diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4281 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 606,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0049504950495049506,
+      "learning_rate": 0,
+      "loss": 2.1216,
+      "step": 1
+    },
+    {
+      "epoch": 0.009900990099009901,
+      "learning_rate": 0,
+      "loss": 2.1427,
+      "step": 2
+    },
+    {
+      "epoch": 0.01485148514851485,
+      "learning_rate": 0,
+      "loss": 2.1346,
+      "step": 3
+    },
+    {
+      "epoch": 0.019801980198019802,
+      "grad_norm": 2.259030342102051,
+      "learning_rate": 0.0,
+      "loss": 2.12,
+      "step": 4
+    },
+    {
+      "epoch": 0.024752475247524754,
+      "grad_norm": 2.259030342102051,
+      "learning_rate": 0.0,
+      "loss": 2.1802,
+      "step": 5
+    },
+    {
+      "epoch": 0.0297029702970297,
+      "grad_norm": 2.259030342102051,
+      "learning_rate": 0.0,
+      "loss": 2.1452,
+      "step": 6
+    },
+    {
+      "epoch": 0.034653465346534656,
+      "grad_norm": 2.259030342102051,
+      "learning_rate": 0.0,
+      "loss": 2.1545,
+      "step": 7
+    },
+    {
+      "epoch": 0.039603960396039604,
+      "grad_norm": 2.1996872425079346,
+      "learning_rate": 3.3722619737900224e-06,
+      "loss": 2.1611,
+      "step": 8
+    },
+    {
+      "epoch": 0.04455445544554455,
+      "grad_norm": 2.1996872425079346,
+      "learning_rate": 3.3722619737900224e-06,
+      "loss": 2.2054,
+      "step": 9
+    },
+    {
+      "epoch": 0.04950495049504951,
+      "grad_norm": 2.1996872425079346,
+      "learning_rate": 3.3722619737900224e-06,
+      "loss": 2.1423,
+      "step": 10
+    },
+    {
+      "epoch": 0.054455445544554455,
+      "grad_norm": 2.1996872425079346,
+      "learning_rate": 3.3722619737900224e-06,
+      "loss": 2.1626,
+      "step": 11
+    },
+    {
+      "epoch": 0.0594059405940594,
+      "grad_norm": 2.2271275520324707,
+      "learning_rate": 5.344908771065097e-06,
+      "loss": 2.1728,
+      "step": 12
+    },
+    {
+      "epoch": 0.06435643564356436,
+      "grad_norm": 2.2271275520324707,
+      "learning_rate": 5.344908771065097e-06,
+      "loss": 2.1859,
+      "step": 13
+    },
+    {
+      "epoch": 0.06930693069306931,
+      "grad_norm": 2.2271275520324707,
+      "learning_rate": 5.344908771065097e-06,
+      "loss": 2.1523,
+      "step": 14
+    },
+    {
+      "epoch": 0.07425742574257425,
+      "grad_norm": 2.2271275520324707,
+      "learning_rate": 5.344908771065097e-06,
+      "loss": 2.2623,
+      "step": 15
+    },
+    {
+      "epoch": 0.07920792079207921,
+      "grad_norm": 2.2643508911132812,
+      "learning_rate": 6.744523947580045e-06,
+      "loss": 2.2004,
+      "step": 16
+    },
+    {
+      "epoch": 0.08415841584158416,
+      "grad_norm": 2.2643508911132812,
+      "learning_rate": 6.744523947580045e-06,
+      "loss": 2.079,
+      "step": 17
+    },
+    {
+      "epoch": 0.0891089108910891,
+      "grad_norm": 2.2643508911132812,
+      "learning_rate": 6.744523947580045e-06,
+      "loss": 2.1093,
+      "step": 18
+    },
+    {
+      "epoch": 0.09405940594059406,
+      "grad_norm": 2.2643508911132812,
+      "learning_rate": 6.744523947580045e-06,
+      "loss": 2.172,
+      "step": 19
+    },
+    {
+      "epoch": 0.09900990099009901,
+      "grad_norm": 2.128260850906372,
+      "learning_rate": 7.830149820263363e-06,
+      "loss": 2.171,
+      "step": 20
+    },
+    {
+      "epoch": 0.10396039603960396,
+      "grad_norm": 2.128260850906372,
+      "learning_rate": 7.830149820263363e-06,
+      "loss": 2.2302,
+      "step": 21
+    },
+    {
+      "epoch": 0.10891089108910891,
+      "grad_norm": 2.128260850906372,
+      "learning_rate": 7.830149820263363e-06,
+      "loss": 2.1686,
+      "step": 22
+    },
+    {
+      "epoch": 0.11386138613861387,
+      "grad_norm": 2.128260850906372,
+      "learning_rate": 7.830149820263363e-06,
+      "loss": 2.1244,
+      "step": 23
+    },
+    {
+      "epoch": 0.1188118811881188,
+      "grad_norm": 2.14156174659729,
+      "learning_rate": 8.717170744855119e-06,
+      "loss": 2.2406,
+      "step": 24
+    },
+    {
+      "epoch": 0.12376237623762376,
+      "grad_norm": 2.14156174659729,
+      "learning_rate": 8.717170744855119e-06,
+      "loss": 2.1582,
+      "step": 25
+    },
+    {
+      "epoch": 0.12871287128712872,
+      "grad_norm": 2.14156174659729,
+      "learning_rate": 8.717170744855119e-06,
+      "loss": 2.1773,
+      "step": 26
+    },
+    {
+      "epoch": 0.13366336633663367,
+      "grad_norm": 2.14156174659729,
+      "learning_rate": 8.717170744855119e-06,
+      "loss": 2.0421,
+      "step": 27
+    },
+    {
+      "epoch": 0.13861386138613863,
+      "grad_norm": 2.0065624713897705,
+      "learning_rate": 9.46713625058711e-06,
+      "loss": 2.1027,
+      "step": 28
+    },
+    {
+      "epoch": 0.14356435643564355,
+      "grad_norm": 2.0065624713897705,
+      "learning_rate": 9.46713625058711e-06,
+      "loss": 1.9643,
+      "step": 29
+    },
+    {
+      "epoch": 0.1485148514851485,
+      "grad_norm": 2.0065624713897705,
+      "learning_rate": 9.46713625058711e-06,
+      "loss": 2.0852,
+      "step": 30
+    },
+    {
+      "epoch": 0.15346534653465346,
+      "grad_norm": 2.0065624713897705,
+      "learning_rate": 9.46713625058711e-06,
+      "loss": 2.0588,
+      "step": 31
+    },
+    {
+      "epoch": 0.15841584158415842,
+      "grad_norm": 1.8585381507873535,
+      "learning_rate": 1.0116785921370066e-05,
+      "loss": 2.1563,
+      "step": 32
+    },
+    {
+      "epoch": 0.16336633663366337,
+      "grad_norm": 1.8585381507873535,
+      "learning_rate": 1.0116785921370066e-05,
+      "loss": 1.9961,
+      "step": 33
+    },
+    {
+      "epoch": 0.16831683168316833,
+      "grad_norm": 1.8585381507873535,
+      "learning_rate": 1.0116785921370066e-05,
+      "loss": 2.1021,
+      "step": 34
+    },
+    {
+      "epoch": 0.17326732673267325,
+      "grad_norm": 1.8585381507873535,
+      "learning_rate": 1.0116785921370066e-05,
+      "loss": 2.0583,
+      "step": 35
+    },
+    {
+      "epoch": 0.1782178217821782,
+      "grad_norm": 1.813769817352295,
+      "learning_rate": 1.0689817542130194e-05,
+      "loss": 1.9911,
+      "step": 36
+    },
+    {
+      "epoch": 0.18316831683168316,
+      "grad_norm": 1.813769817352295,
+      "learning_rate": 1.0689817542130194e-05,
+      "loss": 2.024,
+      "step": 37
+    },
+    {
+      "epoch": 0.18811881188118812,
+      "grad_norm": 1.813769817352295,
+      "learning_rate": 1.0689817542130194e-05,
+      "loss": 2.0567,
+      "step": 38
+    },
+    {
+      "epoch": 0.19306930693069307,
+      "grad_norm": 1.813769817352295,
+      "learning_rate": 1.0689817542130194e-05,
+      "loss": 2.0862,
+      "step": 39
+    },
+    {
+      "epoch": 0.19801980198019803,
+      "grad_norm": 1.8762847185134888,
+      "learning_rate": 1.1202411794053388e-05,
+      "loss": 2.0415,
+      "step": 40
+    },
+    {
+      "epoch": 0.20297029702970298,
+      "grad_norm": 1.8762847185134888,
+      "learning_rate": 1.1202411794053388e-05,
+      "loss": 1.8461,
+      "step": 41
+    },
+    {
+      "epoch": 0.2079207920792079,
+      "grad_norm": 1.8762847185134888,
+      "learning_rate": 1.1202411794053388e-05,
+      "loss": 1.9262,
+      "step": 42
+    },
+    {
+      "epoch": 0.21287128712871287,
+      "grad_norm": 1.8762847185134888,
+      "learning_rate": 1.1202411794053388e-05,
+      "loss": 1.9178,
+      "step": 43
+    },
+    {
+      "epoch": 0.21782178217821782,
+      "grad_norm": 2.218778610229492,
+      "learning_rate": 1.1666109698457423e-05,
+      "loss": 1.8544,
+      "step": 44
+    },
+    {
+      "epoch": 0.22277227722772278,
+      "grad_norm": 2.218778610229492,
+      "learning_rate": 1.1666109698457423e-05,
+      "loss": 1.968,
+      "step": 45
+    },
+    {
+      "epoch": 0.22772277227722773,
+      "grad_norm": 2.218778610229492,
+      "learning_rate": 1.1666109698457423e-05,
+      "loss": 1.9128,
+      "step": 46
+    },
+    {
+      "epoch": 0.23267326732673269,
+      "grad_norm": 2.218778610229492,
+      "learning_rate": 1.1666109698457423e-05,
+      "loss": 1.8448,
+      "step": 47
+    },
+    {
+      "epoch": 0.2376237623762376,
+      "grad_norm": 1.9425584077835083,
+      "learning_rate": 1.2089432718645142e-05,
+      "loss": 1.8355,
+      "step": 48
+    },
+    {
+      "epoch": 0.24257425742574257,
+      "grad_norm": 1.9425584077835083,
+      "learning_rate": 1.2089432718645142e-05,
+      "loss": 1.8293,
+      "step": 49
+    },
+    {
+      "epoch": 0.24752475247524752,
+      "grad_norm": 1.9425584077835083,
+      "learning_rate": 1.2089432718645142e-05,
+      "loss": 1.8464,
+      "step": 50
+    },
+    {
+      "epoch": 0.2524752475247525,
+      "grad_norm": 1.9425584077835083,
+      "learning_rate": 1.2089432718645142e-05,
+      "loss": 2.0073,
+      "step": 51
+    },
+    {
+      "epoch": 0.25742574257425743,
+      "grad_norm": 1.4947453737258911,
+      "learning_rate": 1.2478852147789474e-05,
+      "loss": 1.8608,
+      "step": 52
+    },
+    {
+      "epoch": 0.2623762376237624,
+      "grad_norm": 1.4947453737258911,
+      "learning_rate": 1.2478852147789474e-05,
+      "loss": 1.858,
+      "step": 53
+    },
+    {
+      "epoch": 0.26732673267326734,
+      "grad_norm": 1.4947453737258911,
+      "learning_rate": 1.2478852147789474e-05,
+      "loss": 1.9059,
+      "step": 54
+    },
+    {
+      "epoch": 0.2722772277227723,
+      "grad_norm": 1.4947453737258911,
+      "learning_rate": 1.2478852147789474e-05,
+      "loss": 1.8371,
+      "step": 55
+    },
+    {
+      "epoch": 0.27722772277227725,
+      "grad_norm": 1.1246448755264282,
+      "learning_rate": 1.2839398224377133e-05,
+      "loss": 1.8752,
+      "step": 56
+    },
+    {
+      "epoch": 0.28217821782178215,
+      "grad_norm": 1.1246448755264282,
+      "learning_rate": 1.2839398224377133e-05,
+      "loss": 1.8094,
+      "step": 57
+    },
+    {
+      "epoch": 0.2871287128712871,
+      "grad_norm": 1.1246448755264282,
+      "learning_rate": 1.2839398224377133e-05,
+      "loss": 1.7313,
+      "step": 58
+    },
+    {
+      "epoch": 0.29207920792079206,
+      "grad_norm": 1.1246448755264282,
+      "learning_rate": 1.2839398224377133e-05,
+      "loss": 1.8594,
+      "step": 59
+    },
+    {
+      "epoch": 0.297029702970297,
+      "grad_norm": 1.0055837631225586,
+      "learning_rate": 1.317505859132846e-05,
+      "loss": 1.7862,
+      "step": 60
+    },
+    {
+      "epoch": 0.30198019801980197,
+      "grad_norm": 1.0055837631225586,
+      "learning_rate": 1.317505859132846e-05,
+      "loss": 1.7247,
+      "step": 61
+    },
+    {
+      "epoch": 0.3069306930693069,
+      "grad_norm": 1.0055837631225586,
+      "learning_rate": 1.317505859132846e-05,
+      "loss": 1.7997,
+      "step": 62
+    },
+    {
+      "epoch": 0.3118811881188119,
+      "grad_norm": 1.0055837631225586,
+      "learning_rate": 1.317505859132846e-05,
+      "loss": 1.7965,
+      "step": 63
+    },
+    {
+      "epoch": 0.31683168316831684,
+      "grad_norm": 0.931380033493042,
+      "learning_rate": 1.348904789516009e-05,
+      "loss": 1.7517,
+      "step": 64
+    },
+    {
+      "epoch": 0.3217821782178218,
+      "grad_norm": 0.931380033493042,
+      "learning_rate": 1.348904789516009e-05,
+      "loss": 1.8095,
+      "step": 65
+    },
+    {
+      "epoch": 0.32673267326732675,
+      "grad_norm": 0.931380033493042,
+      "learning_rate": 1.348904789516009e-05,
+      "loss": 1.8155,
+      "step": 66
+    },
+    {
+      "epoch": 0.3316831683168317,
+      "grad_norm": 0.931380033493042,
+      "learning_rate": 1.348904789516009e-05,
+      "loss": 1.7587,
+      "step": 67
+    },
+    {
+      "epoch": 0.33663366336633666,
+      "grad_norm": 1.029356837272644,
+      "learning_rate": 1.3783995508828243e-05,
+      "loss": 1.8325,
+      "step": 68
+    },
+    {
+      "epoch": 0.3415841584158416,
+      "grad_norm": 1.029356837272644,
+      "learning_rate": 1.3783995508828243e-05,
+      "loss": 1.7398,
+      "step": 69
+    },
+    {
+      "epoch": 0.3465346534653465,
+      "grad_norm": 1.029356837272644,
+      "learning_rate": 1.3783995508828243e-05,
+      "loss": 1.8004,
+      "step": 70
+    },
+    {
+      "epoch": 0.35148514851485146,
+      "grad_norm": 1.029356837272644,
+      "learning_rate": 1.3783995508828243e-05,
+      "loss": 1.7193,
+      "step": 71
+    },
+    {
+      "epoch": 0.3564356435643564,
+      "grad_norm": 0.8115963935852051,
+      "learning_rate": 1.4062079515920212e-05,
+      "loss": 1.7698,
+      "step": 72
+    },
+    {
+      "epoch": 0.3613861386138614,
+      "grad_norm": 0.8115963935852051,
+      "learning_rate": 1.4062079515920212e-05,
