning_rate": 1.3739489626047875e-08,
+      "loss": 0.5902,
+      "step": 11056
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.43517032817326057,
+      "learning_rate": 1.3525672798677669e-08,
+      "loss": 0.6724,
+      "step": 11057
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.37363420539476144,
+      "learning_rate": 1.3313532212770874e-08,
+      "loss": 0.6327,
+      "step": 11058
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.36280612142456714,
+      "learning_rate": 1.3103067886122144e-08,
+      "loss": 0.6493,
+      "step": 11059
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3763887563517723,
+      "learning_rate": 1.2894279836379586e-08,
+      "loss": 0.6591,
+      "step": 11060
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3965957568716436,
+      "learning_rate": 1.2687168081055855e-08,
+      "loss": 0.6139,
+      "step": 11061
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.414160896352134,
+      "learning_rate": 1.2481732637521505e-08,
+      "loss": 0.6137,
+      "step": 11062
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3609960427356949,
+      "learning_rate": 1.227797352300497e-08,
+      "loss": 0.6313,
+      "step": 11063
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3812594909816523,
+      "learning_rate": 1.207589075459481e-08,
+      "loss": 0.6677,
+      "step": 11064
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3598525379694029,
+      "learning_rate": 1.1875484349241906e-08,
+      "loss": 0.6122,
+      "step": 11065
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3676389781641404,
+      "learning_rate": 1.1676754323750594e-08,
+      "loss": 0.6208,
+      "step": 11066
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3946329521772464,
+      "learning_rate": 1.1479700694791985e-08,
+      "loss": 0.65,
+      "step": 11067
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4345493316685126,
+      "learning_rate": 1.1284323478890635e-08,
+      "loss": 0.6438,
+      "step": 11068
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4264155177073852,
+      "learning_rate": 1.1090622692431218e-08,
+      "loss": 0.6292,
+      "step": 11069
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.40595352908141413,
+      "learning_rate": 1.089859835166296e-08,
+      "loss": 0.6317,
+      "step": 11070
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3693830200288033,
+      "learning_rate": 1.070825047268631e-08,
+      "loss": 0.6217,
+      "step": 11071
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.40700052792960995,
+      "learning_rate": 1.051957907147072e-08,
+      "loss": 0.6047,
+      "step": 11072
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.44126858545020936,
+      "learning_rate": 1.0332584163834647e-08,
+      "loss": 0.6759,
+      "step": 11073
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4395496266605132,
+      "learning_rate": 1.014726576546332e-08,
+      "loss": 0.6331,
+      "step": 11074
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3833459597200118,
+      "learning_rate": 9.963623891899865e-09,
+      "loss": 0.6203,
+      "step": 11075
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.372635215665833,
+      "learning_rate": 9.781658558545293e-09,
+      "loss": 0.6253,
+      "step": 11076
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3603841352387562,
+      "learning_rate": 9.60136978066073e-09,
+      "loss": 0.5774,
+      "step": 11077
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.37559866708209355,
+      "learning_rate": 9.422757573367413e-09,
+      "loss": 0.6511,
+      "step": 11078
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39594158774466565,
+      "learning_rate": 9.245821951644474e-09,
+      "loss": 0.6492,
+      "step": 11079
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4088574658189101,
+      "learning_rate": 9.07056293033115e-09,
+      "loss": 0.6592,
+      "step": 11080
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.42049913763968694,
+      "learning_rate": 8.896980524129018e-09,
+      "loss": 0.6264,
+      "step": 11081
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3960633730436863,
+      "learning_rate": 8.725074747590878e-09,
+      "loss": 0.6332,
+      "step": 11082
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.41604208050006275,
+      "learning_rate": 8.55484561514075e-09,
+      "loss": 0.6622,
+      "step": 11083
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3908419792656098,
+      "learning_rate": 8.386293141049439e-09,
+      "loss": 0.6261,
+      "step": 11084
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4151410001080487,
+      "learning_rate": 8.219417339456748e-09,
+      "loss": 0.6626,
+      "step": 11085
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.35571355133699273,
+      "learning_rate": 8.054218224358145e-09,
+      "loss": 0.6213,
+      "step": 11086
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.44223842665877294,
+      "learning_rate": 7.890695809606996e-09,
+      "loss": 0.66,
+      "step": 11087
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4043082476325844,
+      "learning_rate": 7.728850108918995e-09,
+      "loss": 0.6386,
+      "step": 11088
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4569683674303862,
