diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.3769363166953528,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 0.9574,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 0.9515,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.21917808219178e-06,
+      "loss": 0.9591,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0958904109589042e-05,
+      "loss": 0.9551,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3698630136986302e-05,
+      "loss": 0.9557,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.643835616438356e-05,
+      "loss": 0.9524,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9178082191780822e-05,
+      "loss": 0.9489,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1917808219178083e-05,
+      "loss": 0.9493,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4657534246575342e-05,
+      "loss": 0.9476,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7397260273972603e-05,
+      "loss": 0.9391,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0136986301369862e-05,
+      "loss": 0.9392,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.287671232876712e-05,
+      "loss": 0.9396,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.561643835616438e-05,
+      "loss": 0.9326,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8356164383561644e-05,
+      "loss": 0.9276,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1095890410958905e-05,
+      "loss": 0.9194,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.383561643835617e-05,
+      "loss": 0.9124,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657534246575342e-05,
+      "loss": 0.902,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9315068493150684e-05,
+      "loss": 0.8943,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2054794520547945e-05,
+      "loss": 0.8736,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.479452054794521e-05,
+      "loss": 0.8599,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.753424657534247e-05,
+      "loss": 0.8389,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0273972602739724e-05,
+      "loss": 0.8196,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.301369863013699e-05,
+      "loss": 0.7979,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.575342465753424e-05,
+      "loss": 0.7722,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.84931506849315e-05,
+      "loss": 0.746,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.123287671232876e-05,
+      "loss": 0.7211,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.397260273972603e-05,
+      "loss": 0.6992,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.671232876712329e-05,
+      "loss": 0.6736,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.945205479452055e-05,
+      "loss": 0.6517,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.219178082191781e-05,
+      "loss": 0.6287,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.493150684931507e-05,
+      "loss": 0.6095,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.767123287671233e-05,
+      "loss": 0.592,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.041095890410958e-05,
+      "loss": 0.5689,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.315068493150684e-05,
+      "loss": 0.5526,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.58904109589041e-05,
+      "loss": 0.536,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.863013698630137e-05,
+      "loss": 0.5246,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010136986301369864,
+      "loss": 0.5052,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010410958904109589,
+      "loss": 0.4971,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010684931506849317,
+      "loss": 0.4868,
+      "step": 39
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00010958904109589041,
+      "loss": 0.4738,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011232876712328768,
+      "loss": 0.4694,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011506849315068494,
+      "loss": 0.4633,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001178082191780822,
+      "loss": 0.4565,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012054794520547945,
+      "loss": 0.451,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001232876712328767,
+      "loss": 0.4446,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012602739726027398,
+      "loss": 0.4439,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012876712328767123,
+      "loss": 0.4383,
+      "step": 47
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013150684931506848,
+      "loss": 0.4368,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013424657534246576,
+      "loss": 0.4224,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000136986301369863,
+      "loss": 0.4143,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013972602739726028,
+      "loss": 0.4087,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014246575342465753,
+      "loss": 0.4082,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001452054794520548,
+      "loss": 0.4045,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014794520547945205,
+      "loss": 0.3977,
+      "step": 54
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015068493150684933,
+      "loss": 0.3875,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015342465753424657,
+      "loss": 0.3883,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015616438356164385,
+      "loss": 0.3857,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001589041095890411,
+      "loss": 0.3822,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016164383561643837,
+      "loss": 0.3802,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016438356164383562,
+      "loss": 0.3814,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001671232876712329,
+      "loss": 0.3795,
+      "step": 61
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00016986301369863014,
+      "loss": 0.3781,
+      "step": 62
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017260273972602742,
+      "loss": 0.3764,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017534246575342467,
+      "loss": 0.375,
+      "step": 64
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017808219178082192,
+      "loss": 0.3696,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018082191780821916,
+      "loss": 0.3689,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018356164383561644,
+      "loss": 0.3657,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001863013698630137,
+      "loss": 0.3634,
+      "step": 68
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00018904109589041096,
+      "loss": 0.3662,
+      "step": 69
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001917808219178082,
+      "loss": 0.36,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019452054794520549,
+      "loss": 0.3576,
+      "step": 71
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019726027397260273,
+      "loss": 0.361,
+      "step": 72
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3545,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999974049780868,
+      "loss": 0.3502,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999896199258152,
+      "loss": 0.3506,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199997664488359,
+      "loss": 0.3485,
+      "step": 76
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019999584799187522,
+      "loss": 0.3488,
+      "step": 77
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999935125125579,
+      "loss": 0.3484,
+      "step": 78
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019999065806252829,
+      "loss": 0.3469,
+      "step": 79
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998728465660105,
+      "loss": 0.3428,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998339231228434,
+      "loss": 0.3401,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999789810497796,
+      "loss": 0.336,
+      "step": 82
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999740508919815,
+      "loss": 0.3409,
+      "step": 83
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999686018644777,
+      "loss": 0.3336,
+      "step": 84
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019996263399554897,
+      "loss": 0.3331,
+      "step": 85
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019995614731616875,
+      "loss": 0.337,
+      "step": 86
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019994914186000328,
+      "loss": 0.3351,
+      "step": 87
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999416176634111,
+      "loss": 0.331,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019993357476544312,
+      "loss": 0.3293,
+      "step": 89
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999250132078424,
+      "loss": 0.3279,
+      "step": 90
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019991593303504376,
+      "loss": 0.329,
+      "step": 91
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019990633429417363,
+      "loss": 0.3254,
+      "step": 92
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019989621703505,
+      "loss": 0.3286,
+      "step": 93
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019988558131018186,
+      "loss": 0.3263,
+      "step": 94
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019987442717476906,
+      "loss": 0.3303,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019986275468670205,
+      "loss": 0.3221,
+      "step": 96
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019985056390656162,
+      "loss": 0.3186,
+      "step": 97
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019983785489761837,
+      "loss": 0.3222,
+      "step": 98
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019982462772583266,
+      "loss": 0.3134,
+      "step": 99
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019981088245985408,
+      "loss": 0.3168,
+      "step": 100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019979661917102115,
+      "loss": 0.3171,
+      "step": 101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000199781837933361,
+      "loss": 0.3154,
+      "step": 102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019976653882358884,
+      "loss": 0.3154,
+      "step": 103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001997507219211078,
+      "loss": 0.3161,
+      "step": 104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019973438730800822,
+      "loss": 0.3115,
+      "step": 105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019971753506906753,
+      "loss": 0.3132,
+      "step": 106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019970016529174947,
+      "loss": 0.3106,
+      "step": 107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001996822780662041,
+      "loss": 0.3156,
+      "step": 108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019966387348526683,
+      "loss": 0.3106,
+      "step": 109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019964495164445824,
+      "loss": 0.3104,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001996255126419835,
+      "loss": 0.3114,
+      "step": 111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001996055565787319,
+      "loss": 0.3044,
+      "step": 112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001995850835582763,
+      "loss": 0.3062,
+      "step": 113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019956409368687258,
+      "loss": 0.3075,
+      "step": 114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000199542587073459,
+      "loss": 0.2978,
+      "step": 115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019952056382965597,
+      "loss": 0.3087,
+      "step": 116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019949802406976495,
+      "loss": 0.3026,
+      "step": 117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019947496791076837,
+      "loss": 0.2965,
+      "step": 118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019945139547232872,
+      "loss": 0.3037,
+      "step": 119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001994273068767879,
+      "loss": 0.301,
+      "step": 120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019940270224916688,
+      "loss": 0.2994,
+      "step": 121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019937758171716468,
+      "loss": 0.2968,
+      "step": 122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000199351945411158,
+      "loss": 0.2999,
+      "step": 123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019932579346420038,
+      "loss": 0.2972,
+      "step": 124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019929912601202151,
+      "loss": 0.2893,
+      "step": 125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019927194319302677,
+      "loss": 0.2973,
+      "step": 126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019924424514829606,
+      "loss": 0.287,
+      "step": 127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019921603202158354,
+      "loss": 0.2933,
+      "step": 128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019918730395931649,
+      "loss": 0.293,
+      "step": 129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019915806111059486,
+      "loss": 0.29,
+      "step": 130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001991283036271903,
+      "loss": 0.2836,
+      "step": 131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990980316635455,
+      "loss": 0.2884,
+      "step": 132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019906724537677316,
+      "loss": 0.2879,
+      "step": 133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019903594492665558,
+      "loss": 0.2853,
+      "step": 134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001990041304756434,
+      "loss": 0.2901,
+      "step": 135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019897180218885507,
+      "loss": 0.281,
+      "step": 136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019893896023407578,