+      "loss": 1.7462,
+      "step": 73
+    },
+    {
+      "epoch": 0.36633663366336633,
+      "grad_norm": 0.8115963935852051,
+      "learning_rate": 1.4062079515920212e-05,
+      "loss": 1.7259,
+      "step": 74
+    },
+    {
+      "epoch": 0.3712871287128713,
+      "grad_norm": 0.8115963935852051,
+      "learning_rate": 1.4062079515920212e-05,
+      "loss": 1.7222,
+      "step": 75
+    },
+    {
+      "epoch": 0.37623762376237624,
+      "grad_norm": 0.6555297374725342,
+      "learning_rate": 1.4325124421002207e-05,
+      "loss": 1.7193,
+      "step": 76
+    },
+    {
+      "epoch": 0.3811881188118812,
+      "grad_norm": 0.6555297374725342,
+      "learning_rate": 1.4325124421002207e-05,
+      "loss": 1.692,
+      "step": 77
+    },
+    {
+      "epoch": 0.38613861386138615,
+      "grad_norm": 0.6555297374725342,
+      "learning_rate": 1.4325124421002207e-05,
+      "loss": 1.7357,
+      "step": 78
+    },
+    {
+      "epoch": 0.3910891089108911,
+      "grad_norm": 0.6555297374725342,
+      "learning_rate": 1.4325124421002207e-05,
+      "loss": 1.6804,
+      "step": 79
+    },
+    {
+      "epoch": 0.39603960396039606,
+      "grad_norm": 0.6669710278511047,
+      "learning_rate": 1.4574673767843407e-05,
+      "loss": 1.7015,
+      "step": 80
+    },
+    {
+      "epoch": 0.400990099009901,
+      "grad_norm": 0.6669710278511047,
+      "learning_rate": 1.4574673767843407e-05,
+      "loss": 1.7281,
+      "step": 81
+    },
+    {
+      "epoch": 0.40594059405940597,
+      "grad_norm": 0.6669710278511047,
+      "learning_rate": 1.4574673767843407e-05,
+      "loss": 1.7246,
+      "step": 82
+    },
+    {
+      "epoch": 0.41089108910891087,
+      "grad_norm": 0.6669710278511047,
+      "learning_rate": 1.4574673767843407e-05,
+      "loss": 1.7016,
+      "step": 83
+    },
+    {
+      "epoch": 0.4158415841584158,
+      "grad_norm": 0.5633140802383423,
+      "learning_rate": 1.4812045021652206e-05,
+      "loss": 1.7358,
+      "step": 84
+    },
+    {
+      "epoch": 0.4207920792079208,
+      "grad_norm": 0.5633140802383423,
+      "learning_rate": 1.4812045021652206e-05,
+      "loss": 1.7006,
+      "step": 85
+    },
+    {
+      "epoch": 0.42574257425742573,
+      "grad_norm": 0.5633140802383423,
+      "learning_rate": 1.4812045021652206e-05,
+      "loss": 1.6776,
+      "step": 86
+    },
+    {
+      "epoch": 0.4306930693069307,
+      "grad_norm": 0.5633140802383423,
+      "learning_rate": 1.4812045021652206e-05,
+      "loss": 1.6905,
+      "step": 87
+    },
+    {
+      "epoch": 0.43564356435643564,
+      "grad_norm": 0.5185403227806091,
+      "learning_rate": 1.5038371672247447e-05,
+      "loss": 1.7344,
+      "step": 88
+    },
+    {
+      "epoch": 0.4405940594059406,
+      "grad_norm": 0.5185403227806091,
+      "learning_rate": 1.5038371672247447e-05,
+      "loss": 1.6872,
+      "step": 89
+    },
+    {
+      "epoch": 0.44554455445544555,
+      "grad_norm": 0.5185403227806091,
+      "learning_rate": 1.5038371672247447e-05,
+      "loss": 1.6632,
+      "step": 90
+    },
+    {
+      "epoch": 0.4504950495049505,
+      "grad_norm": 0.5185403227806091,
+      "learning_rate": 1.5038371672247447e-05,
+      "loss": 1.7155,
+      "step": 91
+    },
+    {
+      "epoch": 0.45544554455445546,
+      "grad_norm": 0.4828384518623352,
+      "learning_rate": 1.5254635970494278e-05,
+      "loss": 1.6871,
+      "step": 92
+    },
+    {
+      "epoch": 0.4603960396039604,
+      "grad_norm": 0.4828384518623352,
+      "learning_rate": 1.5254635970494278e-05,
+      "loss": 1.6906,
+      "step": 93
+    },
+    {
+      "epoch": 0.46534653465346537,
+      "grad_norm": 0.4828384518623352,
+      "learning_rate": 1.5254635970494278e-05,
+      "loss": 1.6122,
+      "step": 94
+    },
+    {
+      "epoch": 0.47029702970297027,
+      "grad_norm": 0.4828384518623352,
+      "learning_rate": 1.5254635970494278e-05,
+      "loss": 1.7374,
+      "step": 95
+    },
+    {
+      "epoch": 0.4752475247524752,
+      "grad_norm": 0.4721103310585022,
+      "learning_rate": 1.5461694692435165e-05,
+      "loss": 1.6655,
+      "step": 96
+    },
+    {
+      "epoch": 0.4801980198019802,
+      "grad_norm": 0.4721103310585022,
+      "learning_rate": 1.5461694692435165e-05,
+      "loss": 1.5872,
+      "step": 97
+    },
+    {
+      "epoch": 0.48514851485148514,
+      "grad_norm": 0.4721103310585022,
+      "learning_rate": 1.5461694692435165e-05,
+      "loss": 1.6449,
+      "step": 98
+    },
+    {
+      "epoch": 0.4900990099009901,
+      "grad_norm": 0.4721103310585022,
+      "learning_rate": 1.5461694692435165e-05,
+      "loss": 1.6695,
+      "step": 99
+    },
+    {
+      "epoch": 0.49504950495049505,
+      "grad_norm": 0.45932599902153015,
+      "learning_rate": 1.5660299640526725e-05,
+      "loss": 1.6836,
+      "step": 100
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.45932599902153015,
+      "learning_rate": 1.5660299640526725e-05,
+      "loss": 1.6853,
+      "step": 101
+    },
+    {
+      "epoch": 0.504950495049505,
+      "grad_norm": 0.45932599902153015,
+      "learning_rate": 1.5660299640526725e-05,
+      "loss": 1.6404,
+      "step": 102
+    },
+    {
+      "epoch": 0.5099009900990099,
+      "grad_norm": 0.45932599902153015,
+      "learning_rate": 1.5660299640526725e-05,
+      "loss": 1.7104,
+      "step": 103
+    },
+    {
+      "epoch": 0.5148514851485149,
+      "grad_norm": 0.44053927063941956,
+      "learning_rate": 1.5851114121579497e-05,
+      "loss": 1.6309,
+      "step": 104
+    },
+    {
+      "epoch": 0.5198019801980198,
+      "grad_norm": 0.44053927063941956,
+      "learning_rate": 1.5851114121579497e-05,
+      "loss": 1.6296,
+      "step": 105
+    },
+    {
+      "epoch": 0.5247524752475248,
+      "grad_norm": 0.44053927063941956,
+      "learning_rate": 1.5851114121579497e-05,
+      "loss": 1.685,
+      "step": 106
+    },
+    {
+      "epoch": 0.5297029702970297,
+      "grad_norm": 0.44053927063941956,
+      "learning_rate": 1.5851114121579497e-05,
+      "loss": 1.6317,
+      "step": 107
+    },
+    {
+      "epoch": 0.5346534653465347,
+      "grad_norm": 0.4426601827144623,
+      "learning_rate": 1.603472631319529e-05,
+      "loss": 1.621,
+      "step": 108
+    },
+    {
+      "epoch": 0.5396039603960396,
+      "grad_norm": 0.4426601827144623,
+      "learning_rate": 1.603472631319529e-05,
+      "loss": 1.5765,
+      "step": 109
+    },
+    {
+      "epoch": 0.5445544554455446,
+      "grad_norm": 0.4426601827144623,
+      "learning_rate": 1.603472631319529e-05,
+      "loss": 1.6865,
+      "step": 110
+    },
+    {
+      "epoch": 0.5495049504950495,
+      "grad_norm": 0.4426601827144623,
+      "learning_rate": 1.603472631319529e-05,
+      "loss": 1.586,
+      "step": 111
+    },
+    {
+      "epoch": 0.5544554455445545,
+      "grad_norm": 0.43527039885520935,
+      "learning_rate": 1.6211660198167156e-05,
+      "loss": 1.6683,
+      "step": 112
+    },
+    {
+      "epoch": 0.5594059405940595,
+      "grad_norm": 0.43527039885520935,
+      "learning_rate": 1.6211660198167156e-05,
+      "loss": 1.5759,
+      "step": 113
+    },
+    {
+      "epoch": 0.5643564356435643,
+      "grad_norm": 0.43527039885520935,
+      "learning_rate": 1.6211660198167156e-05,
+      "loss": 1.6003,
+      "step": 114
+    },
+    {
+      "epoch": 0.5693069306930693,
+      "grad_norm": 0.43527039885520935,
+      "learning_rate": 1.6211660198167156e-05,
+      "loss": 1.5873,
+      "step": 115
+    },
+    {
+      "epoch": 0.5742574257425742,
+      "grad_norm": 0.4649295508861542,
+      "learning_rate": 1.6382384579263326e-05,
+      "loss": 1.6703,
+      "step": 116
+    },
+    {
+      "epoch": 0.5792079207920792,
+      "grad_norm": 0.4649295508861542,
+      "learning_rate": 1.6382384579263326e-05,
+      "loss": 1.6352,
+      "step": 117
+    },
+    {
+      "epoch": 0.5841584158415841,
+      "grad_norm": 0.4649295508861542,
+      "learning_rate": 1.6382384579263326e-05,
+      "loss": 1.6522,
+      "step": 118
+    },
+    {
+      "epoch": 0.5891089108910891,
+      "grad_norm": 0.4649295508861542,
+      "learning_rate": 1.6382384579263326e-05,
+      "loss": 1.6246,
+      "step": 119
+    },
+    {
+      "epoch": 0.594059405940594,
+      "grad_norm": 0.42726704478263855,
+      "learning_rate": 1.6547320565118483e-05,
+      "loss": 1.6479,
+      "step": 120
+    },
+    {
+      "epoch": 0.599009900990099,
+      "grad_norm": 0.42726704478263855,
+      "learning_rate": 1.6547320565118483e-05,
+      "loss": 1.6093,
+      "step": 121
+    },
+    {
+      "epoch": 0.6039603960396039,
+      "grad_norm": 0.42726704478263855,
+      "learning_rate": 1.6547320565118483e-05,
+      "loss": 1.5866,
+      "step": 122
+    },
+    {
+      "epoch": 0.6089108910891089,
+      "grad_norm": 0.42726704478263855,
+      "learning_rate": 1.6547320565118483e-05,
+      "loss": 1.646,
+      "step": 123
+    },
+    {
+      "epoch": 0.6138613861386139,
+      "grad_norm": 0.4181938171386719,
+      "learning_rate": 1.670684782820849e-05,
+      "loss": 1.6139,
+      "step": 124
+    },
+    {
+      "epoch": 0.6188118811881188,
+      "grad_norm": 0.4181938171386719,
+      "learning_rate": 1.670684782820849e-05,
+      "loss": 1.6139,
+      "step": 125
+    },
+    {
+      "epoch": 0.6237623762376238,
+      "grad_norm": 0.4181938171386719,
+      "learning_rate": 1.670684782820849e-05,
+      "loss": 1.6142,
+      "step": 126
+    },
+    {
+      "epoch": 0.6287128712871287,
+      "grad_norm": 0.4181938171386719,
+      "learning_rate": 1.670684782820849e-05,
+      "loss": 1.6083,
+      "step": 127
+    },
+    {
+      "epoch": 0.6336633663366337,
+      "grad_norm": 0.40746423602104187,
+      "learning_rate": 1.6861309868950113e-05,
+      "loss": 1.5874,
+      "step": 128
+    },
+    {
+      "epoch": 0.6386138613861386,
+      "grad_norm": 0.40746423602104187,
+      "learning_rate": 1.6861309868950113e-05,
+      "loss": 1.6076,
+      "step": 129
+    },
+    {
+      "epoch": 0.6435643564356436,
+      "grad_norm": 0.40746423602104187,
+      "learning_rate": 1.6861309868950113e-05,
+      "loss": 1.578,
+      "step": 130
+    },
+    {
+      "epoch": 0.6485148514851485,
+      "grad_norm": 0.40746423602104187,
+      "learning_rate": 1.6861309868950113e-05,
+      "loss": 1.5159,
+      "step": 131
+    },
+    {
+      "epoch": 0.6534653465346535,
+      "grad_norm": 0.4084773659706116,
+      "learning_rate": 1.701101846952252e-05,
+      "loss": 1.5989,
+      "step": 132
+    },
+    {
+      "epoch": 0.6584158415841584,
+      "grad_norm": 0.4084773659706116,
+      "learning_rate": 1.701101846952252e-05,
+      "loss": 1.6173,
+      "step": 133
+    },
+    {
+      "epoch": 0.6633663366336634,
+      "grad_norm": 0.4084773659706116,
+      "learning_rate": 1.701101846952252e-05,
+      "loss": 1.6067,
+      "step": 134
+    },
+    {
+      "epoch": 0.6683168316831684,
+      "grad_norm": 0.4084773659706116,
+      "learning_rate": 1.701101846952252e-05,
+      "loss": 1.571,
+      "step": 135
+    },
+    {
+      "epoch": 0.6732673267326733,
+      "grad_norm": 0.3729560971260071,
+      "learning_rate": 1.7156257482618265e-05,
+      "loss": 1.561,
+      "step": 136
+    },
+    {
+      "epoch": 0.6782178217821783,
+      "grad_norm": 0.3729560971260071,
+      "learning_rate": 1.7156257482618265e-05,
+      "loss": 1.5984,
+      "step": 137
+    },
+    {
+      "epoch": 0.6831683168316832,
+      "grad_norm": 0.3729560971260071,
+      "learning_rate": 1.7156257482618265e-05,
+      "loss": 1.5748,
+      "step": 138
+    },
+    {
+      "epoch": 0.6881188118811881,
+      "grad_norm": 0.3729560971260071,
+      "learning_rate": 1.7156257482618265e-05,
+      "loss": 1.5892,
+      "step": 139
+    },
+    {
+      "epoch": 0.693069306930693,
+      "grad_norm": 0.36557894945144653,
+      "learning_rate": 1.7297286070850474e-05,
+      "loss": 1.5574,
+      "step": 140
+    },
+    {
+      "epoch": 0.698019801980198,
+      "grad_norm": 0.36557894945144653,
+      "learning_rate": 1.7297286070850474e-05,
+      "loss": 1.5777,
+      "step": 141
+    },
+    {
+      "epoch": 0.7029702970297029,
+      "grad_norm": 0.36557894945144653,
+      "learning_rate": 1.7297286070850474e-05,
+      "loss": 1.4983,
+      "step": 142
+    },
+    {
+      "epoch": 0.7079207920792079,
+      "grad_norm": 0.36557894945144653,
+      "learning_rate": 1.7297286070850474e-05,
+      "loss": 1.5714,
+      "step": 143
+    },
+    {
+      "epoch": 0.7128712871287128,
+      "grad_norm": 0.3284773528575897,
+      "learning_rate": 1.7434341489710237e-05,
+      "loss": 1.5375,
+      "step": 144
+    },
+    {
+      "epoch": 0.7178217821782178,
+      "grad_norm": 0.3284773528575897,