+      "learning_rate": 7.568681135869948e-09,
+      "loss": 0.6843,
+      "step": 11089
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3827458153924817,
+      "learning_rate": 7.410188903889115e-09,
+      "loss": 0.661,
+      "step": 11090
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3745391741103453,
+      "learning_rate": 7.2533734262703046e-09,
+      "loss": 0.621,
+      "step": 11091
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4027864041366409,
+      "learning_rate": 7.09823471616744e-09,
+      "loss": 0.6313,
+      "step": 11092
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.41999498291077414,
+      "learning_rate": 6.944772786587894e-09,
+      "loss": 0.66,
+      "step": 11093
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.35018232492149265,
+      "learning_rate": 6.7929876504080335e-09,
+      "loss": 0.6292,
+      "step": 11094
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.404894717317082,
+      "learning_rate": 6.642879320353235e-09,
+      "loss": 0.6353,
+      "step": 11095
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39696253864802106,
+      "learning_rate": 6.494447809013426e-09,
+      "loss": 0.6397,
+      "step": 11096
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3826066433088956,
+      "learning_rate": 6.34769312883865e-09,
+      "loss": 0.6303,
+      "step": 11097
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4140887225014502,
+      "learning_rate": 6.202615292136838e-09,
+      "loss": 0.6337,
+      "step": 11098
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4219044380821125,
+      "learning_rate": 6.059214311073813e-09,
+      "loss": 0.6632,
+      "step": 11099
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.35951003760166195,
+      "learning_rate": 5.917490197679954e-09,
+      "loss": 0.5753,
+      "step": 11100
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4289953838729734,
+      "learning_rate": 5.7774429638390865e-09,
+      "loss": 0.6306,
+      "step": 11101
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3810419250404913,
+      "learning_rate": 5.6390726212973705e-09,
+      "loss": 0.631,
+      "step": 11102
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39022781081719876,
+      "learning_rate": 5.502379181658857e-09,
+      "loss": 0.6189,
+      "step": 11103
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.36753706908293243,
+      "learning_rate": 5.367362656387709e-09,
+      "loss": 0.6328,
+      "step": 11104
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.37074072529789326,
+      "learning_rate": 5.234023056810422e-09,
+      "loss": 0.6477,
+      "step": 11105
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39378225870777833,
+      "learning_rate": 5.102360394106942e-09,
+      "loss": 0.5804,
+      "step": 11106
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.41773755189203005,
+      "learning_rate": 4.972374679321768e-09,
+      "loss": 0.6649,
+      "step": 11107
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.41992952993386645,
+      "learning_rate": 4.844065923355068e-09,
+      "loss": 0.6491,
+      "step": 11108
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4151058671936696,
+      "learning_rate": 4.717434136967125e-09,
+      "loss": 0.6589,
+      "step": 11109
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.41424282951393776,
+      "learning_rate": 4.592479330782773e-09,
+      "loss": 0.6603,
+      "step": 11110
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4164898097559135,
+      "learning_rate": 4.469201515275856e-09,
+      "loss": 0.623,
+      "step": 11111
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39547220586263065,
+      "learning_rate": 4.347600700789212e-09,
+      "loss": 0.6362,
+      "step": 11112
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.41225758984499583,
+      "learning_rate": 4.227676897521349e-09,
+      "loss": 0.6207,
+      "step": 11113
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.40793356425984045,
+      "learning_rate": 4.109430115530888e-09,
+      "loss": 0.658,
+      "step": 11114
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.43590285331951595,
+      "learning_rate": 3.9928603647321205e-09,
+      "loss": 0.6469,
+      "step": 11115
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39963807091029907,
+      "learning_rate": 3.877967654901671e-09,
+      "loss": 0.6125,
+      "step": 11116
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3742216197341699,
+      "learning_rate": 3.764751995678495e-09,
+      "loss": 0.6352,
+      "step": 11117
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3975737488913313,
+      "learning_rate": 3.653213396555e-09,
+      "loss": 0.6381,
+      "step": 11118
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3495200979851979,
+      "learning_rate": 3.543351866888145e-09,
+      "loss": 0.5754,
+      "step": 11119
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.40463309935357716,
+      "learning_rate": 3.4351674158905613e-09,
+      "loss": 0.6395,
+      "step": 11120
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.377366542266821,
+      "learning_rate": 3.3286600526372114e-09,
+      "loss": 0.6443,
+      "step": 11121