+      "loss": 0.2869,
+      "step": 137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989056047817567,
+      "loss": 0.286,
+      "step": 138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001988717360050141,
+      "loss": 0.2887,
+      "step": 139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019883735407962846,
+      "loss": 0.2919,
+      "step": 140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019880245918404342,
+      "loss": 0.2876,
+      "step": 141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000198767051499365,
+      "loss": 0.2875,
+      "step": 142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019873113120936074,
+      "loss": 0.2783,
+      "step": 143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019869469850045842,
+      "loss": 0.2903,
+      "step": 144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019865775356174545,
+      "loss": 0.2831,
+      "step": 145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019862029658496762,
+      "loss": 0.2825,
+      "step": 146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019858232776452837,
+      "loss": 0.2847,
+      "step": 147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019854384729748746,
+      "loss": 0.2796,
+      "step": 148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019850485538356027,
+      "loss": 0.275,
+      "step": 149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001984653522251165,
+      "loss": 0.2796,
+      "step": 150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019842533802717923,
+      "loss": 0.2793,
+      "step": 151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019838481299742398,
+      "loss": 0.2821,
+      "step": 152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001983437773461774,
+      "loss": 0.2774,
+      "step": 153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019830223128641637,
+      "loss": 0.2754,
+      "step": 154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019826017503376666,
+      "loss": 0.2749,
+      "step": 155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019821760880650214,
+      "loss": 0.2783,
+      "step": 156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019817453282554333,
+      "loss": 0.2779,
+      "step": 157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019813094731445654,
+      "loss": 0.2837,
+      "step": 158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019808685249945245,
+      "loss": 0.2744,
+      "step": 159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019804224860938506,
+      "loss": 0.2814,
+      "step": 160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001979971358757505,
+      "loss": 0.2804,
+      "step": 161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001979515145326859,
+      "loss": 0.2769,
+      "step": 162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000197905384816968,
+      "loss": 0.2718,
+      "step": 163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019785874696801202,
+      "loss": 0.2781,
+      "step": 164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019781160122787046,
+      "loss": 0.273,
+      "step": 165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019776394784123177,
+      "loss": 0.2792,
+      "step": 166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019771578705541916,
+      "loss": 0.273,
+      "step": 167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019766711912038915,
+      "loss": 0.2701,
+      "step": 168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001976179442887305,
+      "loss": 0.2718,
+      "step": 169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019756826281566272,
+      "loss": 0.2736,
+      "step": 170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019751807495903484,
+      "loss": 0.266,
+      "step": 171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019746738097932407,
+      "loss": 0.2677,
+      "step": 172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001974161811396343,
+      "loss": 0.2649,
+      "step": 173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019736447570569503,
+      "loss": 0.2775,
+      "step": 174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001973122649458597,
+      "loss": 0.2638,
+      "step": 175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019725954913110442,
+      "loss": 0.2693,
+      "step": 176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001972063285350266,
+      "loss": 0.2662,
+      "step": 177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019715260343384347,
+      "loss": 0.2653,
+      "step": 178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019709837410639063,
+      "loss": 0.2686,
+      "step": 179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001970436408341207,
+      "loss": 0.2726,
+      "step": 180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019698840390110176,
+      "loss": 0.2601,
+      "step": 181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001969326635940159,
+      "loss": 0.27,
+      "step": 182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019687642020215775,
+      "loss": 0.2664,
+      "step": 183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019681967401743297,
+      "loss": 0.2672,
+      "step": 184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019676242533435678,
+      "loss": 0.2688,
+      "step": 185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019670467445005233,
+      "loss": 0.2646,
+      "step": 186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019664642166424928,
+      "loss": 0.2654,
+      "step": 187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019658766727928206,
+      "loss": 0.2656,
+      "step": 188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019652841160008858,
+      "loss": 0.2657,
+      "step": 189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001964686549342084,
+      "loss": 0.2665,
+      "step": 190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019640839759178116,
+      "loss": 0.2663,
+      "step": 191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019634763988554522,
+      "loss": 0.2628,
+      "step": 192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019628638213083565,
+      "loss": 0.2627,
+      "step": 193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019622462464558295,
+      "loss": 0.263,
+      "step": 194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019616236775031113,
+      "loss": 0.2595,
+      "step": 195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019609961176813624,
+      "loss": 0.2636,
+      "step": 196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001960363570247645,
+      "loss": 0.259,
+      "step": 197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001959726038484909,
+      "loss": 0.2602,
+      "step": 198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019590835257019714,
+      "loss": 0.2635,
+      "step": 199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019584360352335023,
+      "loss": 0.2558,
+      "step": 200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001957783570440005,
+      "loss": 0.2589,
+      "step": 201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001957126134707801,
+      "loss": 0.2644,
+      "step": 202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019564637314490108,
+      "loss": 0.2639,
+      "step": 203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001955796364101535,
+      "loss": 0.2541,
+      "step": 204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019551240361290407,
+      "loss": 0.2566,
+      "step": 205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019544467510209388,
+      "loss": 0.2549,
+      "step": 206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001953764512292369,
+      "loss": 0.2551,
+      "step": 207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019530773234841803,
+      "loss": 0.2559,
+      "step": 208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019523851881629126,
+      "loss": 0.2479,
+      "step": 209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001951688109920778,
+      "loss": 0.2587,
+      "step": 210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019509860923756442,
+      "loss": 0.2556,
+      "step": 211
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019502791391710125,
+      "loss": 0.2572,
+      "step": 212
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019495672539760007,
+      "loss": 0.2557,
+      "step": 213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019488504404853248,
+      "loss": 0.2532,
+      "step": 214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019481287024192775,
+      "loss": 0.254,
+      "step": 215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019474020435237117,
+      "loss": 0.2538,
+      "step": 216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019466704675700185,
+      "loss": 0.2588,
+      "step": 217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019459339783551094,
+      "loss": 0.2573,
+      "step": 218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019451925797013954,
+      "loss": 0.2533,
+      "step": 219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019444462754567682,
+      "loss": 0.2539,
+      "step": 220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019436950694945798,
+      "loss": 0.2561,
+      "step": 221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019429389657136213,
+      "loss": 0.2569,
+      "step": 222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019421779680381054,
+      "loss": 0.2497,
+      "step": 223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019414120804176426,
+      "loss": 0.2493,
+      "step": 224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019406413068272238,
+      "loss": 0.2527,
+      "step": 225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019398656512671972,
+      "loss": 0.2541,
+      "step": 226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019390851177632497,
+      "loss": 0.2541,
+      "step": 227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019382997103663838,
+      "loss": 0.2503,
+      "step": 228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001937509433152899,
+      "loss": 0.2586,
+      "step": 229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001936714290224368,
+      "loss": 0.2519,
+      "step": 230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019359142857076176,
+      "loss": 0.2466,
+      "step": 231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019351094237547066,
+      "loss": 0.2428,
+      "step": 232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001934299708542904,
+      "loss": 0.248,
+      "step": 233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019334851442746664,
+      "loss": 0.2404,
+      "step": 234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019326657351776186,
+      "loss": 0.2463,
+      "step": 235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000193184148550453,
+      "loss": 0.2506,
+      "step": 236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019310123995332917,
+      "loss": 0.2477,
+      "step": 237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019301784815668974,
+      "loss": 0.2445,
+      "step": 238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019293397359334167,
+      "loss": 0.2452,
+      "step": 239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019284961669859766,
+      "loss": 0.2489,
+      "step": 240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019276477791027374,
+      "loss": 0.2467,
+      "step": 241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001926794576686869,
+      "loss": 0.2474,
+      "step": 242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001925936564166529,
+      "loss": 0.2465,
+      "step": 243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019250737459948405,
+      "loss": 0.241,
+      "step": 244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019242061266498675,
+      "loss": 0.2478,
+      "step": 245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019233337106345925,
+      "loss": 0.2395,
+      "step": 246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019224565024768926,
+      "loss": 0.2371,
+      "step": 247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019215745067295169,
+      "loss": 0.2484,
+      "step": 248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019206877279700612,
+      "loss": 0.2417,
+      "step": 249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019197961708009473,
+      "loss": 0.231,
+      "step": 250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019188998398493953,
+      "loss": 0.2405,
+      "step": 251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019179987397674022,
+      "loss": 0.242,
+      "step": 252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001917092875231717,
+      "loss": 0.2422,
+      "step": 253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019161822509438162,
+      "loss": 0.2435,
+      "step": 254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000191526687162988,
+      "loss": 0.2374,
+      "step": 255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001914346742040767,
+      "loss": 0.2419,
+      "step": 256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019134218669519896,
+      "loss": 0.2448,
+      "step": 257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019124922511636912,
+      "loss": 0.2363,
+      "step": 258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019115578995006173,
+      "loss": 0.2348,
+      "step": 259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019106188168120948,
+      "loss": 0.2426,
+      "step": 260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019096750079720037,
+      "loss": 0.2466,