+      "learning_rate": 1.7434341489710237e-05,
+      "loss": 1.5888,
+      "step": 145
+    },
+    {
+      "epoch": 0.7227722772277227,
+      "grad_norm": 0.3284773528575897,
+      "learning_rate": 1.7434341489710237e-05,
+      "loss": 1.5763,
+      "step": 146
+    },
+    {
+      "epoch": 0.7277227722772277,
+      "grad_norm": 0.3284773528575897,
+      "learning_rate": 1.7434341489710237e-05,
+      "loss": 1.5336,
+      "step": 147
+    },
+    {
+      "epoch": 0.7326732673267327,
+      "grad_norm": 0.3244377672672272,
+      "learning_rate": 1.7567641489142956e-05,
+      "loss": 1.5122,
+      "step": 148
+    },
+    {
+      "epoch": 0.7376237623762376,
+      "grad_norm": 0.3244377672672272,
+      "learning_rate": 1.7567641489142956e-05,
+      "loss": 1.5999,
+      "step": 149
+    },
+    {
+      "epoch": 0.7425742574257426,
+      "grad_norm": 0.3244377672672272,
+      "learning_rate": 1.7567641489142956e-05,
+      "loss": 1.5643,
+      "step": 150
+    },
+    {
+      "epoch": 0.7475247524752475,
+      "grad_norm": 0.3244377672672272,
+      "learning_rate": 1.7567641489142956e-05,
+      "loss": 1.5742,
+      "step": 151
+    },
+    {
+      "epoch": 0.7524752475247525,
+      "grad_norm": 0.33302003145217896,
+      "learning_rate": 1.769738639479223e-05,
+      "loss": 1.6125,
+      "step": 152
+    },
+    {
+      "epoch": 0.7574257425742574,
+      "grad_norm": 0.33302003145217896,
+      "learning_rate": 1.769738639479223e-05,
+      "loss": 1.5294,
+      "step": 153
+    },
+    {
+      "epoch": 0.7623762376237624,
+      "grad_norm": 0.33302003145217896,
+      "learning_rate": 1.769738639479223e-05,
+      "loss": 1.6211,
+      "step": 154
+    },
+    {
+      "epoch": 0.7673267326732673,
+      "grad_norm": 0.33302003145217896,
+      "learning_rate": 1.769738639479223e-05,
+      "loss": 1.5182,
+      "step": 155
+    },
+    {
+      "epoch": 0.7722772277227723,
+      "grad_norm": 0.31815779209136963,
+      "learning_rate": 1.782376091885457e-05,
+      "loss": 1.5432,
+      "step": 156
+    },
+    {
+      "epoch": 0.7772277227722773,
+      "grad_norm": 0.31815779209136963,
+      "learning_rate": 1.782376091885457e-05,
+      "loss": 1.6547,
+      "step": 157
+    },
+    {
+      "epoch": 0.7821782178217822,
+      "grad_norm": 0.31815779209136963,
+      "learning_rate": 1.782376091885457e-05,
+      "loss": 1.5951,
+      "step": 158
+    },
+    {
+      "epoch": 0.7871287128712872,
+      "grad_norm": 0.31815779209136963,
+      "learning_rate": 1.782376091885457e-05,
+      "loss": 1.5594,
+      "step": 159
+    },
+    {
+      "epoch": 0.7920792079207921,
+      "grad_norm": 0.30909690260887146,
+      "learning_rate": 1.794693574163343e-05,
+      "loss": 1.5609,
+      "step": 160
+    },
+    {
+      "epoch": 0.7970297029702971,
+      "grad_norm": 0.30909690260887146,
+      "learning_rate": 1.794693574163343e-05,
+      "loss": 1.5435,
+      "step": 161
+    },
+    {
+      "epoch": 0.801980198019802,
+      "grad_norm": 0.30909690260887146,
+      "learning_rate": 1.794693574163343e-05,
+      "loss": 1.584,
+      "step": 162
+    },
+    {
+      "epoch": 0.806930693069307,
+      "grad_norm": 0.30909690260887146,
+      "learning_rate": 1.794693574163343e-05,
+      "loss": 1.5258,
+      "step": 163
+    },
+    {
+      "epoch": 0.8118811881188119,
+      "grad_norm": 0.3037148118019104,
+      "learning_rate": 1.8067068897776073e-05,
+      "loss": 1.6069,
+      "step": 164
+    },
+    {
+      "epoch": 0.8168316831683168,
+      "grad_norm": 0.3037148118019104,
+      "learning_rate": 1.8067068897776073e-05,
+      "loss": 1.5888,
+      "step": 165
+    },
+    {
+      "epoch": 0.8217821782178217,
+      "grad_norm": 0.3037148118019104,
+      "learning_rate": 1.8067068897776073e-05,
+      "loss": 1.5755,
+      "step": 166
+    },
+    {
+      "epoch": 0.8267326732673267,
+      "grad_norm": 0.3037148118019104,
+      "learning_rate": 1.8067068897776073e-05,
+      "loss": 1.6131,
+      "step": 167
+    },
+    {
+      "epoch": 0.8316831683168316,
+      "grad_norm": 0.30614173412323,
+      "learning_rate": 1.8184306995442228e-05,
+      "loss": 1.5445,
+      "step": 168
+    },
+    {
+      "epoch": 0.8366336633663366,
+      "grad_norm": 0.30614173412323,
+      "learning_rate": 1.8184306995442228e-05,
+      "loss": 1.5598,
+      "step": 169
+    },
+    {
+      "epoch": 0.8415841584158416,
+      "grad_norm": 0.30614173412323,
+      "learning_rate": 1.8184306995442228e-05,
+      "loss": 1.5947,
+      "step": 170
+    },
+    {
+      "epoch": 0.8465346534653465,
+      "grad_norm": 0.30614173412323,
+      "learning_rate": 1.8184306995442228e-05,
+      "loss": 1.5464,
+      "step": 171
+    },
+    {
+      "epoch": 0.8514851485148515,
+      "grad_norm": 0.2994905412197113,
+      "learning_rate": 1.829878629199893e-05,
+      "loss": 1.5968,
+      "step": 172
+    },
+    {
+      "epoch": 0.8564356435643564,
+      "grad_norm": 0.2994905412197113,
+      "learning_rate": 1.829878629199893e-05,
+      "loss": 1.5762,
+      "step": 173
+    },
+    {
+      "epoch": 0.8613861386138614,
+      "grad_norm": 0.2994905412197113,
+      "learning_rate": 1.829878629199893e-05,
+      "loss": 1.6057,
+      "step": 174
+    },
+    {
+      "epoch": 0.8663366336633663,
+      "grad_norm": 0.2994905412197113,
+      "learning_rate": 1.829878629199893e-05,
+      "loss": 1.5344,
+      "step": 175
+    },
+    {
+      "epoch": 0.8712871287128713,
+      "grad_norm": 0.29959049820899963,
+      "learning_rate": 1.8410633646037467e-05,
+      "loss": 1.645,
+      "step": 176
+    },
+    {
+      "epoch": 0.8762376237623762,
+      "grad_norm": 0.29959049820899963,
+      "learning_rate": 1.8410633646037467e-05,
+      "loss": 1.527,
+      "step": 177
+    },
+    {
+      "epoch": 0.8811881188118812,
+      "grad_norm": 0.29959049820899963,
+      "learning_rate": 1.8410633646037467e-05,
+      "loss": 1.5353,
+      "step": 178
+    },
+    {
+      "epoch": 0.8861386138613861,
+      "grad_norm": 0.29959049820899963,
+      "learning_rate": 1.8410633646037467e-05,
+      "loss": 1.528,
+      "step": 179
+    },
+    {
+      "epoch": 0.8910891089108911,
+      "grad_norm": 0.2880352735519409,
+      "learning_rate": 1.8519967362393555e-05,
+      "loss": 1.5629,
+      "step": 180
+    },
+    {
+      "epoch": 0.8960396039603961,
+      "grad_norm": 0.2880352735519409,
+      "learning_rate": 1.8519967362393555e-05,
+      "loss": 1.5101,
+      "step": 181
+    },
+    {
+      "epoch": 0.900990099009901,
+      "grad_norm": 0.2880352735519409,
+      "learning_rate": 1.8519967362393555e-05,
+      "loss": 1.531,
+      "step": 182
+    },
+    {
+      "epoch": 0.905940594059406,
+      "grad_norm": 0.2880352735519409,
+      "learning_rate": 1.8519967362393555e-05,
+      "loss": 1.4733,
+      "step": 183
+    },
+    {
+      "epoch": 0.9108910891089109,
+      "grad_norm": 0.3000313341617584,
+      "learning_rate": 1.86268979442843e-05,
+      "loss": 1.5998,
+      "step": 184
+    },
+    {
+      "epoch": 0.9158415841584159,
+      "grad_norm": 0.3000313341617584,
+      "learning_rate": 1.86268979442843e-05,
+      "loss": 1.5109,
+      "step": 185
+    },
+    {
+      "epoch": 0.9207920792079208,
+      "grad_norm": 0.3000313341617584,
+      "learning_rate": 1.86268979442843e-05,
+      "loss": 1.5017,
+      "step": 186
+    },
+    {
+      "epoch": 0.9257425742574258,
+      "grad_norm": 0.3000313341617584,
+      "learning_rate": 1.86268979442843e-05,
+      "loss": 1.5368,
+      "step": 187
+    },
+    {
+      "epoch": 0.9306930693069307,
+      "grad_norm": 0.2862202227115631,
+      "learning_rate": 1.8731528764550483e-05,
+      "loss": 1.4827,
+      "step": 188
+    },
+    {
+      "epoch": 0.9356435643564357,
+      "grad_norm": 0.2862202227115631,
+      "learning_rate": 1.8731528764550483e-05,
+      "loss": 1.549,
+      "step": 189
+    },
+    {
+      "epoch": 0.9405940594059405,
+      "grad_norm": 0.2862202227115631,
+      "learning_rate": 1.8731528764550483e-05,
+      "loss": 1.5179,
+      "step": 190
+    },
+    {
+      "epoch": 0.9455445544554455,
+      "grad_norm": 0.2862202227115631,
+      "learning_rate": 1.8731528764550483e-05,
+      "loss": 1.5286,
+      "step": 191
+    },
+    {
+      "epoch": 0.9504950495049505,
+      "grad_norm": 0.3076187074184418,
+      "learning_rate": 1.883395666622519e-05,
+      "loss": 1.6123,
+      "step": 192
+    },
+    {
+      "epoch": 0.9554455445544554,
+      "grad_norm": 0.3076187074184418,
+      "learning_rate": 1.883395666622519e-05,
+      "loss": 1.4996,
+      "step": 193
+    },
+    {
+      "epoch": 0.9603960396039604,
+      "grad_norm": 0.3076187074184418,
+      "learning_rate": 1.883395666622519e-05,
+      "loss": 1.5736,
+      "step": 194
+    },
+    {
+      "epoch": 0.9653465346534653,
+      "grad_norm": 0.3076187074184418,
+      "learning_rate": 1.883395666622519e-05,
+      "loss": 1.5538,
+      "step": 195
+    },
+    {
+      "epoch": 0.9702970297029703,
+      "grad_norm": 0.2966230511665344,
+      "learning_rate": 1.893427250117422e-05,
+      "loss": 1.5405,
+      "step": 196
+    },
+    {
+      "epoch": 0.9752475247524752,
+      "grad_norm": 0.2966230511665344,
+      "learning_rate": 1.893427250117422e-05,
+      "loss": 1.5288,
+      "step": 197
+    },
+    {
+      "epoch": 0.9801980198019802,
+      "grad_norm": 0.2966230511665344,
+      "learning_rate": 1.893427250117422e-05,
+      "loss": 1.5328,
+      "step": 198
+    },
+    {
+      "epoch": 0.9851485148514851,
+      "grad_norm": 0.2966230511665344,
+      "learning_rate": 1.893427250117422e-05,
+      "loss": 1.5754,
+      "step": 199
+    },
+    {
+      "epoch": 0.9900990099009901,
+      "grad_norm": 0.3104676604270935,
+      "learning_rate": 1.903256161431675e-05,
+      "loss": 1.5506,
+      "step": 200
+    },
+    {
+      "epoch": 0.995049504950495,
+      "grad_norm": 0.3104676604270935,
+      "learning_rate": 1.903256161431675e-05,
+      "loss": 1.5486,
+      "step": 201
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3104676604270935,
+      "learning_rate": 1.903256161431675e-05,
+      "loss": 1.4932,
+      "step": 202
+    },
+    {
+      "epoch": 1.004950495049505,
+      "grad_norm": 0.3104676604270935,
+      "learning_rate": 1.903256161431675e-05,
+      "loss": 1.4559,
+      "step": 203
+    },
+    {
+      "epoch": 1.00990099009901,
+      "grad_norm": 0.305254191160202,
+      "learning_rate": 1.912890427989334e-05,
+      "loss": 1.4787,
+      "step": 204
+    },
+    {
+      "epoch": 1.0148514851485149,
+      "grad_norm": 0.305254191160202,
+      "learning_rate": 1.912890427989334e-05,
+      "loss": 1.5086,
+      "step": 205
+    },
+    {
+      "epoch": 1.0198019801980198,
+      "grad_norm": 0.305254191160202,
+      "learning_rate": 1.912890427989334e-05,
+      "loss": 1.4966,
+      "step": 206
+    },
+    {
+      "epoch": 1.0247524752475248,
+      "grad_norm": 0.305254191160202,
+      "learning_rate": 1.912890427989334e-05,
+      "loss": 1.4825,
+      "step": 207
+    },
+    {
+      "epoch": 1.0297029702970297,
+      "grad_norm": 0.2713533341884613,
+      "learning_rate": 1.922337609536952e-05,
+      "loss": 1.5957,
+      "step": 208
+    },
+    {
+      "epoch": 1.0346534653465347,
+      "grad_norm": 0.2713533341884613,
+      "learning_rate": 1.922337609536952e-05,
+      "loss": 1.463,
+      "step": 209
+    },
+    {
+      "epoch": 1.0396039603960396,
+      "grad_norm": 0.2713533341884613,
+      "learning_rate": 1.922337609536952e-05,
+      "loss": 1.5056,
+      "step": 210
+    },
+    {
+      "epoch": 1.0445544554455446,
+      "grad_norm": 0.2713533341884613,
+      "learning_rate": 1.922337609536952e-05,
+      "loss": 1.5194,
+      "step": 211
+    },
+    {
+      "epoch": 1.0495049504950495,
+      "grad_norm": 0.30546605587005615,
+      "learning_rate": 1.9316048337817536e-05,
+      "loss": 1.518,
+      "step": 212
+    },
+    {
+      "epoch": 1.0544554455445545,
+      "grad_norm": 0.30546605587005615,
+      "learning_rate": 1.9316048337817536e-05,
+      "loss": 1.4553,
+      "step": 213
+    },
+    {
+      "epoch": 1.0594059405940595,
+      "grad_norm": 0.30546605587005615,
+      "learning_rate": 1.9316048337817536e-05,
+      "loss": 1.5377,
+      "step": 214
+    },
+    {
+      "epoch": 1.0643564356435644,
+      "grad_norm": 0.30546605587005615,
+      "learning_rate": 1.9316048337817536e-05,
+      "loss": 1.5212,
+      "step": 215
+    },
+    {
+      "epoch": 1.0693069306930694,
+      "grad_norm": 0.299664705991745,
+      "learning_rate": 1.9406988286985313e-05,
+      "loss": 1.4681,
+      "step": 216
+    },
+    {
+      "epoch": 1.0742574257425743,