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.44113179434182337,
+      "learning_rate": 3.223829786056509e-09,
+      "loss": 0.6197,
+      "step": 11122
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.41024794814422033,
+      "learning_rate": 3.120676624945862e-09,
+      "loss": 0.5627,
+      "step": 11123
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.6217415990344561,
+      "learning_rate": 3.0192005779516863e-09,
+      "loss": 6.5482,
+      "step": 11124
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.37249832056222576,
+      "learning_rate": 2.9194016535871727e-09,
+      "loss": 0.6196,
+      "step": 11125
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3598060858056972,
+      "learning_rate": 2.8212798602234027e-09,
+      "loss": 0.6238,
+      "step": 11126
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.362067362121588,
+      "learning_rate": 2.7248352060871286e-09,
+      "loss": 0.6307,
+      "step": 11127
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.380447109842645,
+      "learning_rate": 2.630067699269656e-09,
+      "loss": 0.6386,
+      "step": 11128
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.37013802651361577,
+      "learning_rate": 2.5369773477179618e-09,
+      "loss": 0.618,
+      "step": 11129
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.38098524442026116,
+      "learning_rate": 2.445564159236913e-09,
+      "loss": 0.6317,
+      "step": 11130
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.383333187104593,
+      "learning_rate": 2.355828141498151e-09,
+      "loss": 0.6141,
+      "step": 11131
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.423086505423099,
+      "learning_rate": 2.267769302022327e-09,
+      "loss": 0.6611,
+      "step": 11132
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4291362247199439,
+      "learning_rate": 2.181387648199085e-09,
+      "loss": 0.6701,
+      "step": 11133
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.44781521066401525,
+      "learning_rate": 2.0966831872715197e-09,
+      "loss": 0.6896,
+      "step": 11134
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3959551417174536,
+      "learning_rate": 2.0136559263450596e-09,
+      "loss": 0.6366,
+      "step": 11135
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3575885696726365,
+      "learning_rate": 1.9323058723785813e-09,
+      "loss": 0.6226,
+      "step": 11136
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.42687771479286174,
+      "learning_rate": 1.8526330321999575e-09,
+      "loss": 0.6825,
+      "step": 11137
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.36482137772265927,
+      "learning_rate": 1.7746374124905097e-09,
+      "loss": 0.6308,
+      "step": 11138
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4095363464307073,
+      "learning_rate": 1.6983190197894517e-09,
+      "loss": 0.6508,
+      "step": 11139
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4001151103312084,
+      "learning_rate": 1.6236778604983295e-09,
+      "loss": 0.6683,
+      "step": 11140
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4006107900090801,
+      "learning_rate": 1.5507139408765803e-09,
+      "loss": 0.589,
+      "step": 11141
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.31566788936591333,
+      "learning_rate": 1.4794272670459742e-09,
+      "loss": 0.6196,
+      "step": 11142
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3892833414207321,
+      "learning_rate": 1.4098178449839517e-09,
+      "loss": 0.6398,
+      "step": 11143
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.36444178946963696,
+      "learning_rate": 1.3418856805258451e-09,
+      "loss": 0.631,
+      "step": 11144
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4097736464619722,
+      "learning_rate": 1.27563077937376e-09,
+      "loss": 0.6171,
+      "step": 11145
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.37342621404775833,
+      "learning_rate": 1.2110531470810317e-09,
+      "loss": 0.6695,
+      "step": 11146
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.41244691856285975,
+      "learning_rate": 1.1481527890655486e-09,
+      "loss": 0.6452,
+      "step": 11147
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4145287857542926,
+      "learning_rate": 1.0869297106030908e-09,
+      "loss": 0.5926,
+      "step": 11148
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4128749573369694,
+      "learning_rate": 1.0273839168251087e-09,
+      "loss": 0.6322,
+      "step": 11149
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3396831726609409,
+      "learning_rate": 9.695154127298267e-10,
+      "loss": 0.6206,
+      "step": 11150
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3503212719559611,
+      "learning_rate": 9.133242031666988e-10,
+      "loss": 0.6307,
+      "step": 11151
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3801947048707954,
+      "learning_rate": 8.588102928519526e-10,
+      "loss": 0.6068,
+      "step": 11152
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.40171410650426653,
+      "learning_rate": 8.059736863530454e-10,
+      "loss": 0.6469,
+      "step": 11153
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.2694113991580271,
+      "learning_rate": 7.548143881064285e-10,
+      "loss": 6.4536,