+      "step": 261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019087264778787534,
+      "loss": 0.2395,
+      "step": 262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019077732314552566,
+      "loss": 0.2374,
+      "step": 263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019068152736489036,
+      "loss": 0.2342,
+      "step": 264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019058526094315378,
+      "loss": 0.237,
+      "step": 265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001904885243799429,
+      "loss": 0.2383,
+      "step": 266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019039131817732462,
+      "loss": 0.2366,
+      "step": 267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001902936428398035,
+      "loss": 0.2435,
+      "step": 268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019019549887431877,
+      "loss": 0.2434,
+      "step": 269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001900968867902419,
+      "loss": 0.2337,
+      "step": 270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018999780709937398,
+      "loss": 0.2427,
+      "step": 271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001898982603159429,
+      "loss": 0.2432,
+      "step": 272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018979824695660087,
+      "loss": 0.2446,
+      "step": 273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018969776754042156,
+      "loss": 0.2327,
+      "step": 274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001895968225888976,
+      "loss": 0.2369,
+      "step": 275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018949541262593762,
+      "loss": 0.2376,
+      "step": 276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018939353817786387,
+      "loss": 0.2286,
+      "step": 277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018929119977340917,
+      "loss": 0.2304,
+      "step": 278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001891883979437143,
+      "loss": 0.2369,
+      "step": 279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018908513322232528,
+      "loss": 0.244,
+      "step": 280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018898140614519054,
+      "loss": 0.2336,
+      "step": 281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018887721725065814,
+      "loss": 0.239,
+      "step": 282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018877256707947306,
+      "loss": 0.2373,
+      "step": 283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018866745617477423,
+      "loss": 0.2373,
+      "step": 284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018856188508209183,
+      "loss": 0.2363,
+      "step": 285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018845585434934452,
+      "loss": 0.2302,
+      "step": 286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018834936452683638,
+      "loss": 0.2376,
+      "step": 287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018824241616725434,
+      "loss": 0.2335,
+      "step": 288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000188135009825665,
+      "loss": 0.2338,
+      "step": 289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018802714605951199,
+      "loss": 0.24,
+      "step": 290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018791882542861302,
+      "loss": 0.2311,
+      "step": 291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001878100484951569,
+      "loss": 0.2396,
+      "step": 292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018770081582370068,
+      "loss": 0.2399,
+      "step": 293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001875911279811667,
+      "loss": 0.2289,
+      "step": 294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018748098553683968,
+      "loss": 0.2337,
+      "step": 295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001873703890623637,
+      "loss": 0.2329,
+      "step": 296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018725933913173938,
+      "loss": 0.2349,
+      "step": 297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018714783632132068,
+      "loss": 0.2287,
+      "step": 298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018703588120981207,
+      "loss": 0.2297,
+      "step": 299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018692347437826548,
+      "loss": 0.2325,
+      "step": 300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018681061641007737,
+      "loss": 0.2305,
+      "step": 301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001866973078909854,
+      "loss": 0.2362,
+      "step": 302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018658354940906586,
+      "loss": 0.2274,
+      "step": 303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018646934155473022,
+      "loss": 0.2359,
+      "step": 304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018635468492072228,
+      "loss": 0.2317,
+      "step": 305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018623958010211493,
+      "loss": 0.2353,
+      "step": 306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001861240276963073,
+      "loss": 0.2239,
+      "step": 307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018600802830302134,
+      "loss": 0.2303,
+      "step": 308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001858915825242991,
+      "loss": 0.2236,
+      "step": 309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018577469096449925,
+      "loss": 0.2359,
+      "step": 310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018565735423029404,
+      "loss": 0.2309,
+      "step": 311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018553957293066632,
+      "loss": 0.2349,
+      "step": 312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018542134767690616,
+      "loss": 0.2293,
+      "step": 313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018530267908260784,
+      "loss": 0.2323,
+      "step": 314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018518356776366657,
+      "loss": 0.2266,
+      "step": 315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018506401433827528,
+      "loss": 0.2331,
+      "step": 316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018494401942692153,
+      "loss": 0.229,
+      "step": 317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018482358365238413,
+      "loss": 0.2221,
+      "step": 318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018470270763973004,
+      "loss": 0.2229,
+      "step": 319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018458139201631108,
+      "loss": 0.2292,
+      "step": 320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018445963741176065,
+      "loss": 0.2254,
+      "step": 321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018433744445799045,
+      "loss": 0.2258,
+      "step": 322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001842148137891873,
+      "loss": 0.2306,
+      "step": 323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018409174604180976,
+      "loss": 0.2307,
+      "step": 324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001839682418545848,
+      "loss": 0.226,
+      "step": 325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018384430186850454,
+      "loss": 0.2365,
+      "step": 326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000183719926726823,
+      "loss": 0.2244,
+      "step": 327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018359511707505258,
+      "loss": 0.2305,
+      "step": 328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018346987356096086,
+      "loss": 0.2307,
+      "step": 329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018334419683456717,
+      "loss": 0.2241,
+      "step": 330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001832180875481392,
+      "loss": 0.222,
+      "step": 331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018309154635618965,
+      "loss": 0.2265,
+      "step": 332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018296457391547296,
+      "loss": 0.2302,
+      "step": 333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018283717088498155,
+      "loss": 0.2245,
+      "step": 334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001827093379259428,
+      "loss": 0.2204,
+      "step": 335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018258107570181533,
+      "loss": 0.2264,
+      "step": 336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018245238487828573,
+      "loss": 0.2243,
+      "step": 337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000182323266123265,
+      "loss": 0.2294,
+      "step": 338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018219372010688515,
+      "loss": 0.2228,
+      "step": 339
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018206374750149567,
+      "loss": 0.2234,
+      "step": 340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018193334898166007,
+      "loss": 0.22,
+      "step": 341
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018180252522415242,
+      "loss": 0.2198,
+      "step": 342
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018167127690795368,
+      "loss": 0.2311,
+      "step": 343
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001815396047142485,
+      "loss": 0.2246,
+      "step": 344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001814075093264212,
+      "loss": 0.2313,
+      "step": 345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018127499143005268,
+      "loss": 0.2158,
+      "step": 346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018114205171291663,
+      "loss": 0.2341,
+      "step": 347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000181008690864976,
+      "loss": 0.225,
+      "step": 348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018087490957837944,
+      "loss": 0.2264,
+      "step": 349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018074070854745772,
+      "loss": 0.2259,
+      "step": 350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018060608846872005,
+      "loss": 0.2319,
+      "step": 351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018047105004085053,
+      "loss": 0.2299,
+      "step": 352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018033559396470454,
+      "loss": 0.2185,
+      "step": 353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018019972094330503,
+      "loss": 0.2209,
+      "step": 354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018006343168183893,
+      "loss": 0.2178,
+      "step": 355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001799267268876535,
+      "loss": 0.2144,
+      "step": 356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001797896072702526,
+      "loss": 0.2143,
+      "step": 357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017965207354129307,
+      "loss": 0.2178,
+      "step": 358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00017951412641458098,
+      "loss": 0.223,
+      "step": 359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000179375766606068,
+      "loss": 0.2273,
+      "step": 360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017923699483384753,
+      "loss": 0.2213,
+      "step": 361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017909781181815117,
+      "loss": 0.2194,
+      "step": 362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001789582182813449,
+      "loss": 0.2126,
+      "step": 363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017881821494792528,
+      "loss": 0.2166,
+      "step": 364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017867780254451576,
+      "loss": 0.2169,
+      "step": 365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017853698179986282,
+      "loss": 0.2192,
+      "step": 366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017839575344483238,
+      "loss": 0.229,
+      "step": 367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001782541182124057,
+      "loss": 0.219,
+      "step": 368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001781120768376759,
+      "loss": 0.2202,
+      "step": 369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017796963005784394,
+      "loss": 0.2156,
+      "step": 370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001778267786122148,
+      "loss": 0.2216,
+      "step": 371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001776835232421938,
+      "loss": 0.2122,
+      "step": 372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017753986469128257,
+      "loss": 0.2185,
+      "step": 373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017739580370507532,
+      "loss": 0.2228,
+      "step": 374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001772513410312548,
+      "loss": 0.2271,
+      "step": 375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017710647741958868,
+      "loss": 0.2102,
+      "step": 376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017696121362192544,
+      "loss": 0.2106,
+      "step": 377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017681555039219054,
+      "loss": 0.224,
+      "step": 378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017666948848638257,
+      "loss": 0.2224,
+      "step": 379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017652302866256916,
+      "loss": 0.2206,
+      "step": 380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017637617168088325,
+      "loss": 0.221,
+      "step": 381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000176228918303519,
+      "loss": 0.2221,
+      "step": 382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017608126929472795,
+      "loss": 0.2201,
+      "step": 383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017593322542081485,
+      "loss": 0.2207,
+      "step": 384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017578478745013392,
+      "loss": 0.2203,
+      "step": 385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017563595615308474,