+      "grad_norm": 0.299664705991745,
+      "learning_rate": 1.9406988286985313e-05,
+      "loss": 1.4718,
+      "step": 217
+    },
+    {
+      "epoch": 1.0792079207920793,
+      "grad_norm": 0.299664705991745,
+      "learning_rate": 1.9406988286985313e-05,
+      "loss": 1.4783,
+      "step": 218
+    },
+    {
+      "epoch": 1.0841584158415842,
+      "grad_norm": 0.299664705991745,
+      "learning_rate": 1.9406988286985313e-05,
+      "loss": 1.4054,
+      "step": 219
+    },
+    {
+      "epoch": 1.0891089108910892,
+      "grad_norm": 0.2781918942928314,
+      "learning_rate": 1.9496259518720788e-05,
+      "loss": 1.4988,
+      "step": 220
+    },
+    {
+      "epoch": 1.0940594059405941,
+      "grad_norm": 0.2781918942928314,
+      "learning_rate": 1.9496259518720788e-05,
+      "loss": 1.4985,
+      "step": 221
+    },
+    {
+      "epoch": 1.099009900990099,
+      "grad_norm": 0.2781918942928314,
+      "learning_rate": 1.9496259518720788e-05,
+      "loss": 1.4837,
+      "step": 222
+    },
+    {
+      "epoch": 1.103960396039604,
+      "grad_norm": 0.2781918942928314,
+      "learning_rate": 1.9496259518720788e-05,
+      "loss": 1.526,
+      "step": 223
+    },
+    {
+      "epoch": 1.108910891089109,
+      "grad_norm": 0.284312903881073,
+      "learning_rate": 1.958392217195718e-05,
+      "loss": 1.4355,
+      "step": 224
+    },
+    {
+      "epoch": 1.113861386138614,
+      "grad_norm": 0.284312903881073,
+      "learning_rate": 1.958392217195718e-05,
+      "loss": 1.4934,
+      "step": 225
+    },
+    {
+      "epoch": 1.118811881188119,
+      "grad_norm": 0.284312903881073,
+      "learning_rate": 1.958392217195718e-05,
+      "loss": 1.5242,
+      "step": 226
+    },
+    {
+      "epoch": 1.1237623762376239,
+      "grad_norm": 0.284312903881073,
+      "learning_rate": 1.958392217195718e-05,
+      "loss": 1.5275,
+      "step": 227
+    },
+    {
+      "epoch": 1.1287128712871288,
+      "grad_norm": 0.2917441725730896,
+      "learning_rate": 1.9670033192067303e-05,
+      "loss": 1.4717,
+      "step": 228
+    },
+    {
+      "epoch": 1.1336633663366338,
+      "grad_norm": 0.2917441725730896,
+      "learning_rate": 1.9670033192067303e-05,
+      "loss": 1.4589,
+      "step": 229
+    },
+    {
+      "epoch": 1.1386138613861387,
+      "grad_norm": 0.2917441725730896,
+      "learning_rate": 1.9670033192067303e-05,
+      "loss": 1.4698,
+      "step": 230
+    },
+    {
+      "epoch": 1.1435643564356435,
+      "grad_norm": 0.2917441725730896,
+      "learning_rate": 1.9670033192067303e-05,
+      "loss": 1.5495,
+      "step": 231
+    },
+    {
+      "epoch": 1.1485148514851484,
+      "grad_norm": 0.27202197909355164,
+      "learning_rate": 1.9754646553053346e-05,
+      "loss": 1.5051,
+      "step": 232
+    },
+    {
+      "epoch": 1.1534653465346534,
+      "grad_norm": 0.27202197909355164,
+      "learning_rate": 1.9754646553053346e-05,
+      "loss": 1.4921,
+      "step": 233
+    },
+    {
+      "epoch": 1.1584158415841583,
+      "grad_norm": 0.27202197909355164,
+      "learning_rate": 1.9754646553053346e-05,
+      "loss": 1.5619,
+      "step": 234
+    },
+    {
+      "epoch": 1.1633663366336633,
+      "grad_norm": 0.27202197909355164,
+      "learning_rate": 1.9754646553053346e-05,
+      "loss": 1.5186,
+      "step": 235
+    },
+    {
+      "epoch": 1.1683168316831682,
+      "grad_norm": 0.28283485770225525,
+      "learning_rate": 1.983781346074312e-05,
+      "loss": 1.5185,
+      "step": 236
+    },
+    {
+      "epoch": 1.1732673267326732,
+      "grad_norm": 0.28283485770225525,
+      "learning_rate": 1.983781346074312e-05,
+      "loss": 1.4358,
+      "step": 237
+    },
+    {
+      "epoch": 1.1782178217821782,
+      "grad_norm": 0.28283485770225525,
+      "learning_rate": 1.983781346074312e-05,
+      "loss": 1.5415,
+      "step": 238
+    },
+    {
+      "epoch": 1.183168316831683,
+      "grad_norm": 0.28283485770225525,
+      "learning_rate": 1.983781346074312e-05,
+      "loss": 1.5427,
+      "step": 239
+    },
+    {
+      "epoch": 1.188118811881188,
+      "grad_norm": 0.28792083263397217,
+      "learning_rate": 1.9919582538908503e-05,
+      "loss": 1.456,
+      "step": 240
+    },
+    {
+      "epoch": 1.193069306930693,
+      "grad_norm": 0.28792083263397217,
+      "learning_rate": 1.9919582538908503e-05,
+      "loss": 1.4901,
+      "step": 241
+    },
+    {
+      "epoch": 1.198019801980198,
+      "grad_norm": 0.28792083263397217,
+      "learning_rate": 1.9919582538908503e-05,
+      "loss": 1.438,
+      "step": 242
+    },
+    {
+      "epoch": 1.202970297029703,
+      "grad_norm": 0.28792083263397217,
+      "learning_rate": 1.9919582538908503e-05,
+      "loss": 1.5229,
+      "step": 243
+    },
+    {
+      "epoch": 1.2079207920792079,
+      "grad_norm": 0.2743774354457855,
+      "learning_rate": 2e-05,
+      "loss": 1.5547,
+      "step": 244
+    },
+    {
+      "epoch": 1.2128712871287128,
+      "grad_norm": 0.2743774354457855,
+      "learning_rate": 2e-05,
+      "loss": 1.5215,
+      "step": 245
+    },
+    {
+      "epoch": 1.2178217821782178,
+      "grad_norm": 0.2743774354457855,
+      "learning_rate": 2e-05,
+      "loss": 1.452,
+      "step": 246
+    },
+    {
+      "epoch": 1.2227722772277227,
+      "grad_norm": 0.2743774354457855,
+      "learning_rate": 2e-05,
+      "loss": 1.4676,
+      "step": 247
+    },
+    {
+      "epoch": 1.2277227722772277,
+      "grad_norm": 0.2722848951816559,
+      "learning_rate": 2e-05,
+      "loss": 1.4569,
+      "step": 248
+    },
+    {
+      "epoch": 1.2326732673267327,
+      "grad_norm": 0.2722848951816559,
+      "learning_rate": 2e-05,
+      "loss": 1.5278,
+      "step": 249
+    },
+    {
+      "epoch": 1.2376237623762376,
+      "grad_norm": 0.2722848951816559,
+      "learning_rate": 2e-05,
+      "loss": 1.4752,
+      "step": 250
+    },
+    {
+      "epoch": 1.2425742574257426,
+      "grad_norm": 0.2722848951816559,
+      "learning_rate": 2e-05,
+      "loss": 1.5639,
+      "step": 251
+    },
+    {
+      "epoch": 1.2475247524752475,
+      "grad_norm": 0.2844066917896271,
+      "learning_rate": 2e-05,
+      "loss": 1.5385,
+      "step": 252
+    },
+    {
+      "epoch": 1.2524752475247525,
+      "grad_norm": 0.2844066917896271,
+      "learning_rate": 2e-05,
+      "loss": 1.5141,
+      "step": 253
+    },
+    {
+      "epoch": 1.2574257425742574,
+      "grad_norm": 0.2844066917896271,
+      "learning_rate": 2e-05,
+      "loss": 1.4676,
+      "step": 254
+    },
+    {
+      "epoch": 1.2623762376237624,
+      "grad_norm": 0.2844066917896271,
+      "learning_rate": 2e-05,
+      "loss": 1.4632,
+      "step": 255
+    },
+    {
+      "epoch": 1.2673267326732673,
+      "grad_norm": 0.2843905985355377,
+      "learning_rate": 2e-05,
+      "loss": 1.4868,
+      "step": 256
+    },
+    {
+      "epoch": 1.2722772277227723,
+      "grad_norm": 0.2843905985355377,
+      "learning_rate": 2e-05,
+      "loss": 1.5614,
+      "step": 257
+    },
+    {
+      "epoch": 1.2772277227722773,
+      "grad_norm": 0.2843905985355377,
+      "learning_rate": 2e-05,
+      "loss": 1.5404,
+      "step": 258
+    },
+    {
+      "epoch": 1.2821782178217822,
+      "grad_norm": 0.2843905985355377,
+      "learning_rate": 2e-05,
+      "loss": 1.4836,
+      "step": 259
+    },
+    {
+      "epoch": 1.2871287128712872,
+      "grad_norm": 0.2771119773387909,
+      "learning_rate": 2e-05,
+      "loss": 1.5339,
+      "step": 260
+    },
+    {
+      "epoch": 1.2920792079207921,
+      "grad_norm": 0.2771119773387909,
+      "learning_rate": 2e-05,
+      "loss": 1.4168,
+      "step": 261
+    },
+    {
+      "epoch": 1.297029702970297,
+      "grad_norm": 0.2771119773387909,
+      "learning_rate": 2e-05,
+      "loss": 1.4641,
+      "step": 262
+    },
+    {
+      "epoch": 1.301980198019802,
+      "grad_norm": 0.2771119773387909,
+      "learning_rate": 2e-05,
+      "loss": 1.4616,
+      "step": 263
+    },
+    {
+      "epoch": 1.306930693069307,
+      "grad_norm": 0.27680638432502747,
+      "learning_rate": 2e-05,
+      "loss": 1.4962,
+      "step": 264
+    },
+    {
+      "epoch": 1.311881188118812,
+      "grad_norm": 0.27680638432502747,
+      "learning_rate": 2e-05,
+      "loss": 1.455,
+      "step": 265
+    },
+    {
+      "epoch": 1.316831683168317,
+      "grad_norm": 0.27680638432502747,
+      "learning_rate": 2e-05,
+      "loss": 1.4401,
+      "step": 266
+    },
+    {
+      "epoch": 1.3217821782178218,
+      "grad_norm": 0.27680638432502747,
+      "learning_rate": 2e-05,
+      "loss": 1.571,
+      "step": 267
+    },
+    {
+      "epoch": 1.3267326732673268,
+      "grad_norm": 0.2692127823829651,
+      "learning_rate": 2e-05,
+      "loss": 1.5209,
+      "step": 268
+    },
+    {
+      "epoch": 1.3316831683168318,
+      "grad_norm": 0.2692127823829651,
+      "learning_rate": 2e-05,
+      "loss": 1.5326,
+      "step": 269
+    },
+    {
+      "epoch": 1.3366336633663367,
+      "grad_norm": 0.2692127823829651,
+      "learning_rate": 2e-05,
+      "loss": 1.496,
+      "step": 270
+    },
+    {
+      "epoch": 1.3415841584158417,
+      "grad_norm": 0.2692127823829651,
+      "learning_rate": 2e-05,
+      "loss": 1.4306,
+      "step": 271
+    },
+    {
+      "epoch": 1.3465346534653464,
+      "grad_norm": 0.26907190680503845,
+      "learning_rate": 2e-05,
+      "loss": 1.4874,
+      "step": 272
+    },
+    {
+      "epoch": 1.3514851485148514,
+      "grad_norm": 0.26907190680503845,
+      "learning_rate": 2e-05,
+      "loss": 1.4535,
+      "step": 273
+    },
+    {
+      "epoch": 1.3564356435643563,
+      "grad_norm": 0.26907190680503845,
+      "learning_rate": 2e-05,
+      "loss": 1.4721,
+      "step": 274
+    },
+    {
+      "epoch": 1.3613861386138613,
+      "grad_norm": 0.26907190680503845,
+      "learning_rate": 2e-05,
+      "loss": 1.3758,
+      "step": 275
+    },
+    {
+      "epoch": 1.3663366336633662,
+      "grad_norm": 0.26644808053970337,
+      "learning_rate": 2e-05,
+      "loss": 1.4439,
+      "step": 276
+    },
+    {
+      "epoch": 1.3712871287128712,
+      "grad_norm": 0.26644808053970337,
+      "learning_rate": 2e-05,
+      "loss": 1.4347,
+      "step": 277
+    },
+    {
+      "epoch": 1.3762376237623761,
+      "grad_norm": 0.26644808053970337,
+      "learning_rate": 2e-05,
+      "loss": 1.4604,
+      "step": 278
+    },
+    {
+      "epoch": 1.381188118811881,
+      "grad_norm": 0.26644808053970337,
+      "learning_rate": 2e-05,
+      "loss": 1.4547,
+      "step": 279
+    },
+    {
+      "epoch": 1.386138613861386,
+      "grad_norm": 0.2679310441017151,
+      "learning_rate": 2e-05,
+      "loss": 1.518,
+      "step": 280
+    },
+    {
+      "epoch": 1.391089108910891,
+      "grad_norm": 0.2679310441017151,
+      "learning_rate": 2e-05,
+      "loss": 1.4687,
+      "step": 281
+    },
+    {
+      "epoch": 1.396039603960396,
+      "grad_norm": 0.2679310441017151,
+      "learning_rate": 2e-05,
+      "loss": 1.448,
+      "step": 282
+    },
+    {
+      "epoch": 1.400990099009901,
+      "grad_norm": 0.2679310441017151,
+      "learning_rate": 2e-05,
+      "loss": 1.5097,
+      "step": 283
+    },
+    {
+      "epoch": 1.4059405940594059,
+      "grad_norm": 0.2934781014919281,
+      "learning_rate": 2e-05,
+      "loss": 1.4957,
+      "step": 284
+    },
+    {
+      "epoch": 1.4108910891089108,
+      "grad_norm": 0.2934781014919281,
+      "learning_rate": 2e-05,
+      "loss": 1.4581,
+      "step": 285
+    },
+    {
+      "epoch": 1.4158415841584158,
+      "grad_norm": 0.2934781014919281,
+      "learning_rate": 2e-05,
+      "loss": 1.4936,
+      "step": 286
+    },
+    {
+      "epoch": 1.4207920792079207,
+      "grad_norm": 0.2934781014919281,
+      "learning_rate": 2e-05,
+      "loss": 1.4964,
+      "step": 287
+    },
+    {
+      "epoch": 1.4257425742574257,
+      "grad_norm": 0.2842659056186676,
+      "learning_rate": 2e-05,
+      "loss": 1.3665,
+      "step": 288
+    },
+    {
+      "epoch": 1.4306930693069306,
+      "grad_norm": 0.2842659056186676,
+      "learning_rate": 2e-05,
+      "loss": 1.4713,
+      "step": 289
+    },
+    {
+      "epoch": 1.4356435643564356,
+      "grad_norm": 0.2842659056186676,
+      "learning_rate": 2e-05,
+      "loss": 1.4624,
+      "step": 290
+    },
+    {
+      "epoch": 1.4405940594059405,
+      "grad_norm": 0.2842659056186676,
+      "learning_rate": 2e-05,
+      "loss": 1.4777,
+      "step": 291
+    },
+    {
+      "epoch": 1.4455445544554455,
+      "grad_norm": 0.27639469504356384,
+      "learning_rate": 2e-05,
+      "loss": 1.4559,
+      "step": 292
+    },
+    {
+      "epoch": 1.4504950495049505,
+      "grad_norm": 0.27639469504356384,