+      "step": 11154
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.46073399427268985,
+      "learning_rate": 7.053324024020036e-10,
+      "loss": 0.5934,
+      "step": 11155
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3882023077801949,
+      "learning_rate": 6.575277333853436e-10,
+      "loss": 0.6403,
+      "step": 11156
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3862363133516602,
+      "learning_rate": 6.114003850710148e-10,
+      "loss": 0.637,
+      "step": 11157
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3557080280773298,
+      "learning_rate": 5.669503613225935e-10,
+      "loss": 0.6116,
+      "step": 11158
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3677787225724615,
+      "learning_rate": 5.241776658726494e-10,
+      "loss": 0.6251,
+      "step": 11159
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.41756400307666725,
+      "learning_rate": 4.830823023072028e-10,
+      "loss": 0.6551,
+      "step": 11160
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3836098761818286,
+      "learning_rate": 4.4366427407238623e-10,
+      "loss": 0.6484,
+      "step": 11161
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.39615285101361314,
+      "learning_rate": 4.059235844744436e-10,
+      "loss": 0.6291,
+      "step": 11162
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.34716510619595947,
+      "learning_rate": 3.6986023667751056e-10,
+      "loss": 0.6178,
+      "step": 11163
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3995399787946349,
+      "learning_rate": 3.3547423370583475e-10,
+      "loss": 0.6558,
+      "step": 11164
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3870971144898501,
+      "learning_rate": 3.02765578445996e-10,
+      "loss": 0.6226,
+      "step": 11165
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3740703834802233,
+      "learning_rate": 2.717342736380246e-10,
+      "loss": 0.6017,
+      "step": 11166
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4684021188529609,
+      "learning_rate": 2.423803218887244e-10,
+      "loss": 0.672,
+      "step": 11167
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3399966881272195,
+      "learning_rate": 2.1470372565390863e-10,
+      "loss": 0.621,
+      "step": 11168
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3825651392594987,
+      "learning_rate": 1.8870448726060474e-10,
+      "loss": 0.633,
+      "step": 11169
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3820118917570173,
+      "learning_rate": 1.6438260888707036e-10,
+      "loss": 0.6107,
+      "step": 11170
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.38498999182149585,
+      "learning_rate": 1.4173809257389538e-10,
+      "loss": 0.679,
+      "step": 11171
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.35690281604518415,
+      "learning_rate": 1.207709402173407e-10,
+      "loss": 0.6257,
+      "step": 11172
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.38249156495648046,
+      "learning_rate": 1.0148115357822008e-10,
+      "loss": 0.6655,
+      "step": 11173
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3641912730117998,
+      "learning_rate": 8.386873427523867e-11,
+      "loss": 0.6067,
+      "step": 11174
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.401256429341402,
+      "learning_rate": 6.79336837849931e-11,
+      "loss": 0.6297,
+      "step": 11175
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.42027177844683916,
+      "learning_rate": 5.367600344197144e-11,
+      "loss": 0.6567,
+      "step": 11176
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3458356258239677,
+      "learning_rate": 4.10956944429941e-11,
+      "loss": 0.56,
+      "step": 11177
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.40714444407687594,
+      "learning_rate": 3.0192757844993427e-11,
+      "loss": 0.637,
+      "step": 11178
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3829080527698318,
+      "learning_rate": 2.0967194560572723e-11,
+      "loss": 0.6566,
+      "step": 11179
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.3680279970499762,
+      "learning_rate": 1.3419005362447223e-11,
+      "loss": 6.4288,
+      "step": 11180
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.37082134283500745,
+      "learning_rate": 7.548190885664496e-12,
+      "loss": 0.6392,
+      "step": 11181
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3805053386155754,
+      "learning_rate": 3.3547516209431195e-12,
+      "loss": 0.6193,
+      "step": 11182
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6832338936400978,
+      "learning_rate": 8.386879235544598e-13,
+      "loss": 2.0539,
+      "step": 11183
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.4511987014053773,
+      "learning_rate": 0.0,
+      "loss": 4.8548,
+      "step": 11184
+    },
+    {
+      "epoch": 1.0,
+      "step": 11184,
+      "total_flos": 4525840170385408.0,
+      "train_loss": 0.028412628092820382,
+      "train_runtime": 5271.7341,
+      "train_samples_per_second": 271.547,
+      "train_steps_per_second": 2.122
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 11184,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 300,
+  "total_flos": 4525840170385408.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}