+      "loss": 0.2185,
+      "step": 386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017548673230210823,
+      "loss": 0.2165,
+      "step": 387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001753371166716828,
+      "loss": 0.22,
+      "step": 388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017518711003832002,
+      "loss": 0.2183,
+      "step": 389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000175036713180561,
+      "loss": 0.2172,
+      "step": 390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017488592687897193,
+      "loss": 0.2158,
+      "step": 391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017473475191614037,
+      "loss": 0.2182,
+      "step": 392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017458318907667098,
+      "loss": 0.2178,
+      "step": 393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001744312391471816,
+      "loss": 0.2121,
+      "step": 394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017427890291629893,
+      "loss": 0.219,
+      "step": 395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017412618117465477,
+      "loss": 0.2223,
+      "step": 396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001739730747148816,
+      "loss": 0.2072,
+      "step": 397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017381958433160865,
+      "loss": 0.2116,
+      "step": 398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001736657108214578,
+      "loss": 0.218,
+      "step": 399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017351145498303925,
+      "loss": 0.2135,
+      "step": 400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001733568176169476,
+      "loss": 0.2258,
+      "step": 401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001732017995257575,
+      "loss": 0.2171,
+      "step": 402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017304640151401967,
+      "loss": 0.2191,
+      "step": 403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017289062438825665,
+      "loss": 0.2213,
+      "step": 404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001727344689569585,
+      "loss": 0.2187,
+      "step": 405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017257793603057871,
+      "loss": 0.2161,
+      "step": 406
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017242102642153016,
+      "loss": 0.2086,
+      "step": 407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017226374094418044,
+      "loss": 0.2179,
+      "step": 408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001721060804148482,
+      "loss": 0.2133,
+      "step": 409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017194804565179842,
+      "loss": 0.2181,
+      "step": 410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017178963747523847,
+      "loss": 0.2096,
+      "step": 411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017163085670731371,
+      "loss": 0.2106,
+      "step": 412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017147170417210333,
+      "loss": 0.2104,
+      "step": 413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017131218069561593,
+      "loss": 0.2118,
+      "step": 414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017115228710578534,
+      "loss": 0.2154,
+      "step": 415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001709920242324663,
+      "loss": 0.2168,
+      "step": 416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001708313929074302,
+      "loss": 0.2146,
+      "step": 417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017067039396436058,
+      "loss": 0.2104,
+      "step": 418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017050902823884903,
+      "loss": 0.2151,
+      "step": 419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017034729656839078,
+      "loss": 0.2172,
+      "step": 420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017018519979238023,
+      "loss": 0.2161,
+      "step": 421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001700227387521068,
+      "loss": 0.2247,
+      "step": 422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016985991429075036,
+      "loss": 0.2132,
+      "step": 423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016969672725337706,
+      "loss": 0.2185,
+      "step": 424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016953317848693474,
+      "loss": 0.2201,
+      "step": 425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016936926884024864,
+      "loss": 0.2191,
+      "step": 426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016920499916401707,
+      "loss": 0.2153,
+      "step": 427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001690403703108068,
+      "loss": 0.2109,
+      "step": 428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016887538313504883,
+      "loss": 0.2104,
+      "step": 429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016871003849303382,
+      "loss": 0.2087,
+      "step": 430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001685443372429077,
+      "loss": 0.2122,
+      "step": 431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001683782802446672,
+      "loss": 0.2141,
+      "step": 432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001682118683601555,
+      "loss": 0.2258,
+      "step": 433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016804510245305745,
+      "loss": 0.2038,
+      "step": 434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016787798338889552,
+      "loss": 0.2114,
+      "step": 435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016771051203502493,
+      "loss": 0.2145,
+      "step": 436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016754268926062938,
+      "loss": 0.2168,
+      "step": 437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016737451593671636,
+      "loss": 0.2113,
+      "step": 438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00016720599293611286,
+      "loss": 0.2227,
+      "step": 439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001670371211334606,
+      "loss": 0.2083,
+      "step": 440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016686790140521164,
+      "loss": 0.2179,
+      "step": 441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001666983346296238,
+      "loss": 0.2079,
+      "step": 442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001665284216867561,
+      "loss": 0.2033,
+      "step": 443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001663581634584641,
+      "loss": 0.2051,
+      "step": 444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00016618756082839554,
+      "loss": 0.2155,
+      "step": 445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001660166146819855,
+      "loss": 0.217,
+      "step": 446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001658453259064519,
+      "loss": 0.2056,
+      "step": 447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016567369539079114,
+      "loss": 0.2103,
+      "step": 448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000165501724025773,
+      "loss": 0.2088,
+      "step": 449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001653294127039365,
+      "loss": 0.2202,
+      "step": 450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001651567623195849,
+      "loss": 0.2069,
+      "step": 451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00016498377376878126,
+      "loss": 0.2068,
+      "step": 452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001648104479493437,
+      "loss": 0.2106,
+      "step": 453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001646367857608409,
+      "loss": 0.2083,
+      "step": 454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016446278810458716,
+      "loss": 0.2062,
+      "step": 455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000164288455883638,
+      "loss": 0.2122,
+      "step": 456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016411379000278524,
+      "loss": 0.2124,
+      "step": 457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016393879136855248,
+      "loss": 0.2132,
+      "step": 458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001637634608891903,
+      "loss": 0.2113,
+      "step": 459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016358779947467158,
+      "loss": 0.2122,
+      "step": 460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00016341180803668674,
+      "loss": 0.204,
+      "step": 461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016323548748863907,
+      "loss": 0.2105,
+      "step": 462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016305883874563994,
+      "loss": 0.2064,
+      "step": 463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001628818627245041,
+      "loss": 0.2069,
+      "step": 464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016270456034374474,
+      "loss": 0.2033,
+      "step": 465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016252693252356916,
+      "loss": 0.2038,
+      "step": 466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016234898018587337,
+      "loss": 0.2063,
+      "step": 467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00016217070425423788,
+      "loss": 0.2057,
+      "step": 468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001619921056539226,
+      "loss": 0.2088,
+      "step": 469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016181318531186206,
+      "loss": 0.2072,
+      "step": 470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001616339441566607,
+      "loss": 0.2049,
+      "step": 471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000161454383118588,
+      "loss": 0.2056,
+      "step": 472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016127450312957353,
+      "loss": 0.2113,
+      "step": 473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016109430512320237,
+      "loss": 0.2056,
+      "step": 474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00016091379003471007,
+      "loss": 0.2093,
+      "step": 475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016073295880097784,
+      "loss": 0.2047,
+      "step": 476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001605518123605277,
+      "loss": 0.2115,
+      "step": 477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016037035165351768,
+      "loss": 0.2111,
+      "step": 478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001601885776217367,
+      "loss": 0.2113,
+      "step": 479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00016000649120860003,
+      "loss": 0.2071,
+      "step": 480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015982409335914407,
+      "loss": 0.2106,
+      "step": 481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00015964138502002175,
+      "loss": 0.2118,
+      "step": 482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015945836713949726,
+      "loss": 0.2077,
+      "step": 483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015927504066744148,
+      "loss": 0.2,
+      "step": 484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001590914065553268,
+      "loss": 0.2084,
+      "step": 485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015890746575622231,
+      "loss": 0.2071,
+      "step": 486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015872321922478884,
+      "loss": 0.2072,
+      "step": 487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015853866791727396,
+      "loss": 0.2059,
+      "step": 488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015835381279150705,
+      "loss": 0.2092,
+      "step": 489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00015816865480689426,
+      "loss": 0.2093,
+      "step": 490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001579831949244137,
+      "loss": 0.2155,
+      "step": 491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015779743410661033,
+      "loss": 0.2019,
+      "step": 492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015761137331759084,
+      "loss": 0.2012,
+      "step": 493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015742501352301893,
+      "loss": 0.2073,
+      "step": 494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015723835569011007,
+      "loss": 0.2046,
+      "step": 495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015705140078762665,
+      "loss": 0.2131,
+      "step": 496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00015686414978587277,
+      "loss": 0.2071,
+      "step": 497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001566766036566893,
+      "loss": 0.2097,
+      "step": 498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015648876337344896,
+      "loss": 0.2041,
+      "step": 499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015630062991105098,
+      "loss": 0.2006,
+      "step": 500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001561122042459163,
+      "loss": 0.2113,
+      "step": 501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015592348735598237,
+      "loss": 0.2008,
+      "step": 502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015573448022069815,
+      "loss": 0.1983,
+      "step": 503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00015554518382101892,
+      "loss": 0.2065,
+      "step": 504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015535559913940126,
+      "loss": 0.2074,
+      "step": 505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015516572715979806,
+      "loss": 0.2103,
+      "step": 506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015497556886765316,
+      "loss": 0.2082,
+      "step": 507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015478512524989645,
+      "loss": 0.201,
+      "step": 508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015459439729493865,
+      "loss": 0.2075,
+      "step": 509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015440338599266622,
+      "loss": 0.2013,