+      "learning_rate": 2e-05,
+      "loss": 1.4857,
+      "step": 293
+    },
+    {
+      "epoch": 1.4554455445544554,
+      "grad_norm": 0.27639469504356384,
+      "learning_rate": 2e-05,
+      "loss": 1.4328,
+      "step": 294
+    },
+    {
+      "epoch": 1.4603960396039604,
+      "grad_norm": 0.27639469504356384,
+      "learning_rate": 2e-05,
+      "loss": 1.4704,
+      "step": 295
+    },
+    {
+      "epoch": 1.4653465346534653,
+      "grad_norm": 0.2794305086135864,
+      "learning_rate": 2e-05,
+      "loss": 1.4608,
+      "step": 296
+    },
+    {
+      "epoch": 1.4702970297029703,
+      "grad_norm": 0.2794305086135864,
+      "learning_rate": 2e-05,
+      "loss": 1.458,
+      "step": 297
+    },
+    {
+      "epoch": 1.4752475247524752,
+      "grad_norm": 0.2794305086135864,
+      "learning_rate": 2e-05,
+      "loss": 1.4629,
+      "step": 298
+    },
+    {
+      "epoch": 1.4801980198019802,
+      "grad_norm": 0.2794305086135864,
+      "learning_rate": 2e-05,
+      "loss": 1.4899,
+      "step": 299
+    },
+    {
+      "epoch": 1.4851485148514851,
+      "grad_norm": 0.29005393385887146,
+      "learning_rate": 2e-05,
+      "loss": 1.4461,
+      "step": 300
+    },
+    {
+      "epoch": 1.49009900990099,
+      "grad_norm": 0.29005393385887146,
+      "learning_rate": 2e-05,
+      "loss": 1.395,
+      "step": 301
+    },
+    {
+      "epoch": 1.495049504950495,
+      "grad_norm": 0.29005393385887146,
+      "learning_rate": 2e-05,
+      "loss": 1.5044,
+      "step": 302
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.29005393385887146,
+      "learning_rate": 2e-05,
+      "loss": 1.4857,
+      "step": 303
+    },
+    {
+      "epoch": 1.504950495049505,
+      "grad_norm": 0.28084444999694824,
+      "learning_rate": 2e-05,
+      "loss": 1.4932,
+      "step": 304
+    },
+    {
+      "epoch": 1.50990099009901,
+      "grad_norm": 0.28084444999694824,
+      "learning_rate": 2e-05,
+      "loss": 1.4494,
+      "step": 305
+    },
+    {
+      "epoch": 1.5148514851485149,
+      "grad_norm": 0.28084444999694824,
+      "learning_rate": 2e-05,
+      "loss": 1.4542,
+      "step": 306
+    },
+    {
+      "epoch": 1.5198019801980198,
+      "grad_norm": 0.28084444999694824,
+      "learning_rate": 2e-05,
+      "loss": 1.4893,
+      "step": 307
+    },
+    {
+      "epoch": 1.5247524752475248,
+      "grad_norm": 0.283642441034317,
+      "learning_rate": 2e-05,
+      "loss": 1.5155,
+      "step": 308
+    },
+    {
+      "epoch": 1.5297029702970297,
+      "grad_norm": 0.283642441034317,
+      "learning_rate": 2e-05,
+      "loss": 1.3895,
+      "step": 309
+    },
+    {
+      "epoch": 1.5346534653465347,
+      "grad_norm": 0.283642441034317,
+      "learning_rate": 2e-05,
+      "loss": 1.488,
+      "step": 310
+    },
+    {
+      "epoch": 1.5396039603960396,
+      "grad_norm": 0.283642441034317,
+      "learning_rate": 2e-05,
+      "loss": 1.5099,
+      "step": 311
+    },
+    {
+      "epoch": 1.5445544554455446,
+      "grad_norm": 0.2822302579879761,
+      "learning_rate": 2e-05,
+      "loss": 1.4694,
+      "step": 312
+    },
+    {
+      "epoch": 1.5495049504950495,
+      "grad_norm": 0.2822302579879761,
+      "learning_rate": 2e-05,
+      "loss": 1.4307,
+      "step": 313
+    },
+    {
+      "epoch": 1.5544554455445545,
+      "grad_norm": 0.2822302579879761,
+      "learning_rate": 2e-05,
+      "loss": 1.4705,
+      "step": 314
+    },
+    {
+      "epoch": 1.5594059405940595,
+      "grad_norm": 0.2822302579879761,
+      "learning_rate": 2e-05,
+      "loss": 1.407,
+      "step": 315
+    },
+    {
+      "epoch": 1.5643564356435644,
+      "grad_norm": 0.2674373984336853,
+      "learning_rate": 2e-05,
+      "loss": 1.5169,
+      "step": 316
+    },
+    {
+      "epoch": 1.5693069306930694,
+      "grad_norm": 0.2674373984336853,
+      "learning_rate": 2e-05,
+      "loss": 1.4667,
+      "step": 317
+    },
+    {
+      "epoch": 1.5742574257425743,
+      "grad_norm": 0.2674373984336853,
+      "learning_rate": 2e-05,
+      "loss": 1.4669,
+      "step": 318
+    },
+    {
+      "epoch": 1.5792079207920793,
+      "grad_norm": 0.2674373984336853,
+      "learning_rate": 2e-05,
+      "loss": 1.4217,
+      "step": 319
+    },
+    {
+      "epoch": 1.5841584158415842,
+      "grad_norm": 0.27052974700927734,
+      "learning_rate": 2e-05,
+      "loss": 1.4244,
+      "step": 320
+    },
+    {
+      "epoch": 1.5891089108910892,
+      "grad_norm": 0.27052974700927734,
+      "learning_rate": 2e-05,
+      "loss": 1.461,
+      "step": 321
+    },
+    {
+      "epoch": 1.5940594059405941,
+      "grad_norm": 0.27052974700927734,
+      "learning_rate": 2e-05,
+      "loss": 1.497,
+      "step": 322
+    },
+    {
+      "epoch": 1.599009900990099,
+      "grad_norm": 0.27052974700927734,
+      "learning_rate": 2e-05,
+      "loss": 1.503,
+      "step": 323
+    },
+    {
+      "epoch": 1.603960396039604,
+      "grad_norm": 0.2938457429409027,
+      "learning_rate": 2e-05,
+      "loss": 1.4156,
+      "step": 324
+    },
+    {
+      "epoch": 1.608910891089109,
+      "grad_norm": 0.2938457429409027,
+      "learning_rate": 2e-05,
+      "loss": 1.4303,
+      "step": 325
+    },
+    {
+      "epoch": 1.613861386138614,
+      "grad_norm": 0.2938457429409027,
+      "learning_rate": 2e-05,
+      "loss": 1.471,
+      "step": 326
+    },
+    {
+      "epoch": 1.618811881188119,
+      "grad_norm": 0.2938457429409027,
+      "learning_rate": 2e-05,
+      "loss": 1.4311,
+      "step": 327
+    },
+    {
+      "epoch": 1.6237623762376239,
+      "grad_norm": 0.2872541844844818,
+      "learning_rate": 2e-05,
+      "loss": 1.4992,
+      "step": 328
+    },
+    {
+      "epoch": 1.6287128712871288,
+      "grad_norm": 0.2872541844844818,
+      "learning_rate": 2e-05,
+      "loss": 1.4592,
+      "step": 329
+    },
+    {
+      "epoch": 1.6336633663366338,
+      "grad_norm": 0.2872541844844818,
+      "learning_rate": 2e-05,
+      "loss": 1.4477,
+      "step": 330
+    },
+    {
+      "epoch": 1.6386138613861387,
+      "grad_norm": 0.2872541844844818,
+      "learning_rate": 2e-05,
+      "loss": 1.3685,
+      "step": 331
+    },
+    {
+      "epoch": 1.6435643564356437,
+      "grad_norm": 0.2954404056072235,
+      "learning_rate": 2e-05,
+      "loss": 1.4691,
+      "step": 332
+    },
+    {
+      "epoch": 1.6485148514851486,
+      "grad_norm": 0.2954404056072235,
+      "learning_rate": 2e-05,
+      "loss": 1.4652,
+      "step": 333
+    },
+    {
+      "epoch": 1.6534653465346536,
+      "grad_norm": 0.2954404056072235,
+      "learning_rate": 2e-05,
+      "loss": 1.4048,
+      "step": 334
+    },
+    {
+      "epoch": 1.6584158415841586,
+      "grad_norm": 0.2954404056072235,
+      "learning_rate": 2e-05,
+      "loss": 1.4597,
+      "step": 335
+    },
+    {
+      "epoch": 1.6633663366336635,
+      "grad_norm": 0.2781623899936676,
+      "learning_rate": 2e-05,
+      "loss": 1.4382,
+      "step": 336
+    },
+    {
+      "epoch": 1.6683168316831685,
+      "grad_norm": 0.2781623899936676,
+      "learning_rate": 2e-05,
+      "loss": 1.5407,
+      "step": 337
+    },
+    {
+      "epoch": 1.6732673267326734,
+      "grad_norm": 0.2781623899936676,
+      "learning_rate": 2e-05,
+      "loss": 1.4404,
+      "step": 338
+    },
+    {
+      "epoch": 1.6782178217821784,
+      "grad_norm": 0.2781623899936676,
+      "learning_rate": 2e-05,
+      "loss": 1.4698,
+      "step": 339
+    },
+    {
+      "epoch": 1.6831683168316833,
+      "grad_norm": 0.3247792422771454,
+      "learning_rate": 2e-05,
+      "loss": 1.4563,
+      "step": 340
+    },
+    {
+      "epoch": 1.688118811881188,
+      "grad_norm": 0.3247792422771454,
+      "learning_rate": 2e-05,
+      "loss": 1.3681,
+      "step": 341
+    },
+    {
+      "epoch": 1.693069306930693,
+      "grad_norm": 0.3247792422771454,
+      "learning_rate": 2e-05,
+      "loss": 1.4239,
+      "step": 342
+    },
+    {
+      "epoch": 1.698019801980198,
+      "grad_norm": 0.3247792422771454,
+      "learning_rate": 2e-05,
+      "loss": 1.4099,
+      "step": 343
+    },
+    {
+      "epoch": 1.702970297029703,
+      "grad_norm": 0.2891463339328766,
+      "learning_rate": 2e-05,
+      "loss": 1.4827,
+      "step": 344
+    },
+    {
+      "epoch": 1.7079207920792079,
+      "grad_norm": 0.2891463339328766,
+      "learning_rate": 2e-05,
+      "loss": 1.4642,
+      "step": 345
+    },
+    {
+      "epoch": 1.7128712871287128,
+      "grad_norm": 0.2891463339328766,
+      "learning_rate": 2e-05,
+      "loss": 1.5105,
+      "step": 346
+    },
+    {
+      "epoch": 1.7178217821782178,
+      "grad_norm": 0.2891463339328766,
+      "learning_rate": 2e-05,
+      "loss": 1.4188,
+      "step": 347
+    },
+    {
+      "epoch": 1.7227722772277227,
+      "grad_norm": 0.2906363308429718,
+      "learning_rate": 2e-05,
+      "loss": 1.4342,
+      "step": 348
+    },
+    {
+      "epoch": 1.7277227722772277,
+      "grad_norm": 0.2906363308429718,
+      "learning_rate": 2e-05,
+      "loss": 1.4774,
+      "step": 349
+    },
+    {
+      "epoch": 1.7326732673267327,
+      "grad_norm": 0.2906363308429718,
+      "learning_rate": 2e-05,
+      "loss": 1.5289,
+      "step": 350
+    },
+    {
+      "epoch": 1.7376237623762376,
+      "grad_norm": 0.2906363308429718,
+      "learning_rate": 2e-05,
+      "loss": 1.4149,
+      "step": 351
+    },
+    {
+      "epoch": 1.7425742574257426,
+      "grad_norm": 0.2809044420719147,
+      "learning_rate": 2e-05,
+      "loss": 1.4506,
+      "step": 352
+    },
+    {
+      "epoch": 1.7475247524752475,
+      "grad_norm": 0.2809044420719147,
+      "learning_rate": 2e-05,
+      "loss": 1.4838,
+      "step": 353
+    },
+    {
+      "epoch": 1.7524752475247525,
+      "grad_norm": 0.2809044420719147,
+      "learning_rate": 2e-05,
+      "loss": 1.4642,
+      "step": 354
+    },
+    {
+      "epoch": 1.7574257425742574,
+      "grad_norm": 0.2809044420719147,
+      "learning_rate": 2e-05,
+      "loss": 1.428,
+      "step": 355
+    },
+    {
+      "epoch": 1.7623762376237624,
+      "grad_norm": 0.2897462248802185,
+      "learning_rate": 2e-05,
+      "loss": 1.459,
+      "step": 356
+    },
+    {
+      "epoch": 1.7673267326732673,
+      "grad_norm": 0.2897462248802185,
+      "learning_rate": 2e-05,
+      "loss": 1.4475,
+      "step": 357
+    },
+    {
+      "epoch": 1.7722772277227723,
+      "grad_norm": 0.2897462248802185,
+      "learning_rate": 2e-05,
+      "loss": 1.4933,
+      "step": 358
+    },
+    {
+      "epoch": 1.7772277227722773,
+      "grad_norm": 0.2897462248802185,
+      "learning_rate": 2e-05,
+      "loss": 1.4318,
+      "step": 359
+    },
+    {
+      "epoch": 1.7821782178217822,
+      "grad_norm": 0.2832702696323395,
+      "learning_rate": 2e-05,
+      "loss": 1.3858,
+      "step": 360
+    },
+    {
+      "epoch": 1.7871287128712872,
+      "grad_norm": 0.2832702696323395,
+      "learning_rate": 2e-05,
+      "loss": 1.4198,
+      "step": 361
+    },
+    {
+      "epoch": 1.7920792079207921,
+      "grad_norm": 0.2832702696323395,
+      "learning_rate": 2e-05,
+      "loss": 1.4236,
+      "step": 362
+    },
+    {
+      "epoch": 1.797029702970297,
+      "grad_norm": 0.2832702696323395,
+      "learning_rate": 2e-05,
+      "loss": 1.4393,
+      "step": 363
+    },
+    {
+      "epoch": 1.801980198019802,
+      "grad_norm": 0.2876037657260895,
+      "learning_rate": 2e-05,
+      "loss": 1.4811,
+      "step": 364
+    },
+    {
+      "epoch": 1.806930693069307,
+      "grad_norm": 0.2876037657260895,
+      "learning_rate": 2e-05,
+      "loss": 1.4662,
+      "step": 365
+    },
+    {
+      "epoch": 1.811881188118812,
+      "grad_norm": 0.2876037657260895,
+      "learning_rate": 2e-05,
+      "loss": 1.4535,
+      "step": 366
+    },
+    {
+      "epoch": 1.8168316831683167,
+      "grad_norm": 0.2876037657260895,
+      "learning_rate": 2e-05,
+      "loss": 1.4433,
+      "step": 367
+    },
+    {
+      "epoch": 1.8217821782178216,
+      "grad_norm": 0.2914987802505493,
+      "learning_rate": 2e-05,
+      "loss": 1.4859,
+      "step": 368
+    },
+    {
+      "epoch": 1.8267326732673266,
+      "grad_norm": 0.2914987802505493,
+      "learning_rate": 2e-05,
+      "loss": 1.4385,
+      "step": 369
+    },
+    {
+      "epoch": 1.8316831683168315,
+      "grad_norm": 0.2914987802505493,
+      "learning_rate": 2e-05,
+      "loss": 1.42,
+      "step": 370
+    },
+    {
+      "epoch": 1.8366336633663365,
+      "grad_norm": 0.2914987802505493,
+      "learning_rate": 2e-05,
+      "loss": 1.4572,
+      "step": 371
+    },
+    {
+      "epoch": 1.8415841584158414,
+      "grad_norm": 0.2872907221317291,
+      "learning_rate": 2e-05,