+      "step": 510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015421209233443617,
+      "loss": 0.1999,
+      "step": 511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00015402051731307093,
+      "loss": 0.1946,
+      "step": 512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001538286619228533,
+      "loss": 0.2062,
+      "step": 513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001536365271595212,
+      "loss": 0.2057,
+      "step": 514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015344411402026245,
+      "loss": 0.2043,
+      "step": 515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015325142350370967,
+      "loss": 0.2,
+      "step": 516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015305845660993503,
+      "loss": 0.2035,
+      "step": 517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00015286521434044526,
+      "loss": 0.2049,
+      "step": 518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001526716976981761,
+      "loss": 0.1993,
+      "step": 519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001524779076874875,
+      "loss": 0.2032,
+      "step": 520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001522838453141581,
+      "loss": 0.2039,
+      "step": 521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015208951158538004,
+      "loss": 0.2059,
+      "step": 522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000151894907509754,
+      "loss": 0.2051,
+      "step": 523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015170003409728356,
+      "loss": 0.2069,
+      "step": 524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015150489235937035,
+      "loss": 0.2007,
+      "step": 525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00015130948330880847,
+      "loss": 0.1982,
+      "step": 526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015111380795977954,
+      "loss": 0.2002,
+      "step": 527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015091786732784716,
+      "loss": 0.1968,
+      "step": 528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015072166242995175,
+      "loss": 0.2004,
+      "step": 529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001505251942844054,
+      "loss": 0.194,
+      "step": 530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00015032846391088635,
+      "loss": 0.2025,
+      "step": 531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001501314723304339,
+      "loss": 0.2056,
+      "step": 532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00014993422056544295,
+      "loss": 0.2025,
+      "step": 533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014973670963965883,
+      "loss": 0.1995,
+      "step": 534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014953894057817188,
+      "loss": 0.2061,
+      "step": 535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001493409144074122,
+      "loss": 0.2047,
+      "step": 536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014914263215514431,
+      "loss": 0.2044,
+      "step": 537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014894409485046177,
+      "loss": 0.1996,
+      "step": 538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014874530352378194,
+      "loss": 0.2001,
+      "step": 539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014854625920684042,
+      "loss": 0.197,
+      "step": 540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00014834696293268603,
+      "loss": 0.2,
+      "step": 541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014814741573567514,
+      "loss": 0.2039,
+      "step": 542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014794761865146648,
+      "loss": 0.2,
+      "step": 543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014774757271701557,
+      "loss": 0.1982,
+      "step": 544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014754727897056967,
+      "loss": 0.1982,
+      "step": 545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001473467384516621,
+      "loss": 0.1939,
+      "step": 546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001471459522011069,
+      "loss": 0.201,
+      "step": 547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014694492126099353,
+      "loss": 0.2011,
+      "step": 548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001467436466746814,
+      "loss": 0.1918,
+      "step": 549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001465421294867944,
+      "loss": 0.2015,
+      "step": 550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014634037074321557,
+      "loss": 0.1982,
+      "step": 551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014613837149108163,
+      "loss": 0.2013,
+      "step": 552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014593613277877758,
+      "loss": 0.2078,
+      "step": 553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001457336556559312,
+      "loss": 0.1964,
+      "step": 554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001455309411734076,
+      "loss": 0.1985,
+      "step": 555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014532799038330385,
+      "loss": 0.1974,
+      "step": 556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014512480433894343,
+      "loss": 0.1954,
+      "step": 557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014492138409487085,
+      "loss": 0.2014,
+      "step": 558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014471773070684599,
+      "loss": 0.1983,
+      "step": 559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00014451384523183903,
+      "loss": 0.2,
+      "step": 560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001443097287280244,
+      "loss": 0.1913,
+      "step": 561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001441053822547757,
+      "loss": 0.1879,
+      "step": 562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014390080687266013,
+      "loss": 0.1963,
+      "step": 563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014369600364343285,
+      "loss": 0.2001,
+      "step": 564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014349097363003163,
+      "loss": 0.2045,
+      "step": 565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001432857178965712,
+      "loss": 0.1973,
+      "step": 566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014308023750833783,
+      "loss": 0.1956,
+      "step": 567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014287453353178372,
+      "loss": 0.2013,
+      "step": 568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014266860703452156,
+      "loss": 0.1972,
+      "step": 569
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00014246245908531882,
+      "loss": 0.1986,
+      "step": 570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001422560907540925,
+      "loss": 0.195,
+      "step": 571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014204950311190318,
+      "loss": 0.1969,
+      "step": 572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014184269723094988,
+      "loss": 0.1956,
+      "step": 573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014163567418456406,
+      "loss": 0.1981,
+      "step": 574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001414284350472045,
+      "loss": 0.199,
+      "step": 575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00014122098089445142,
+      "loss": 0.1963,
+      "step": 576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001410133128030009,
+      "loss": 0.1974,
+      "step": 577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014080543185065943,
+      "loss": 0.205,
+      "step": 578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001405973391163383,
+      "loss": 0.193,
+      "step": 579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001403890356800479,
+      "loss": 0.2024,
+      "step": 580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00014018052262289223,
+      "loss": 0.2003,
+      "step": 581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001399718010270632,
+      "loss": 0.1894,
+      "step": 582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00013976287197583494,
+      "loss": 0.1907,
+      "step": 583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001395537365535585,
+      "loss": 0.1917,
+      "step": 584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013934439584565583,
+      "loss": 0.2007,
+      "step": 585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001391348509386144,
+      "loss": 0.1929,
+      "step": 586
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013892510291998146,
+      "loss": 0.2036,
+      "step": 587
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013871515287835839,
+      "loss": 0.1974,
+      "step": 588
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013850500190339514,
+      "loss": 0.1994,
+      "step": 589
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00013829465108578445,
+      "loss": 0.201,
+      "step": 590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001380841015172563,
+      "loss": 0.2054,
+      "step": 591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001378733542905722,
+      "loss": 0.1986,
+      "step": 592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013766241049951948,
+      "loss": 0.197,
+      "step": 593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013745127123890565,
+      "loss": 0.1922,
+      "step": 594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013723993760455272,
+      "loss": 0.2014,
+      "step": 595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001370284106932915,
+      "loss": 0.1968,
+      "step": 596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013681669160295597,
+      "loss": 0.1924,
+      "step": 597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013660478143237746,
+      "loss": 0.1998,
+      "step": 598
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00013639268128137907,
+      "loss": 0.1959,
+      "step": 599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013618039225076986,
+      "loss": 0.1993,
+      "step": 600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001359679154423392,
+      "loss": 0.1905,
+      "step": 601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013575525195885107,
+      "loss": 0.1888,
+      "step": 602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001355424029040382,
+      "loss": 0.1941,
+      "step": 603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013532936938259656,
+      "loss": 0.1951,
+      "step": 604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001351161525001795,
+      "loss": 0.2033,
+      "step": 605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00013490275336339188,
+      "loss": 0.1895,
+      "step": 606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013468917307978467,
+      "loss": 0.1881,
+      "step": 607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013447541275784887,
+      "loss": 0.192,
+      "step": 608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013426147350700996,
+      "loss": 0.1918,
+      "step": 609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00013404735643762192,
+      "loss": 0.1959,
+      "step": 610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001338330626609618,
+      "loss": 0.1962,
+      "step": 611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001336185932892237,
+      "loss": 0.1919,
+      "step": 612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000133403949435513,
+      "loss": 0.1977,
+      "step": 613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00013318913221384076,
+      "loss": 0.1978,
+      "step": 614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00013297414273911784,
+      "loss": 0.1904,
+      "step": 615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00013275898212714889,
+      "loss": 0.1923,
+      "step": 616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00013254365149462699,
+      "loss": 0.1944,
+      "step": 617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00013232815195912754,
+      "loss": 0.1929,
+      "step": 618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00013211248463910262,
+      "loss": 0.2027,
+      "step": 619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00013189665065387507,
+      "loss": 0.187,
+      "step": 620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00013168065112363264,
+      "loss": 0.1936,
+      "step": 621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00013146448716942245,
+      "loss": 0.1954,
+      "step": 622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001312481599131449,
+      "loss": 0.1875,
+      "step": 623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00013103167047754784,
+      "loss": 0.1844,
+      "step": 624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000130815019986221,
+      "loss": 0.1918,
+      "step": 625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00013059820956358998,
+      "loss": 0.1912,
+      "step": 626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00013038124033491025,
+      "loss": 0.1982,
+      "step": 627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00013016411342626168,
+      "loss": 0.1899,
+      "step": 628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012994682996454247,
+      "loss": 0.1944,
+      "step": 629
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012972939107746325,
+      "loss": 0.1883,
+      "step": 630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001295117978935414,
+      "loss": 0.1917,
+      "step": 631
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001292940515420951,
+      "loss": 0.1952,
+      "step": 632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001290761531532374,
+      "loss": 0.1964,
+      "step": 633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012885810385787055,
+      "loss": 0.1875,
+      "step": 634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00012863990478767994,
+      "loss": 0.1897,
+      "step": 635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012842155707512825,
+      "loss": 0.1872,
+      "step": 636
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012820306185344976,
+      "loss": 0.194,
+      "step": 637
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001279844202566442,
+      "loss": 0.1887,
+      "step": 638
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012776563341947104,
+      "loss": 0.1916,
+      "step": 639
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012754670247744354,
+      "loss": 0.2052,
+      "step": 640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001273276285668229,
+      "loss": 0.1934,
+      "step": 641
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00012710841282461238,
+      "loss": 0.1949,
+      "step": 642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001268890563885512,
+      "loss": 0.1805,
+      "step": 643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001266695603971089,
+      "loss": 0.1903,
+      "step": 644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001264499259894793,
+      "loss": 0.1926,
+      "step": 645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001262301543055746,
+      "loss": 0.1958,
+      "step": 646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001260102464860195,
+      "loss": 0.1908,
+      "step": 647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001257902036721452,
+      "loss": 0.1885,
+      "step": 648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00012557002700598353,
+      "loss": 0.1878,
+      "step": 649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012534971763026104,
+      "loss": 0.1888,
+      "step": 650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012512927668839304,
+      "loss": 0.1902,
+      "step": 651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012490870532447774,
+      "loss": 0.1855,
+      "step": 652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012468800468329013,
+      "loss": 0.1892,
+      "step": 653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012446717591027624,
+      "loss": 0.1916,
+      "step": 654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012424622015154703,
+      "loss": 0.1905,
+      "step": 655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001240251385538726,
+      "loss": 0.1874,
+      "step": 656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00012380393226467615,
+      "loss": 0.197,
+      "step": 657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000123582602432028,
+      "loss": 0.1972,
+      "step": 658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001233611502046397,
+      "loss": 0.19,
+      "step": 659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000123139576731858,
+      "loss": 0.1912,
+      "step": 660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012291788316365888,
+      "loss": 0.1895,
+      "step": 661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012269607065064177,
+      "loss": 0.188,
+      "step": 662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001224741403440233,
+      "loss": 0.1932,
+      "step": 663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00012225209339563145,
+      "loss": 0.1899,
+      "step": 664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00012202993095789966,
+      "loss": 0.1791,
+      "step": 665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00012180765418386068,
+      "loss": 0.1943,
+      "step": 666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00012158526422714076,
+      "loss": 0.1953,
+      "step": 667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00012136276224195348,
+      "loss": 0.1885,
+      "step": 668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00012114014938309393,
+      "loss": 0.1847,
+      "step": 669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00012091742680593254,
+      "loss": 0.19,
+      "step": 670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001206945956664093,
+      "loss": 0.1854,
+      "step": 671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00012047165712102759,
+      "loss": 0.1892,
+      "step": 672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00012024861232684823,
+      "loss": 0.1852,
+      "step": 673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00012002546244148345,
+      "loss": 0.1919,
+      "step": 674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011980220862309097,
+      "loss": 0.1908,
+      "step": 675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011957885203036785,
+      "loss": 0.1891,
+      "step": 676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011935539382254459,
+      "loss": 0.1877,
+      "step": 677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011913183515937916,
+      "loss": 0.1868,
+      "step": 678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00011890817720115075,
+      "loss": 0.1912,
+      "step": 679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011868442110865399,
+      "loss": 0.195,
+      "step": 680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001184605680431928,
+      "loss": 0.1887,
+      "step": 681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001182366191665744,
+      "loss": 0.1917,
+      "step": 682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011801257564110329,
+      "loss": 0.18,
+      "step": 683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011778843862957514,
+      "loss": 0.1857,
+      "step": 684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001175642092952709,
+      "loss": 0.1865,
+      "step": 685
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00011733988880195068,
+      "loss": 0.1868,
+      "step": 686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011711547831384761,
+      "loss": 0.1844,
+      "step": 687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011689097899566198,
+      "loss": 0.1932,
+      "step": 688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011666639201255506,
+      "loss": 0.1868,
+      "step": 689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011644171853014319,
+      "loss": 0.1849,
+      "step": 690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011621695971449154,
+      "loss": 0.1905,
+      "step": 691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011599211673210826,
+      "loss": 0.1866,
+      "step": 692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00011576719074993827,
+      "loss": 0.1852,
+      "step": 693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011554218293535725,
+      "loss": 0.1926,
+      "step": 694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011531709445616563,
+      "loss": 0.1933,
+      "step": 695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011509192648058249,
+      "loss": 0.1931,
+      "step": 696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011486668017723949,
+      "loss": 0.1842,
+      "step": 697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011464135671517482,
+      "loss": 0.1863,
+      "step": 698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001144159572638271,
+      "loss": 0.1864,
+      "step": 699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00011419048299302939,
+      "loss": 0.189,
+      "step": 700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011396493507300303,
+      "loss": 0.1823,
+      "step": 701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011373931467435159,
+      "loss": 0.1899,
+      "step": 702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011351362296805485,
+      "loss": 0.1808,
+      "step": 703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011328786112546269,
+      "loss": 0.1857,
+      "step": 704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011306203031828886,
+      "loss": 0.1834,
+      "step": 705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011283613171860525,
+      "loss": 0.1909,
+      "step": 706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00011261016649883545,
+      "loss": 0.1844,
+      "step": 707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001123841358317489,
+      "loss": 0.185,
+      "step": 708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00011215804089045459,
+      "loss": 0.1884,
+      "step": 709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00011193188284839517,
+      "loss": 0.1911,
+      "step": 710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00011170566287934088,
+      "loss": 0.1893,
+      "step": 711
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00011147938215738323,
+      "loss": 0.1838,
+      "step": 712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00011125304185692907,
+      "loss": 0.1922,
+      "step": 713
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001110266431526945,
+      "loss": 0.1884,
+      "step": 714
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00011080018721969871,
+      "loss": 0.1876,
+      "step": 715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00011057367523325792,
+      "loss": 0.1844,
+      "step": 716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00011034710836897921,
+      "loss": 0.1861,
+      "step": 717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00011012048780275463,
+      "loss": 0.1847,
+      "step": 718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010989381471075482,
+      "loss": 0.1812,
+      "step": 719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010966709026942303,
+      "loss": 0.1719,
+      "step": 720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010944031565546906,
+      "loss": 0.1913,
+      "step": 721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001092134920458631,
+      "loss": 0.1893,
+      "step": 722
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010898662061782965,
+      "loss": 0.1815,
+      "step": 723
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001087597025488413,
+      "loss": 0.1943,
+      "step": 724
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010853273901661285,
+      "loss": 0.1912,
+      "step": 725
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010830573119909493,
+      "loss": 0.1882,
+      "step": 726
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010807868027446808,
+      "loss": 0.1842,
+      "step": 727
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010785158742113655,
+      "loss": 0.1908,
+      "step": 728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010762445381772217,
+      "loss": 0.1908,
+      "step": 729
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010739728064305834,
+      "loss": 0.1923,
+      "step": 730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010717006907618377,
+      "loss": 0.1808,
+      "step": 731
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010694282029633647,
+      "loss": 0.1855,
+      "step": 732
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010671553548294753,
+      "loss": 0.1832,
+      "step": 733
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010648821581563513,
+      "loss": 0.1828,
+      "step": 734
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010626086247419826,
+      "loss": 0.1849,
+      "step": 735
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010603347663861079,
+      "loss": 0.1843,
+      "step": 736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010580605948901514,
+      "loss": 0.1776,
+      "step": 737
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010557861220571625,
+      "loss": 0.1822,
+      "step": 738
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010535113596917556,
+      "loss": 0.1764,
+      "step": 739
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010512363196000465,
+      "loss": 0.1782,
+      "step": 740
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010489610135895933,
+      "loss": 0.1811,
+      "step": 741
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010466854534693335,
+      "loss": 0.1809,
+      "step": 742
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010444096510495243,
+      "loss": 0.1856,
+      "step": 743
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010421336181416797,
+      "loss": 0.1905,
+      "step": 744
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010398573665585105,
+      "loss": 0.1843,
+      "step": 745
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001037580908113862,
+      "loss": 0.1847,
+      "step": 746
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010353042546226537,
+      "loss": 0.1772,
+      "step": 747
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010330274179008161,
+      "loss": 0.1789,
+      "step": 748
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010307504097652323,
+      "loss": 0.1816,
+      "step": 749
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001028473242033674,
+      "loss": 0.1806,
+      "step": 750
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010261959265247419,
+      "loss": 0.179,
+      "step": 751
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001023918475057803,
+      "loss": 0.1787,
+      "step": 752
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010216408994529303,
+      "loss": 0.1842,
+      "step": 753
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010193632115308411,
+      "loss": 0.1839,
+      "step": 754
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010170854231128352,
+      "loss": 0.186,
+      "step": 755
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010148075460207347,
+      "loss": 0.1845,
+      "step": 756
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001012529592076821,
+      "loss": 0.1787,
+      "step": 757
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010102515731037758,
+      "loss": 0.1831,
+      "step": 758