+      "loss": 1.4007,
+      "step": 372
+    },
+    {
+      "epoch": 1.8465346534653464,
+      "grad_norm": 0.2872907221317291,
+      "learning_rate": 2e-05,
+      "loss": 1.4527,
+      "step": 373
+    },
+    {
+      "epoch": 1.8514851485148514,
+      "grad_norm": 0.2872907221317291,
+      "learning_rate": 2e-05,
+      "loss": 1.4357,
+      "step": 374
+    },
+    {
+      "epoch": 1.8564356435643563,
+      "grad_norm": 0.2872907221317291,
+      "learning_rate": 2e-05,
+      "loss": 1.4271,
+      "step": 375
+    },
+    {
+      "epoch": 1.8613861386138613,
+      "grad_norm": 0.28607606887817383,
+      "learning_rate": 2e-05,
+      "loss": 1.4886,
+      "step": 376
+    },
+    {
+      "epoch": 1.8663366336633662,
+      "grad_norm": 0.28607606887817383,
+      "learning_rate": 2e-05,
+      "loss": 1.4755,
+      "step": 377
+    },
+    {
+      "epoch": 1.8712871287128712,
+      "grad_norm": 0.28607606887817383,
+      "learning_rate": 2e-05,
+      "loss": 1.4275,
+      "step": 378
+    },
+    {
+      "epoch": 1.8762376237623761,
+      "grad_norm": 0.28607606887817383,
+      "learning_rate": 2e-05,
+      "loss": 1.4568,
+      "step": 379
+    },
+    {
+      "epoch": 1.881188118811881,
+      "grad_norm": 0.28154703974723816,
+      "learning_rate": 2e-05,
+      "loss": 1.4347,
+      "step": 380
+    },
+    {
+      "epoch": 1.886138613861386,
+      "grad_norm": 0.28154703974723816,
+      "learning_rate": 2e-05,
+      "loss": 1.4496,
+      "step": 381
+    },
+    {
+      "epoch": 1.891089108910891,
+      "grad_norm": 0.28154703974723816,
+      "learning_rate": 2e-05,
+      "loss": 1.4331,
+      "step": 382
+    },
+    {
+      "epoch": 1.896039603960396,
+      "grad_norm": 0.28154703974723816,
+      "learning_rate": 2e-05,
+      "loss": 1.4467,
+      "step": 383
+    },
+    {
+      "epoch": 1.900990099009901,
+      "grad_norm": 0.30775487422943115,
+      "learning_rate": 2e-05,
+      "loss": 1.5031,
+      "step": 384
+    },
+    {
+      "epoch": 1.9059405940594059,
+      "grad_norm": 0.30775487422943115,
+      "learning_rate": 2e-05,
+      "loss": 1.4401,
+      "step": 385
+    },
+    {
+      "epoch": 1.9108910891089108,
+      "grad_norm": 0.30775487422943115,
+      "learning_rate": 2e-05,
+      "loss": 1.4608,
+      "step": 386
+    },
+    {
+      "epoch": 1.9158415841584158,
+      "grad_norm": 0.30775487422943115,
+      "learning_rate": 2e-05,
+      "loss": 1.3846,
+      "step": 387
+    },
+    {
+      "epoch": 1.9207920792079207,
+      "grad_norm": 0.28769612312316895,
+      "learning_rate": 2e-05,
+      "loss": 1.3951,
+      "step": 388
+    },
+    {
+      "epoch": 1.9257425742574257,
+      "grad_norm": 0.28769612312316895,
+      "learning_rate": 2e-05,
+      "loss": 1.4128,
+      "step": 389
+    },
+    {
+      "epoch": 1.9306930693069306,
+      "grad_norm": 0.28769612312316895,
+      "learning_rate": 2e-05,
+      "loss": 1.4753,
+      "step": 390
+    },
+    {
+      "epoch": 1.9356435643564356,
+      "grad_norm": 0.28769612312316895,
+      "learning_rate": 2e-05,
+      "loss": 1.4264,
+      "step": 391
+    },
+    {
+      "epoch": 1.9405940594059405,
+      "grad_norm": 0.3035818636417389,
+      "learning_rate": 2e-05,
+      "loss": 1.4908,
+      "step": 392
+    },
+    {
+      "epoch": 1.9455445544554455,
+      "grad_norm": 0.3035818636417389,
+      "learning_rate": 2e-05,
+      "loss": 1.4205,
+      "step": 393
+    },
+    {
+      "epoch": 1.9504950495049505,
+      "grad_norm": 0.3035818636417389,
+      "learning_rate": 2e-05,
+      "loss": 1.4378,
+      "step": 394
+    },
+    {
+      "epoch": 1.9554455445544554,
+      "grad_norm": 0.3035818636417389,
+      "learning_rate": 2e-05,
+      "loss": 1.4151,
+      "step": 395
+    },
+    {
+      "epoch": 1.9603960396039604,
+      "grad_norm": 0.3105930685997009,
+      "learning_rate": 2e-05,
+      "loss": 1.4885,
+      "step": 396
+    },
+    {
+      "epoch": 1.9653465346534653,
+      "grad_norm": 0.3105930685997009,
+      "learning_rate": 2e-05,
+      "loss": 1.4208,
+      "step": 397
+    },
+    {
+      "epoch": 1.9702970297029703,
+      "grad_norm": 0.3105930685997009,
+      "learning_rate": 2e-05,
+      "loss": 1.4401,
+      "step": 398
+    },
+    {
+      "epoch": 1.9752475247524752,
+      "grad_norm": 0.3105930685997009,
+      "learning_rate": 2e-05,
+      "loss": 1.4563,
+      "step": 399
+    },
+    {
+      "epoch": 1.9801980198019802,
+      "grad_norm": 0.28015851974487305,
+      "learning_rate": 2e-05,
+      "loss": 1.4198,
+      "step": 400
+    },
+    {
+      "epoch": 1.9851485148514851,
+      "grad_norm": 0.28015851974487305,
+      "learning_rate": 2e-05,
+      "loss": 1.4868,
+      "step": 401
+    },
+    {
+      "epoch": 1.99009900990099,
+      "grad_norm": 0.28015851974487305,
+      "learning_rate": 2e-05,
+      "loss": 1.431,
+      "step": 402
+    },
+    {
+      "epoch": 1.995049504950495,
+      "grad_norm": 0.28015851974487305,
+      "learning_rate": 2e-05,
+      "loss": 1.4577,
+      "step": 403
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.28616049885749817,
+      "learning_rate": 2e-05,
+      "loss": 1.3554,
+      "step": 404
+    },
+    {
+      "epoch": 2.004950495049505,
+      "grad_norm": 0.28616049885749817,
+      "learning_rate": 2e-05,
+      "loss": 1.4254,
+      "step": 405
+    },
+    {
+      "epoch": 2.00990099009901,
+      "grad_norm": 0.28616049885749817,
+      "learning_rate": 2e-05,
+      "loss": 1.341,
+      "step": 406
+    },
+    {
+      "epoch": 2.014851485148515,
+      "grad_norm": 0.28616049885749817,
+      "learning_rate": 2e-05,
+      "loss": 1.423,
+      "step": 407
+    },
+    {
+      "epoch": 2.01980198019802,
+      "grad_norm": 0.317325234413147,
+      "learning_rate": 2e-05,
+      "loss": 1.3991,
+      "step": 408
+    },
+    {
+      "epoch": 2.0247524752475248,
+      "grad_norm": 0.317325234413147,
+      "learning_rate": 2e-05,
+      "loss": 1.3587,
+      "step": 409
+    },
+    {
+      "epoch": 2.0297029702970297,
+      "grad_norm": 0.317325234413147,
+      "learning_rate": 2e-05,
+      "loss": 1.4268,
+      "step": 410
+    },
+    {
+      "epoch": 2.0346534653465347,
+      "grad_norm": 0.317325234413147,
+      "learning_rate": 2e-05,
+      "loss": 1.4146,
+      "step": 411
+    },
+    {
+      "epoch": 2.0396039603960396,
+      "grad_norm": 0.3060872256755829,
+      "learning_rate": 2e-05,
+      "loss": 1.4125,
+      "step": 412
+    },
+    {
+      "epoch": 2.0445544554455446,
+      "grad_norm": 0.3060872256755829,
+      "learning_rate": 2e-05,
+      "loss": 1.4337,
+      "step": 413
+    },
+    {
+      "epoch": 2.0495049504950495,
+      "grad_norm": 0.3060872256755829,
+      "learning_rate": 2e-05,
+      "loss": 1.4261,
+      "step": 414
+    },
+    {
+      "epoch": 2.0544554455445545,
+      "grad_norm": 0.3060872256755829,
+      "learning_rate": 2e-05,
+      "loss": 1.36,
+      "step": 415
+    },
+    {
+      "epoch": 2.0594059405940595,
+      "grad_norm": 0.2912452518939972,
+      "learning_rate": 2e-05,
+      "loss": 1.4411,
+      "step": 416
+    },
+    {
+      "epoch": 2.0643564356435644,
+      "grad_norm": 0.2912452518939972,
+      "learning_rate": 2e-05,
+      "loss": 1.463,
+      "step": 417
+    },
+    {
+      "epoch": 2.0693069306930694,
+      "grad_norm": 0.2912452518939972,
+      "learning_rate": 2e-05,
+      "loss": 1.4353,
+      "step": 418
+    },
+    {
+      "epoch": 2.0742574257425743,
+      "grad_norm": 0.2912452518939972,
+      "learning_rate": 2e-05,
+      "loss": 1.4165,
+      "step": 419
+    },
+    {
+      "epoch": 2.0792079207920793,
+      "grad_norm": 0.29850682616233826,
+      "learning_rate": 2e-05,
+      "loss": 1.4432,
+      "step": 420
+    },
+    {
+      "epoch": 2.0841584158415842,
+      "grad_norm": 0.29850682616233826,
+      "learning_rate": 2e-05,
+      "loss": 1.4062,
+      "step": 421
+    },
+    {
+      "epoch": 2.089108910891089,
+      "grad_norm": 0.29850682616233826,
+      "learning_rate": 2e-05,
+      "loss": 1.4478,
+      "step": 422
+    },
+    {
+      "epoch": 2.094059405940594,
+      "grad_norm": 0.29850682616233826,
+      "learning_rate": 2e-05,
+      "loss": 1.3477,
+      "step": 423
+    },
+    {
+      "epoch": 2.099009900990099,
+      "grad_norm": 0.27797263860702515,
+      "learning_rate": 2e-05,
+      "loss": 1.3811,
+      "step": 424
+    },
+    {
+      "epoch": 2.103960396039604,
+      "grad_norm": 0.27797263860702515,
+      "learning_rate": 2e-05,
+      "loss": 1.3697,
+      "step": 425
+    },
+    {
+      "epoch": 2.108910891089109,
+      "grad_norm": 0.27797263860702515,
+      "learning_rate": 2e-05,
+      "loss": 1.3936,
+      "step": 426
+    },
+    {
+      "epoch": 2.113861386138614,
+      "grad_norm": 0.27797263860702515,
+      "learning_rate": 2e-05,
+      "loss": 1.3869,
+      "step": 427
+    },
+    {
+      "epoch": 2.118811881188119,
+      "grad_norm": 0.2979559302330017,
+      "learning_rate": 2e-05,
+      "loss": 1.4054,
+      "step": 428
+    },
+    {
+      "epoch": 2.123762376237624,
+      "grad_norm": 0.2979559302330017,
+      "learning_rate": 2e-05,
+      "loss": 1.3993,
+      "step": 429
+    },
+    {
+      "epoch": 2.128712871287129,
+      "grad_norm": 0.2979559302330017,
+      "learning_rate": 2e-05,
+      "loss": 1.4286,
+      "step": 430
+    },
+    {
+      "epoch": 2.133663366336634,
+      "grad_norm": 0.2979559302330017,
+      "learning_rate": 2e-05,
+      "loss": 1.417,
+      "step": 431
+    },
+    {
+      "epoch": 2.1386138613861387,
+      "grad_norm": 0.28421610593795776,
+      "learning_rate": 2e-05,
+      "loss": 1.3958,
+      "step": 432
+    },
+    {
+      "epoch": 2.1435643564356437,
+      "grad_norm": 0.28421610593795776,
+      "learning_rate": 2e-05,
+      "loss": 1.4134,
+      "step": 433
+    },
+    {
+      "epoch": 2.1485148514851486,
+      "grad_norm": 0.28421610593795776,
+      "learning_rate": 2e-05,
+      "loss": 1.418,
+      "step": 434
+    },
+    {
+      "epoch": 2.1534653465346536,
+      "grad_norm": 0.28421610593795776,
+      "learning_rate": 2e-05,
+      "loss": 1.3599,
+      "step": 435
+    },
+    {
+      "epoch": 2.1584158415841586,
+      "grad_norm": 0.29607468843460083,
+      "learning_rate": 2e-05,
+      "loss": 1.4297,
+      "step": 436
+    },
+    {
+      "epoch": 2.1633663366336635,
+      "grad_norm": 0.29607468843460083,
+      "learning_rate": 2e-05,
+      "loss": 1.3682,
+      "step": 437
+    },
+    {
+      "epoch": 2.1683168316831685,
+      "grad_norm": 0.29607468843460083,
+      "learning_rate": 2e-05,
+      "loss": 1.4431,
+      "step": 438
+    },
+    {
+      "epoch": 2.1732673267326734,
+      "grad_norm": 0.29607468843460083,
+      "learning_rate": 2e-05,
+      "loss": 1.4598,
+      "step": 439
+    },
+    {
+      "epoch": 2.1782178217821784,
+      "grad_norm": 0.30585789680480957,
+      "learning_rate": 2e-05,
+      "loss": 1.3998,
+      "step": 440
+    },
+    {
+      "epoch": 2.1831683168316833,
+      "grad_norm": 0.30585789680480957,
+      "learning_rate": 2e-05,
+      "loss": 1.3854,
+      "step": 441
+    },
+    {
+      "epoch": 2.1881188118811883,
+      "grad_norm": 0.30585789680480957,
+      "learning_rate": 2e-05,
+      "loss": 1.3953,
+      "step": 442
+    },
+    {
+      "epoch": 2.1930693069306932,
+      "grad_norm": 0.30585789680480957,
+      "learning_rate": 2e-05,
+      "loss": 1.3797,
+      "step": 443
+    },
+    {
+      "epoch": 2.198019801980198,
+      "grad_norm": 0.28262895345687866,
+      "learning_rate": 2e-05,
+      "loss": 1.3731,
+      "step": 444
+    },
+    {
+      "epoch": 2.202970297029703,
+      "grad_norm": 0.28262895345687866,
+      "learning_rate": 2e-05,
+      "loss": 1.4521,
+      "step": 445
+    },
+    {
+      "epoch": 2.207920792079208,
+      "grad_norm": 0.28262895345687866,
+      "learning_rate": 2e-05,
+      "loss": 1.4159,
+      "step": 446
+    },
+    {
+      "epoch": 2.212871287128713,
+      "grad_norm": 0.28262895345687866,
+      "learning_rate": 2e-05,
+      "loss": 1.4149,
+      "step": 447
+    },
+    {
+      "epoch": 2.217821782178218,
+      "grad_norm": 0.2773006558418274,
+      "learning_rate": 2e-05,
+      "loss": 1.3996,
+      "step": 448
+    },
+    {
+      "epoch": 2.222772277227723,
+      "grad_norm": 0.2773006558418274,
+      "learning_rate": 2e-05,
+      "loss": 1.3302,
+      "step": 449
+    },
+    {
+      "epoch": 2.227722772277228,
+      "grad_norm": 0.2773006558418274,
+      "learning_rate": 2e-05,
+      "loss": 1.3364,
+      "step": 450
+    },
+    {
+      "epoch": 2.232673267326733,
+      "grad_norm": 0.2773006558418274,
+      "learning_rate": 2e-05,
+      "loss": 1.4101,
+      "step": 451
+    },
+    {
+      "epoch": 2.237623762376238,
+      "grad_norm": 0.2921248972415924,
+      "learning_rate": 2e-05,
+      "loss": 1.3372,
+      "step": 452
+    },
+    {
+      "epoch": 2.2425742574257423,
+      "grad_norm": 0.2921248972415924,
+      "learning_rate": 2e-05,
+      "loss": 1.433,
+      "step": 453
+    },
+    {
+      "epoch": 2.2475247524752477,
+      "grad_norm": 0.2921248972415924,
+      "learning_rate": 2e-05,
+      "loss": 1.3784,
+      "step": 454
+    },
+    {
+      "epoch": 2.2524752475247523,
+      "grad_norm": 0.2921248972415924,
+      "learning_rate": 2e-05,
+      "loss": 1.3843,
+      "step": 455
+    },
+    {
+      "epoch": 2.2574257425742577,
+      "grad_norm": 0.2769683599472046,
+      "learning_rate": 2e-05,
+      "loss": 1.3851,
+      "step": 456
+    },
+    {
+      "epoch": 2.262376237623762,
+      "grad_norm": 0.2769683599472046,
+      "learning_rate": 2e-05,
+      "loss": 1.416,
+      "step": 457
+    },
+    {
+      "epoch": 2.2673267326732676,
+      "grad_norm": 0.2769683599472046,
+      "learning_rate": 2e-05,
+      "loss": 1.3678,
+      "step": 458
+    },
+    {
+      "epoch": 2.272277227722772,
+      "grad_norm": 0.2769683599472046,
+      "learning_rate": 2e-05,
+      "loss": 1.3452,
+      "step": 459
+    },
+    {
+      "epoch": 2.2772277227722775,
+      "grad_norm": 0.285204142332077,
+      "learning_rate": 2e-05,
+      "loss": 1.4036,
+      "step": 460
+    },
+    {
+      "epoch": 2.282178217821782,
+      "grad_norm": 0.285204142332077,
+      "learning_rate": 2e-05,
+      "loss": 1.3349,
+      "step": 461
+    },
+    {
+      "epoch": 2.287128712871287,
+      "grad_norm": 0.285204142332077,
+      "learning_rate": 2e-05,
+      "loss": 1.4302,
+      "step": 462
+    },
+    {
+      "epoch": 2.292079207920792,
+      "grad_norm": 0.285204142332077,
+      "learning_rate": 2e-05,
+      "loss": 1.3687,
+      "step": 463
+    },
+    {
+      "epoch": 2.297029702970297,
+      "grad_norm": 0.27696162462234497,
+      "learning_rate": 2e-05,
+      "loss": 1.4433,
+      "step": 464
+    },
+    {
+      "epoch": 2.301980198019802,
+      "grad_norm": 0.27696162462234497,
+      "learning_rate": 2e-05,
+      "loss": 1.4201,
+      "step": 465
+    },
+    {
+      "epoch": 2.3069306930693068,
+      "grad_norm": 0.27696162462234497,
+      "learning_rate": 2e-05,
+      "loss": 1.4355,
+      "step": 466
+    },
+    {
+      "epoch": 2.3118811881188117,
+      "grad_norm": 0.27696162462234497,
+      "learning_rate": 2e-05,
+      "loss": 1.4331,
+      "step": 467
+    },
+    {
+      "epoch": 2.3168316831683167,
+      "grad_norm": 0.2807972729206085,
+      "learning_rate": 2e-05,
+      "loss": 1.3849,
+      "step": 468
+    },
+    {
+      "epoch": 2.3217821782178216,
+      "grad_norm": 0.2807972729206085,
+      "learning_rate": 2e-05,
+      "loss": 1.3458,
+      "step": 469
+    },
+    {
+      "epoch": 2.3267326732673266,
+      "grad_norm": 0.2807972729206085,
+      "learning_rate": 2e-05,
+      "loss": 1.4451,
+      "step": 470
+    },
+    {
+      "epoch": 2.3316831683168315,
+      "grad_norm": 0.2807972729206085,
+      "learning_rate": 2e-05,
+      "loss": 1.3855,
+      "step": 471
+    },
+    {
+      "epoch": 2.3366336633663365,
+      "grad_norm": 0.27739667892456055,
+      "learning_rate": 2e-05,
+      "loss": 1.4201,
+      "step": 472
+    },
+    {
+      "epoch": 2.3415841584158414,
+      "grad_norm": 0.27739667892456055,
+      "learning_rate": 2e-05,
+      "loss": 1.3699,
+      "step": 473
+    },
+    {
+      "epoch": 2.3465346534653464,
+      "grad_norm": 0.27739667892456055,
+      "learning_rate": 2e-05,
+      "loss": 1.3877,
+      "step": 474
+    },
+    {
+      "epoch": 2.3514851485148514,
+      "grad_norm": 0.27739667892456055,
+      "learning_rate": 2e-05,
+      "loss": 1.3925,
+      "step": 475
+    },
+    {
+      "epoch": 2.3564356435643563,
+      "grad_norm": 0.2919694483280182,
+      "learning_rate": 2e-05,
+      "loss": 1.3989,
+      "step": 476
+    },
+    {
+      "epoch": 2.3613861386138613,
+      "grad_norm": 0.2919694483280182,
+      "learning_rate": 2e-05,
+      "loss": 1.4233,
+      "step": 477
+    },
+    {
+      "epoch": 2.366336633663366,
+      "grad_norm": 0.2919694483280182,
+      "learning_rate": 2e-05,
+      "loss": 1.3902,
+      "step": 478
+    },
+    {
+      "epoch": 2.371287128712871,
+      "grad_norm": 0.2919694483280182,
+      "learning_rate": 2e-05,
+      "loss": 1.4339,
+      "step": 479
+    },
+    {
+      "epoch": 2.376237623762376,
+      "grad_norm": 0.276358962059021,
+      "learning_rate": 2e-05,
+      "loss": 1.3965,
+      "step": 480
+    },
+    {
+      "epoch": 2.381188118811881,
+      "grad_norm": 0.276358962059021,
+      "learning_rate": 2e-05,
+      "loss": 1.3959,
+      "step": 481
+    },
+    {
+      "epoch": 2.386138613861386,
+      "grad_norm": 0.276358962059021,
+      "learning_rate": 2e-05,
+      "loss": 1.3689,
+      "step": 482
+    },
+    {
+      "epoch": 2.391089108910891,
+      "grad_norm": 0.276358962059021,
+      "learning_rate": 2e-05,
+      "loss": 1.411,
+      "step": 483
+    },
+    {
+      "epoch": 2.396039603960396,
+      "grad_norm": 0.2838156223297119,
+      "learning_rate": 2e-05,
+      "loss": 1.4069,
+      "step": 484
+    },
+    {
+      "epoch": 2.400990099009901,
+      "grad_norm": 0.2838156223297119,
+      "learning_rate": 2e-05,
+      "loss": 1.3636,
+      "step": 485
+    },
+    {
+      "epoch": 2.405940594059406,
+      "grad_norm": 0.2838156223297119,
+      "learning_rate": 2e-05,
+      "loss": 1.3428,
+      "step": 486
+    },
+    {
+      "epoch": 2.410891089108911,
+      "grad_norm": 0.2838156223297119,
+      "learning_rate": 2e-05,
+      "loss": 1.3681,
+      "step": 487
+    },
+    {
+      "epoch": 2.4158415841584158,
+      "grad_norm": 0.2741771340370178,
+      "learning_rate": 2e-05,
+      "loss": 1.3872,
+      "step": 488
+    },
+    {
+      "epoch": 2.4207920792079207,
+      "grad_norm": 0.2741771340370178,
+      "learning_rate": 2e-05,
+      "loss": 1.357,
+      "step": 489
+    },
+    {
+      "epoch": 2.4257425742574257,
+      "grad_norm": 0.2741771340370178,
+      "learning_rate": 2e-05,
+      "loss": 1.3631,
+      "step": 490
+    },
+    {
+      "epoch": 2.4306930693069306,
+      "grad_norm": 0.2741771340370178,
+      "learning_rate": 2e-05,
+      "loss": 1.4068,
+      "step": 491
+    },
+    {
+      "epoch": 2.4356435643564356,
+      "grad_norm": 0.2892126441001892,
+      "learning_rate": 2e-05,
+      "loss": 1.3679,
+      "step": 492
+    },
+    {
+      "epoch": 2.4405940594059405,
+      "grad_norm": 0.2892126441001892,
+      "learning_rate": 2e-05,
+      "loss": 1.3479,
+      "step": 493
+    },
+    {
+      "epoch": 2.4455445544554455,
+      "grad_norm": 0.2892126441001892,
+      "learning_rate": 2e-05,
+      "loss": 1.4556,
+      "step": 494
+    },
+    {
+      "epoch": 2.4504950495049505,
+      "grad_norm": 0.2892126441001892,
+      "learning_rate": 2e-05,
+      "loss": 1.3408,
+      "step": 495
+    },
+    {
+      "epoch": 2.4554455445544554,
+      "grad_norm": 0.28148889541625977,
+      "learning_rate": 2e-05,
+      "loss": 1.3008,
+      "step": 496
+    },
+    {
+      "epoch": 2.4603960396039604,
+      "grad_norm": 0.28148889541625977,
+      "learning_rate": 2e-05,
+      "loss": 1.4694,
+      "step": 497
+    },
+    {
+      "epoch": 2.4653465346534653,
+      "grad_norm": 0.28148889541625977,
+      "learning_rate": 2e-05,
+      "loss": 1.3694,
+      "step": 498
+    },
+    {
+      "epoch": 2.4702970297029703,
+      "grad_norm": 0.28148889541625977,
+      "learning_rate": 2e-05,
+      "loss": 1.4251,
+      "step": 499
+    },
+    {
+      "epoch": 2.4752475247524752,
+      "grad_norm": 0.29076042771339417,
+      "learning_rate": 2e-05,
+      "loss": 1.3562,
+      "step": 500
+    },
+    {
+      "epoch": 2.48019801980198,
+      "grad_norm": 0.29076042771339417,
+      "learning_rate": 2e-05,
+      "loss": 1.3657,
+      "step": 501
+    },
+    {
+      "epoch": 2.485148514851485,
+      "grad_norm": 0.29076042771339417,
+      "learning_rate": 2e-05,
+      "loss": 1.3946,
+      "step": 502
+    },
+    {
+      "epoch": 2.49009900990099,
+      "grad_norm": 0.29076042771339417,
+      "learning_rate": 2e-05,
+      "loss": 1.3492,
+      "step": 503
+    },
+    {
+      "epoch": 2.495049504950495,
+      "grad_norm": 0.27691689133644104,
+      "learning_rate": 2e-05,
+      "loss": 1.3678,
+      "step": 504
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.27691689133644104,
+      "learning_rate": 2e-05,
+      "loss": 1.4146,
+      "step": 505
+    },
+    {
+      "epoch": 2.504950495049505,
+      "grad_norm": 0.27691689133644104,
+      "learning_rate": 2e-05,
+      "loss": 1.4215,
+      "step": 506
+    },
+    {
+      "epoch": 2.50990099009901,
+      "grad_norm": 0.27691689133644104,
+      "learning_rate": 2e-05,
+      "loss": 1.3759,
+      "step": 507
+    },
+    {
+      "epoch": 2.514851485148515,
+      "grad_norm": 0.27186834812164307,
+      "learning_rate": 2e-05,
+      "loss": 1.339,
+      "step": 508
+    },
+    {
+      "epoch": 2.51980198019802,
+      "grad_norm": 0.27186834812164307,
+      "learning_rate": 2e-05,
+      "loss": 1.3958,
+      "step": 509
+    },
+    {
+      "epoch": 2.5247524752475248,
+      "grad_norm": 0.27186834812164307,
+      "learning_rate": 2e-05,
+      "loss": 1.3675,
+      "step": 510
+    },
+    {
+      "epoch": 2.5297029702970297,
+      "grad_norm": 0.27186834812164307,
+      "learning_rate": 2e-05,
+      "loss": 1.3766,
+      "step": 511
+    },
+    {
+      "epoch": 2.5346534653465347,
+      "grad_norm": 0.3018421232700348,
+      "learning_rate": 2e-05,
+      "loss": 1.3505,
+      "step": 512
+    },
+    {
+      "epoch": 2.5396039603960396,
+      "grad_norm": 0.3018421232700348,
+      "learning_rate": 2e-05,
+      "loss": 1.3001,
+      "step": 513
+    },
+    {
+      "epoch": 2.5445544554455446,
+      "grad_norm": 0.3018421232700348,
+      "learning_rate": 2e-05,
+      "loss": 1.3361,
+      "step": 514
+    },
+    {
+      "epoch": 2.5495049504950495,
+      "grad_norm": 0.3018421232700348,
+      "learning_rate": 2e-05,
+      "loss": 1.4107,
+      "step": 515
+    },
+    {
+      "epoch": 2.5544554455445545,
+      "grad_norm": 0.2854144871234894,
+      "learning_rate": 2e-05,
+      "loss": 1.4126,
+      "step": 516
+    },
+    {
+      "epoch": 2.5594059405940595,
+      "grad_norm": 0.2854144871234894,
+      "learning_rate": 2e-05,
+      "loss": 1.3747,
+      "step": 517
+    },
+    {
+      "epoch": 2.5643564356435644,
+      "grad_norm": 0.2854144871234894,
+      "learning_rate": 2e-05,
+      "loss": 1.3408,
+      "step": 518
+    },
+    {
+      "epoch": 2.5693069306930694,
+      "grad_norm": 0.2854144871234894,
+      "learning_rate": 2e-05,
+      "loss": 1.3512,
+      "step": 519
+    },
+    {
+      "epoch": 2.5742574257425743,
+      "grad_norm": 0.2874051332473755,
+      "learning_rate": 2e-05,
+      "loss": 1.4058,
+      "step": 520
+    },
+    {
+      "epoch": 2.5792079207920793,
+      "grad_norm": 0.2874051332473755,
+      "learning_rate": 2e-05,
+      "loss": 1.4259,
+      "step": 521
+    },
+    {
+      "epoch": 2.5841584158415842,
+      "grad_norm": 0.2874051332473755,
+      "learning_rate": 2e-05,
+      "loss": 1.3752,
+      "step": 522
+    },
+    {
+      "epoch": 2.589108910891089,
+      "grad_norm": 0.2874051332473755,
+      "learning_rate": 2e-05,
+      "loss": 1.3503,
+      "step": 523
+    },
+    {
+      "epoch": 2.594059405940594,
+      "grad_norm": 0.27819254994392395,
+      "learning_rate": 2e-05,
+      "loss": 1.3721,
+      "step": 524
+    },
+    {
+      "epoch": 2.599009900990099,
+      "grad_norm": 0.27819254994392395,
+      "learning_rate": 2e-05,
+      "loss": 1.4515,
+      "step": 525
+    },
+    {
+      "epoch": 2.603960396039604,
+      "grad_norm": 0.27819254994392395,
+      "learning_rate": 2e-05,
+      "loss": 1.4081,
+      "step": 526
+    },
+    {
+      "epoch": 2.608910891089109,
+      "grad_norm": 0.27819254994392395,
+      "learning_rate": 2e-05,
+      "loss": 1.3186,
+      "step": 527
+    },
+    {
+      "epoch": 2.613861386138614,
+      "grad_norm": 0.27905669808387756,
+      "learning_rate": 2e-05,
+      "loss": 1.3056,
+      "step": 528
+    },
+    {
+      "epoch": 2.618811881188119,
+      "grad_norm": 0.27905669808387756,
+      "learning_rate": 2e-05,
+      "loss": 1.3097,
+      "step": 529
+    },
+    {
+      "epoch": 2.623762376237624,
+      "grad_norm": 0.27905669808387756,
+      "learning_rate": 2e-05,
+      "loss": 1.3153,
+      "step": 530