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010079735009246167,
+      "loss": 0.1844,
+      "step": 759
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010056953873626384,
+      "loss": 0.1811,
+      "step": 760
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010034172442413501,
+      "loss": 0.1827,
+      "step": 761
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010011390833844143,
+      "loss": 0.1795,
+      "step": 762
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.988609166155859e-05,
+      "loss": 0.1814,
+      "step": 763
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.965827557586498e-05,
+      "loss": 0.1856,
+      "step": 764
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.943046126373618e-05,
+      "loss": 0.1849,
+      "step": 765
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.920264990753837e-05,
+      "loss": 0.183,
+      "step": 766
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.897484268962243e-05,
+      "loss": 0.1766,
+      "step": 767
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.874704079231791e-05,
+      "loss": 0.1752,
+      "step": 768
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.851924539792656e-05,
+      "loss": 0.1797,
+      "step": 769
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.82914576887165e-05,
+      "loss": 0.1772,
+      "step": 770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.806367884691594e-05,
+      "loss": 0.1861,
+      "step": 771
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.783591005470698e-05,
+      "loss": 0.1751,
+      "step": 772
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.760815249421973e-05,
+      "loss": 0.1825,
+      "step": 773
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.738040734752582e-05,
+      "loss": 0.172,
+      "step": 774
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.715267579663262e-05,
+      "loss": 0.1778,
+      "step": 775
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.692495902347678e-05,
+      "loss": 0.1787,
+      "step": 776
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.669725820991841e-05,
+      "loss": 0.1781,
+      "step": 777
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.646957453773469e-05,
+      "loss": 0.1853,
+      "step": 778
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.62419091886138e-05,
+      "loss": 0.1813,
+      "step": 779
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.601426334414898e-05,
+      "loss": 0.1767,
+      "step": 780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.578663818583203e-05,
+      "loss": 0.1883,
+      "step": 781
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.555903489504761e-05,
+      "loss": 0.1803,
+      "step": 782
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.533145465306667e-05,
+      "loss": 0.1713,
+      "step": 783
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.510389864104069e-05,
+      "loss": 0.1856,
+      "step": 784
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.487636803999538e-05,
+      "loss": 0.1849,
+      "step": 785
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.464886403082445e-05,
+      "loss": 0.1804,
+      "step": 786
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.442138779428376e-05,
+      "loss": 0.1844,
+      "step": 787
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.419394051098489e-05,
+      "loss": 0.1794,
+      "step": 788
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.396652336138922e-05,
+      "loss": 0.1744,
+      "step": 789
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.373913752580175e-05,
+      "loss": 0.1794,
+      "step": 790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.351178418436488e-05,
+      "loss": 0.1757,
+      "step": 791
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.328446451705249e-05,
+      "loss": 0.1804,
+      "step": 792
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.305717970366358e-05,
+      "loss": 0.1828,
+      "step": 793
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.282993092381625e-05,
+      "loss": 0.1732,
+      "step": 794
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.260271935694168e-05,
+      "loss": 0.1738,
+      "step": 795
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.237554618227785e-05,
+      "loss": 0.1782,
+      "step": 796
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.214841257886349e-05,
+      "loss": 0.1843,
+      "step": 797
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.192131972553191e-05,
+      "loss": 0.1773,
+      "step": 798
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.16942688009051e-05,
+      "loss": 0.1828,
+      "step": 799
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.146726098338719e-05,
+      "loss": 0.1724,
+      "step": 800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.12402974511587e-05,
+      "loss": 0.1786,
+      "step": 801
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.101337938217038e-05,
+      "loss": 0.1789,
+      "step": 802
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.078650795413692e-05,
+      "loss": 0.1788,
+      "step": 803
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.055968434453097e-05,
+      "loss": 0.1765,
+      "step": 804
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.0332909730577e-05,
+      "loss": 0.1866,
+      "step": 805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.01061852892452e-05,
+      "loss": 0.1789,
+      "step": 806
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.98795121972454e-05,
+      "loss": 0.1739,
+      "step": 807
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.965289163102078e-05,
+      "loss": 0.1726,
+      "step": 808
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.942632476674211e-05,
+      "loss": 0.1748,
+      "step": 809
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.919981278030133e-05,
+      "loss": 0.1817,
+      "step": 810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.89733568473055e-05,
+      "loss": 0.1795,
+      "step": 811
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.874695814307094e-05,
+      "loss": 0.1714,
+      "step": 812
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.852061784261678e-05,
+      "loss": 0.1768,
+      "step": 813
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.829433712065914e-05,
+      "loss": 0.1742,
+      "step": 814
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.806811715160485e-05,
+      "loss": 0.179,
+      "step": 815
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.784195910954545e-05,
+      "loss": 0.1783,
+      "step": 816
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.761586416825117e-05,
+      "loss": 0.1806,
+      "step": 817
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.738983350116454e-05,
+      "loss": 0.1773,
+      "step": 818
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.716386828139477e-05,
+      "loss": 0.1843,
+      "step": 819
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.693796968171113e-05,
+      "loss": 0.1713,
+      "step": 820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.671213887453735e-05,
+      "loss": 0.1734,
+      "step": 821
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.648637703194516e-05,
+      "loss": 0.1757,
+      "step": 822
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.62606853256484e-05,
+      "loss": 0.1711,
+      "step": 823
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.603506492699699e-05,
+      "loss": 0.1713,
+      "step": 824
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.58095170069706e-05,
+      "loss": 0.171,
+      "step": 825
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.55840427361729e-05,
+      "loss": 0.1721,
+      "step": 826
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.535864328482523e-05,
+      "loss": 0.1815,
+      "step": 827
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.513331982276053e-05,
+      "loss": 0.1711,
+      "step": 828
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.490807351941753e-05,
+      "loss": 0.1791,
+      "step": 829
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.468290554383436e-05,
+      "loss": 0.1808,
+      "step": 830
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.445781706464277e-05,
+      "loss": 0.1731,
+      "step": 831
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.423280925006178e-05,
+      "loss": 0.1788,
+      "step": 832
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.400788326789175e-05,
+      "loss": 0.1748,
+      "step": 833
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.378304028550848e-05,
+      "loss": 0.1804,
+      "step": 834
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.355828146985684e-05,
+      "loss": 0.1789,
+      "step": 835
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.333360798744496e-05,
+      "loss": 0.1781,
+      "step": 836
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.310902100433809e-05,
+      "loss": 0.1785,
+      "step": 837
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.288452168615242e-05,
+      "loss": 0.1745,
+      "step": 838
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.266011119804936e-05,
+      "loss": 0.1694,
+      "step": 839
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.243579070472909e-05,
+      "loss": 0.179,
+      "step": 840
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.221156137042489e-05,
+      "loss": 0.173,
+      "step": 841
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.198742435889674e-05,
+      "loss": 0.1753,
+      "step": 842
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.176338083342561e-05,
+      "loss": 0.1719,
+      "step": 843
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.153943195680723e-05,
+      "loss": 0.178,
+      "step": 844
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.131557889134602e-05,
+      "loss": 0.1763,
+      "step": 845
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.109182279884928e-05,
+      "loss": 0.1677,
+      "step": 846
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.086816484062085e-05,
+      "loss": 0.1858,
+      "step": 847
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.064460617745542e-05,
+      "loss": 0.176,
+      "step": 848
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.042114796963219e-05,
+      "loss": 0.1759,
+      "step": 849
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.019779137690906e-05,
+      "loss": 0.1783,
+      "step": 850
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.997453755851658e-05,
+      "loss": 0.1799,
+      "step": 851
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.975138767315178e-05,
+      "loss": 0.1772,
+      "step": 852
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.952834287897242e-05,
+      "loss": 0.1726,
+      "step": 853
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.930540433359071e-05,
+      "loss": 0.1804,
+      "step": 854
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.908257319406747e-05,
+      "loss": 0.1709,
+      "step": 855
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.88598506169061e-05,
+      "loss": 0.1746,
+      "step": 856
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.863723775804651e-05,
+      "loss": 0.1751,
+      "step": 857
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.841473577285925e-05,
+      "loss": 0.1833,
+      "step": 858
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.819234581613934e-05,
+      "loss": 0.1726,
+      "step": 859
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.797006904210035e-05,
+      "loss": 0.1757,
+      "step": 860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.774790660436858e-05,
+      "loss": 0.1757,
+      "step": 861
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.752585965597673e-05,
+      "loss": 0.1722,
+      "step": 862
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.730392934935825e-05,
+      "loss": 0.1787,
+      "step": 863
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.708211683634112e-05,
+      "loss": 0.1711,
+      "step": 864
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.686042326814205e-05,
+      "loss": 0.1742,
+      "step": 865
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.663884979536035e-05,
+      "loss": 0.1736,
+      "step": 866
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.641739756797202e-05,
+      "loss": 0.1815,
+      "step": 867
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.619606773532386e-05,
+      "loss": 0.1773,
+      "step": 868
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.59748614461274e-05,
+      "loss": 0.1657,
+      "step": 869
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.5753779848453e-05,
+      "loss": 0.1714,
+      "step": 870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.553282408972382e-05,
+      "loss": 0.1727,
+      "step": 871
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.531199531670988e-05,
+      "loss": 0.1762,
+      "step": 872
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.50912946755223e-05,
+      "loss": 0.171,
+      "step": 873
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.487072331160696e-05,
+      "loss": 0.1729,
+      "step": 874
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.465028236973897e-05,
+      "loss": 0.168,
+      "step": 875
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.442997299401652e-05,
+      "loss": 0.1708,
+      "step": 876
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.420979632785483e-05,
+      "loss": 0.1651,
+      "step": 877
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.398975351398053e-05,
+      "loss": 0.1762,
+      "step": 878
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.37698456944254e-05,
+      "loss": 0.167,
+      "step": 879
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.355007401052072e-05,
+      "loss": 0.1689,