+    },
+    {
+      "epoch": 2.628712871287129,
+      "grad_norm": 0.27905669808387756,
+      "learning_rate": 2e-05,
+      "loss": 1.345,
+      "step": 531
+    },
+    {
+      "epoch": 2.633663366336634,
+      "grad_norm": 0.29262369871139526,
+      "learning_rate": 2e-05,
+      "loss": 1.3759,
+      "step": 532
+    },
+    {
+      "epoch": 2.6386138613861387,
+      "grad_norm": 0.29262369871139526,
+      "learning_rate": 2e-05,
+      "loss": 1.3587,
+      "step": 533
+    },
+    {
+      "epoch": 2.6435643564356437,
+      "grad_norm": 0.29262369871139526,
+      "learning_rate": 2e-05,
+      "loss": 1.3036,
+      "step": 534
+    },
+    {
+      "epoch": 2.6485148514851486,
+      "grad_norm": 0.29262369871139526,
+      "learning_rate": 2e-05,
+      "loss": 1.3814,
+      "step": 535
+    },
+    {
+      "epoch": 2.6534653465346536,
+      "grad_norm": 0.26941347122192383,
+      "learning_rate": 2e-05,
+      "loss": 1.2526,
+      "step": 536
+    },
+    {
+      "epoch": 2.6584158415841586,
+      "grad_norm": 0.26941347122192383,
+      "learning_rate": 2e-05,
+      "loss": 1.323,
+      "step": 537
+    },
+    {
+      "epoch": 2.6633663366336635,
+      "grad_norm": 0.26941347122192383,
+      "learning_rate": 2e-05,
+      "loss": 1.3945,
+      "step": 538
+    },
+    {
+      "epoch": 2.6683168316831685,
+      "grad_norm": 0.26941347122192383,
+      "learning_rate": 2e-05,
+      "loss": 1.3668,
+      "step": 539
+    },
+    {
+      "epoch": 2.6732673267326734,
+      "grad_norm": 0.2864089012145996,
+      "learning_rate": 2e-05,
+      "loss": 1.3519,
+      "step": 540
+    },
+    {
+      "epoch": 2.6782178217821784,
+      "grad_norm": 0.2864089012145996,
+      "learning_rate": 2e-05,
+      "loss": 1.402,
+      "step": 541
+    },
+    {
+      "epoch": 2.6831683168316833,
+      "grad_norm": 0.2864089012145996,
+      "learning_rate": 2e-05,
+      "loss": 1.3737,
+      "step": 542
+    },
+    {
+      "epoch": 2.6881188118811883,
+      "grad_norm": 0.2864089012145996,
+      "learning_rate": 2e-05,
+      "loss": 1.345,
+      "step": 543
+    },
+    {
+      "epoch": 2.693069306930693,
+      "grad_norm": 0.2727324664592743,
+      "learning_rate": 2e-05,
+      "loss": 1.3741,
+      "step": 544
+    },
+    {
+      "epoch": 2.698019801980198,
+      "grad_norm": 0.2727324664592743,
+      "learning_rate": 2e-05,
+      "loss": 1.3779,
+      "step": 545
+    },
+    {
+      "epoch": 2.7029702970297027,
+      "grad_norm": 0.2727324664592743,
+      "learning_rate": 2e-05,
+      "loss": 1.4057,
+      "step": 546
+    },
+    {
+      "epoch": 2.707920792079208,
+      "grad_norm": 0.2727324664592743,
+      "learning_rate": 2e-05,
+      "loss": 1.3702,
+      "step": 547
+    },
+    {
+      "epoch": 2.7128712871287126,
+      "grad_norm": 0.2906956970691681,
+      "learning_rate": 2e-05,
+      "loss": 1.4395,
+      "step": 548
+    },
+    {
+      "epoch": 2.717821782178218,
+      "grad_norm": 0.2906956970691681,
+      "learning_rate": 2e-05,
+      "loss": 1.3275,
+      "step": 549
+    },
+    {
+      "epoch": 2.7227722772277225,
+      "grad_norm": 0.2906956970691681,
+      "learning_rate": 2e-05,
+      "loss": 1.396,
+      "step": 550
+    },
+    {
+      "epoch": 2.727722772277228,
+      "grad_norm": 0.2906956970691681,
+      "learning_rate": 2e-05,
+      "loss": 1.3898,
+      "step": 551
+    },
+    {
+      "epoch": 2.7326732673267324,
+      "grad_norm": 0.2827296555042267,
+      "learning_rate": 2e-05,
+      "loss": 1.46,
+      "step": 552
+    },
+    {
+      "epoch": 2.737623762376238,
+      "grad_norm": 0.2827296555042267,
+      "learning_rate": 2e-05,
+      "loss": 1.3724,
+      "step": 553
+    },
+    {
+      "epoch": 2.7425742574257423,
+      "grad_norm": 0.2827296555042267,
+      "learning_rate": 2e-05,
+      "loss": 1.4224,
+      "step": 554
+    },
+    {
+      "epoch": 2.7475247524752477,
+      "grad_norm": 0.2827296555042267,
+      "learning_rate": 2e-05,
+      "loss": 1.3255,
+      "step": 555
+    },
+    {
+      "epoch": 2.7524752475247523,
+      "grad_norm": 0.2894042730331421,
+      "learning_rate": 2e-05,
+      "loss": 1.3698,
+      "step": 556
+    },
+    {
+      "epoch": 2.7574257425742577,
+      "grad_norm": 0.2894042730331421,
+      "learning_rate": 2e-05,
+      "loss": 1.3767,
+      "step": 557
+    },
+    {
+      "epoch": 2.762376237623762,
+      "grad_norm": 0.2894042730331421,
+      "learning_rate": 2e-05,
+      "loss": 1.3385,
+      "step": 558
+    },
+    {
+      "epoch": 2.7673267326732676,
+      "grad_norm": 0.2894042730331421,
+      "learning_rate": 2e-05,
+      "loss": 1.3326,
+      "step": 559
+    },
+    {
+      "epoch": 2.772277227722772,
+      "grad_norm": 0.28423652052879333,
+      "learning_rate": 2e-05,
+      "loss": 1.3213,
+      "step": 560
+    },
+    {
+      "epoch": 2.7772277227722775,
+      "grad_norm": 0.28423652052879333,
+      "learning_rate": 2e-05,
+      "loss": 1.3641,
+      "step": 561
+    },
+    {
+      "epoch": 2.782178217821782,
+      "grad_norm": 0.28423652052879333,
+      "learning_rate": 2e-05,
+      "loss": 1.37,
+      "step": 562
+    },
+    {
+      "epoch": 2.7871287128712874,
+      "grad_norm": 0.28423652052879333,
+      "learning_rate": 2e-05,
+      "loss": 1.321,
+      "step": 563
+    },
+    {
+      "epoch": 2.792079207920792,
+      "grad_norm": 0.27789822220802307,
+      "learning_rate": 2e-05,
+      "loss": 1.4119,
+      "step": 564
+    },
+    {
+      "epoch": 2.7970297029702973,
+      "grad_norm": 0.27789822220802307,
+      "learning_rate": 2e-05,
+      "loss": 1.351,
+      "step": 565
+    },
+    {
+      "epoch": 2.801980198019802,
+      "grad_norm": 0.27789822220802307,
+      "learning_rate": 2e-05,
+      "loss": 1.3619,
+      "step": 566
+    },
+    {
+      "epoch": 2.806930693069307,
+      "grad_norm": 0.27789822220802307,
+      "learning_rate": 2e-05,
+      "loss": 1.3856,
+      "step": 567
+    },
+    {
+      "epoch": 2.8118811881188117,
+      "grad_norm": 0.28625088930130005,
+      "learning_rate": 2e-05,
+      "loss": 1.3674,
+      "step": 568
+    },
+    {
+      "epoch": 2.8168316831683167,
+      "grad_norm": 0.28625088930130005,
+      "learning_rate": 2e-05,
+      "loss": 1.4114,
+      "step": 569
+    },
+    {
+      "epoch": 2.8217821782178216,
+      "grad_norm": 0.28625088930130005,
+      "learning_rate": 2e-05,
+      "loss": 1.4165,
+      "step": 570
+    },
+    {
+      "epoch": 2.8267326732673266,
+      "grad_norm": 0.28625088930130005,
+      "learning_rate": 2e-05,
+      "loss": 1.3547,
+      "step": 571
+    },
+    {
+      "epoch": 2.8316831683168315,
+      "grad_norm": 0.30029016733169556,
+      "learning_rate": 2e-05,
+      "loss": 1.3799,
+      "step": 572
+    },
+    {
+      "epoch": 2.8366336633663365,
+      "grad_norm": 0.30029016733169556,
+      "learning_rate": 2e-05,
+      "loss": 1.4145,
+      "step": 573
+    },
+    {
+      "epoch": 2.8415841584158414,
+      "grad_norm": 0.30029016733169556,
+      "learning_rate": 2e-05,
+      "loss": 1.3959,
+      "step": 574
+    },
+    {
+      "epoch": 2.8465346534653464,
+      "grad_norm": 0.30029016733169556,
+      "learning_rate": 2e-05,
+      "loss": 1.3416,
+      "step": 575
+    },
+    {
+      "epoch": 2.8514851485148514,
+      "grad_norm": 0.29112035036087036,
+      "learning_rate": 2e-05,
+      "loss": 1.3418,
+      "step": 576
+    },
+    {
+      "epoch": 2.8564356435643563,
+      "grad_norm": 0.29112035036087036,
+      "learning_rate": 2e-05,
+      "loss": 1.3522,
+      "step": 577
+    },
+    {
+      "epoch": 2.8613861386138613,
+      "grad_norm": 0.29112035036087036,
+      "learning_rate": 2e-05,
+      "loss": 1.3506,
+      "step": 578
+    },
+    {
+      "epoch": 2.866336633663366,
+      "grad_norm": 0.29112035036087036,
+      "learning_rate": 2e-05,
+      "loss": 1.3808,
+      "step": 579
+    },
+    {
+      "epoch": 2.871287128712871,
+      "grad_norm": 0.2819567024707794,
+      "learning_rate": 2e-05,
+      "loss": 1.3871,
+      "step": 580
+    },
+    {
+      "epoch": 2.876237623762376,
+      "grad_norm": 0.2819567024707794,
+      "learning_rate": 2e-05,
+      "loss": 1.3984,
+      "step": 581
+    },
+    {
+      "epoch": 2.881188118811881,
+      "grad_norm": 0.2819567024707794,
+      "learning_rate": 2e-05,
+      "loss": 1.3584,
+      "step": 582
+    },
+    {
+      "epoch": 2.886138613861386,
+      "grad_norm": 0.2819567024707794,
+      "learning_rate": 2e-05,
+      "loss": 1.3971,
+      "step": 583
+    },
+    {
+      "epoch": 2.891089108910891,
+      "grad_norm": 0.28793245553970337,
+      "learning_rate": 2e-05,
+      "loss": 1.4094,
+      "step": 584
+    },
+    {
+      "epoch": 2.896039603960396,
+      "grad_norm": 0.28793245553970337,
+      "learning_rate": 2e-05,
+      "loss": 1.396,
+      "step": 585
+    },
+    {
+      "epoch": 2.900990099009901,
+      "grad_norm": 0.28793245553970337,
+      "learning_rate": 2e-05,
+      "loss": 1.323,
+      "step": 586
+    },
+    {
+      "epoch": 2.905940594059406,
+      "grad_norm": 0.28793245553970337,
+      "learning_rate": 2e-05,
+      "loss": 1.3936,
+      "step": 587
+    },
+    {
+      "epoch": 2.910891089108911,
+      "grad_norm": 0.28640323877334595,
+      "learning_rate": 2e-05,
+      "loss": 1.3721,
+      "step": 588
+    },
+    {
+      "epoch": 2.9158415841584158,
+      "grad_norm": 0.28640323877334595,
+      "learning_rate": 2e-05,
+      "loss": 1.3501,
+      "step": 589
+    },
+    {
+      "epoch": 2.9207920792079207,
+      "grad_norm": 0.28640323877334595,
+      "learning_rate": 2e-05,
+      "loss": 1.3953,
+      "step": 590
+    },
+    {
+      "epoch": 2.9257425742574257,
+      "grad_norm": 0.28640323877334595,
+      "learning_rate": 2e-05,
+      "loss": 1.3597,
+      "step": 591
+    },
+    {
+      "epoch": 2.9306930693069306,
+      "grad_norm": 0.28491514921188354,
+      "learning_rate": 2e-05,
+      "loss": 1.3689,
+      "step": 592
+    },
+    {
+      "epoch": 2.9356435643564356,
+      "grad_norm": 0.28491514921188354,
+      "learning_rate": 2e-05,
+      "loss": 1.3209,
+      "step": 593
+    },
+    {
+      "epoch": 2.9405940594059405,
+      "grad_norm": 0.28491514921188354,
+      "learning_rate": 2e-05,
+      "loss": 1.4021,
+      "step": 594
+    },
+    {
+      "epoch": 2.9455445544554455,
+      "grad_norm": 0.28491514921188354,
+      "learning_rate": 2e-05,
+      "loss": 1.36,
+      "step": 595
+    },
+    {
+      "epoch": 2.9504950495049505,
+      "grad_norm": 0.2926208972930908,
+      "learning_rate": 2e-05,
+      "loss": 1.3803,
+      "step": 596
+    },
+    {
+      "epoch": 2.9554455445544554,
+      "grad_norm": 0.2926208972930908,
+      "learning_rate": 2e-05,
+      "loss": 1.3806,
+      "step": 597
+    },
+    {
+      "epoch": 2.9603960396039604,
+      "grad_norm": 0.2926208972930908,
+      "learning_rate": 2e-05,
+      "loss": 1.3569,
+      "step": 598
+    },
+    {
+      "epoch": 2.9653465346534653,
+      "grad_norm": 0.2926208972930908,
+      "learning_rate": 2e-05,
+      "loss": 1.3718,
+      "step": 599
+    },
+    {
+      "epoch": 2.9702970297029703,
+      "grad_norm": 0.2797441780567169,
+      "learning_rate": 2e-05,
+      "loss": 1.3376,
+      "step": 600
+    },
+    {
+      "epoch": 2.9752475247524752,
+      "grad_norm": 0.2797441780567169,
+      "learning_rate": 2e-05,
+      "loss": 1.3455,
+      "step": 601
+    },
+    {
+      "epoch": 2.98019801980198,
+      "grad_norm": 0.2797441780567169,
+      "learning_rate": 2e-05,
+      "loss": 1.4179,
+      "step": 602
+    },
+    {
+      "epoch": 2.985148514851485,
+      "grad_norm": 0.2797441780567169,
+      "learning_rate": 2e-05,
+      "loss": 1.3065,
+      "step": 603
+    },
+    {
+      "epoch": 2.99009900990099,
+      "grad_norm": 0.29282307624816895,
+      "learning_rate": 2e-05,
+      "loss": 1.3871,
+      "step": 604
+    },
+    {
+      "epoch": 2.995049504950495,
+      "grad_norm": 0.29282307624816895,
+      "learning_rate": 2e-05,
+      "loss": 1.3707,
+      "step": 605
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.29282307624816895,
+      "learning_rate": 2e-05,
+      "loss": 1.3113,
+      "step": 606
+    },
+    {
+      "epoch": 3.0,
+      "step": 606,
+      "total_flos": 3.877969000755364e+17,
+      "train_loss": 1.531072208000095,
+      "train_runtime": 382.5341,
+      "train_samples_per_second": 403.737,
+      "train_steps_per_second": 1.584
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 606,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 3000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.877969000755364e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}