+      "step": 880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.333043960289113e-05,
+      "loss": 0.1727,
+      "step": 881
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.311094361144881e-05,
+      "loss": 0.1781,
+      "step": 882
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.289158717538765e-05,
+      "loss": 0.1729,
+      "step": 883
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.267237143317706e-05,
+      "loss": 0.1741,
+      "step": 884
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.245329752255647e-05,
+      "loss": 0.1776,
+      "step": 885
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.223436658052898e-05,
+      "loss": 0.1713,
+      "step": 886
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.201557974335583e-05,
+      "loss": 0.1737,
+      "step": 887
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.179693814655026e-05,
+      "loss": 0.1676,
+      "step": 888
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.157844292487173e-05,
+      "loss": 0.175,
+      "step": 889
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.13600952123201e-05,
+      "loss": 0.1718,
+      "step": 890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.114189614212944e-05,
+      "loss": 0.1799,
+      "step": 891
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.092384684676262e-05,
+      "loss": 0.1701,
+      "step": 892
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.070594845790497e-05,
+      "loss": 0.1639,
+      "step": 893
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.048820210645862e-05,
+      "loss": 0.1742,
+      "step": 894
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.027060892253679e-05,
+      "loss": 0.1752,
+      "step": 895
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.005317003545754e-05,
+      "loss": 0.1735,
+      "step": 896
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.983588657373833e-05,
+      "loss": 0.173,
+      "step": 897
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.96187596650898e-05,
+      "loss": 0.1677,
+      "step": 898
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.940179043641005e-05,
+      "loss": 0.1682,
+      "step": 899
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.918498001377901e-05,
+      "loss": 0.1714,
+      "step": 900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.896832952245217e-05,
+      "loss": 0.1738,
+      "step": 901
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.875184008685514e-05,
+      "loss": 0.1675,
+      "step": 902
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.853551283057757e-05,
+      "loss": 0.1675,
+      "step": 903
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.831934887636738e-05,
+      "loss": 0.1703,
+      "step": 904
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.8103349346125e-05,
+      "loss": 0.1774,
+      "step": 905
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.788751536089739e-05,
+      "loss": 0.1677,
+      "step": 906
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.767184804087248e-05,
+      "loss": 0.1691,
+      "step": 907
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.745634850537302e-05,
+      "loss": 0.1736,
+      "step": 908
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.724101787285114e-05,
+      "loss": 0.1692,
+      "step": 909
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.702585726088222e-05,
+      "loss": 0.1709,
+      "step": 910
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.681086778615922e-05,
+      "loss": 0.1676,
+      "step": 911
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.659605056448702e-05,
+      "loss": 0.1732,
+      "step": 912
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.638140671077633e-05,
+      "loss": 0.1682,
+      "step": 913
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.616693733903823e-05,
+      "loss": 0.1722,
+      "step": 914
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.595264356237812e-05,
+      "loss": 0.1734,
+      "step": 915
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.573852649299009e-05,
+      "loss": 0.1727,
+      "step": 916
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.552458724215114e-05,
+      "loss": 0.1696,
+      "step": 917
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.531082692021532e-05,
+      "loss": 0.1753,
+      "step": 918
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.509724663660813e-05,
+      "loss": 0.1678,
+      "step": 919
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.488384749982053e-05,
+      "loss": 0.1685,
+      "step": 920
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.467063061740345e-05,
+      "loss": 0.1707,
+      "step": 921
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.445759709596182e-05,
+      "loss": 0.169,
+      "step": 922
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.424474804114895e-05,
+      "loss": 0.1732,
+      "step": 923
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.403208455766081e-05,
+      "loss": 0.1704,
+      "step": 924
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.381960774923017e-05,
+      "loss": 0.166,
+      "step": 925
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.360731871862093e-05,
+      "loss": 0.1705,
+      "step": 926
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.339521856762254e-05,
+      "loss": 0.1729,
+      "step": 927
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.318330839704405e-05,
+      "loss": 0.1702,
+      "step": 928
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.297158930670852e-05,
+      "loss": 0.1642,
+      "step": 929
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.276006239544729e-05,
+      "loss": 0.1704,
+      "step": 930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.254872876109438e-05,
+      "loss": 0.1682,
+      "step": 931
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.233758950048056e-05,
+      "loss": 0.1726,
+      "step": 932
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.21266457094278e-05,
+      "loss": 0.1704,
+      "step": 933
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.191589848274368e-05,
+      "loss": 0.1734,
+      "step": 934
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.170534891421556e-05,
+      "loss": 0.1593,
+      "step": 935
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.149499809660488e-05,
+      "loss": 0.1669,
+      "step": 936
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.128484712164164e-05,
+      "loss": 0.1685,
+      "step": 937
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.107489708001855e-05,
+      "loss": 0.1708,
+      "step": 938
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.086514906138563e-05,
+      "loss": 0.1715,
+      "step": 939
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.065560415434417e-05,
+      "loss": 0.1697,
+      "step": 940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.044626344644151e-05,
+      "loss": 0.1643,
+      "step": 941
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.023712802416508e-05,
+      "loss": 0.1665,
+      "step": 942
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.0028198972936836e-05,
+      "loss": 0.1777,
+      "step": 943
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.981947737710779e-05,
+      "loss": 0.1649,
+      "step": 944
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.9610964319952104e-05,
+      "loss": 0.1724,
+      "step": 945
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.940266088366173e-05,
+      "loss": 0.1639,
+      "step": 946
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.919456814934061e-05,
+      "loss": 0.1731,
+      "step": 947
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.8986687196999135e-05,
+      "loss": 0.1661,
+      "step": 948
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.877901910554862e-05,
+      "loss": 0.1647,
+      "step": 949
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.8571564952795475e-05,
+      "loss": 0.1661,
+      "step": 950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.8364325815435916e-05,
+      "loss": 0.1728,
+      "step": 951
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.815730276905014e-05,
+      "loss": 0.1671,
+      "step": 952
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.7950496888096795e-05,
+      "loss": 0.1689,
+      "step": 953
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.7743909245907535e-05,
+      "loss": 0.1674,
+      "step": 954
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.753754091468115e-05,
+      "loss": 0.1688,
+      "step": 955
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.7331392965478493e-05,
+      "loss": 0.1671,
+      "step": 956
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.712546646821627e-05,
+      "loss": 0.1756,
+      "step": 957
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.6919762491662164e-05,
+      "loss": 0.1706,
+      "step": 958
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.671428210342884e-05,
+      "loss": 0.1632,
+      "step": 959
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.650902636996837e-05,
+      "loss": 0.1615,
+      "step": 960
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.6303996356567177e-05,
+      "loss": 0.1653,
+      "step": 961
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.6099193127339864e-05,
+      "loss": 0.1656,
+      "step": 962
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.589461774522433e-05,
+      "loss": 0.1641,
+      "step": 963
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.5690271271975644e-05,
+      "loss": 0.1694,
+      "step": 964
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.548615476816097e-05,
+      "loss": 0.1718,
+      "step": 965
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.528226929315401e-05,
+      "loss": 0.1663,
+      "step": 966
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.507861590512916e-05,
+      "loss": 0.1752,
+      "step": 967
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.48751956610566e-05,
+      "loss": 0.1638,
+      "step": 968
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.467200961669619e-05,
+      "loss": 0.1655,
+      "step": 969
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.446905882659243e-05,
+      "loss": 0.1663,
+      "step": 970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.426634434406883e-05,
+      "loss": 0.1683,
+      "step": 971
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.40638672212224e-05,
+      "loss": 0.1655,
+      "step": 972
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.3861628508918384e-05,
+      "loss": 0.1668,
+      "step": 973
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.3659629256784424e-05,
+      "loss": 0.1613,
+      "step": 974
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.345787051320564e-05,
+      "loss": 0.1659,
+      "step": 975
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.325635332531864e-05,
+      "loss": 0.1625,
+      "step": 976
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.305507873900649e-05,
+      "loss": 0.1718,
+      "step": 977
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.2854047798893125e-05,
+      "loss": 0.1603,
+      "step": 978
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.26532615483379e-05,
+      "loss": 0.1663,
+      "step": 979
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.245272102943034e-05,
+      "loss": 0.1628,
+      "step": 980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.225242728298445e-05,
+      "loss": 0.1644,
+      "step": 981
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.2052381348533564e-05,
+      "loss": 0.1604,
+      "step": 982
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.1852584264324866e-05,
+      "loss": 0.1689,
+      "step": 983
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.165303706731397e-05,
+      "loss": 0.1591,
+      "step": 984
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.1453740793159586e-05,
+      "loss": 0.1703,
+      "step": 985
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.12546964762181e-05,
+      "loss": 0.168,
+      "step": 986
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.105590514953824e-05,
+      "loss": 0.1646,
+      "step": 987
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.085736784485571e-05,
+      "loss": 0.1598,
+      "step": 988
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.065908559258782e-05,
+      "loss": 0.1692,
+      "step": 989
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.046105942182815e-05,
+      "loss": 0.1702,
+      "step": 990
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.026329036034119e-05,
+      "loss": 0.1679,
+      "step": 991
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.006577943455706e-05,
+      "loss": 0.1637,
+      "step": 992
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.9868527669566113e-05,
+      "loss": 0.161,
+      "step": 993
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.967153608911366e-05,
+      "loss": 0.168,
+      "step": 994
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.947480571559462e-05,
+      "loss": 0.1756,
+      "step": 995
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.927833757004826e-05,
+      "loss": 0.1685,
+      "step": 996
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.908213267215287e-05,
+      "loss": 0.1678,
+      "step": 997
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.888619204022047e-05,
+      "loss": 0.1703,
+      "step": 998
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.869051669119154e-05,
+      "loss": 0.1676,
+      "step": 999
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.8495107640629675e-05,
+      "loss": 0.155,
+      "step": 1000
+    }
+  ],
+  "max_steps": 1452,
+  "num_train_epochs": 2,
+  "total_flos": 1.6985260825363814e+18,
+  "trial_name": null,
+  "trial_params": null
+}