diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9250 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9990224828934506,
+  "global_step": 1533,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002998043052837573,
+      "loss": 9.8511,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002998043052837573,
+      "loss": 6.188,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029960861056751466,
+      "loss": 5.9422,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000299412915851272,
+      "loss": 4.4474,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029921722113502935,
+      "loss": 3.7152,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029902152641878666,
+      "loss": 3.7503,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000298825831702544,
+      "loss": 3.443,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029863013698630135,
+      "loss": 3.4837,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029843444227005867,
+      "loss": 3.4489,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000298238747553816,
+      "loss": 3.3525,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029804305283757335,
+      "loss": 3.3712,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029784735812133067,
+      "loss": 3.3725,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029765166340508804,
+      "loss": 3.3204,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029745596868884536,
+      "loss": 3.3487,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002972602739726027,
+      "loss": 3.3468,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00029706457925636004,
+      "loss": 3.3251,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002968688845401174,
+      "loss": 3.3036,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029667318982387473,
+      "loss": 3.3156,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029647749510763204,
+      "loss": 3.3083,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002962818003913894,
+      "loss": 3.3087,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029608610567514673,
+      "loss": 3.3111,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029589041095890405,
+      "loss": 3.3252,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002956947162426614,
+      "loss": 3.3096,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029549902152641873,
+      "loss": 3.3336,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002953033268101761,
+      "loss": 3.3167,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002951076320939334,
+      "loss": 3.2763,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002949119373776908,
+      "loss": 3.3215,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002947162426614481,
+      "loss": 3.3226,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002945205479452055,
+      "loss": 3.3105,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002943248532289628,
+      "loss": 3.3021,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002941291585127201,
+      "loss": 3.2943,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002939334637964774,
+      "loss": 3.3157,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002937377690802348,
+      "loss": 3.2878,
+      "step": 33
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002935420743639921,
+      "loss": 3.3081,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002933463796477495,
+      "loss": 3.2917,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002931506849315068,
+      "loss": 3.2955,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029295499021526417,
+      "loss": 3.3038,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002927592954990215,
+      "loss": 3.2788,
+      "step": 38
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029256360078277886,
+      "loss": 3.3037,
+      "step": 39
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029236790606653617,
+      "loss": 3.2877,
+      "step": 40
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029217221135029354,
+      "loss": 3.2931,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029197651663405086,
+      "loss": 3.3072,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002917808219178082,
+      "loss": 3.3041,
+      "step": 43
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002915851272015655,
+      "loss": 3.3031,
+      "step": 44
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029138943248532286,
+      "loss": 3.3022,
+      "step": 45
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002911937377690802,
+      "loss": 3.2758,
+      "step": 46
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029099804305283755,
+      "loss": 3.3018,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029080234833659486,
+      "loss": 3.2911,
+      "step": 48
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029060665362035223,
+      "loss": 3.3105,
+      "step": 49
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029041095890410955,
+      "loss": 3.3076,
+      "step": 50
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002902152641878669,
+      "loss": 3.2728,
+      "step": 51
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029001956947162424,
+      "loss": 3.2984,
+      "step": 52
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00028982387475538155,
+      "loss": 3.2977,
+      "step": 53
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002896281800391389,
+      "loss": 3.2912,
+      "step": 54
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028943248532289624,
+      "loss": 3.2878,
+      "step": 55
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028923679060665356,
+      "loss": 3.2831,
+      "step": 56
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002890410958904109,
+      "loss": 3.291,
+      "step": 57
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00028884540117416824,
+      "loss": 3.3009,
+      "step": 58
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002886497064579256,
+      "loss": 3.3219,
+      "step": 59
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00028845401174168293,
+      "loss": 3.2959,
+      "step": 60
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002882583170254403,
+      "loss": 3.2987,
+      "step": 61
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002880626223091976,
+      "loss": 3.3099,
+      "step": 62
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000287866927592955,
+      "loss": 3.3113,
+      "step": 63
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002876712328767123,
+      "loss": 3.2871,
+      "step": 64
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002874755381604696,
+      "loss": 3.3086,
+      "step": 65
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000287279843444227,
+      "loss": 3.2915,
+      "step": 66
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002870841487279843,
+      "loss": 3.3023,
+      "step": 67
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002868884540117416,
+      "loss": 3.2963,
+      "step": 68
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000286692759295499,
+      "loss": 3.3007,
+      "step": 69
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002864970645792563,
+      "loss": 3.298,
+      "step": 70
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002863013698630137,
+      "loss": 3.3059,
+      "step": 71
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000286105675146771,
+      "loss": 3.2833,
+      "step": 72
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028590998043052837,
+      "loss": 3.2876,
+      "step": 73
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002857142857142857,
+      "loss": 3.3049,
+      "step": 74
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028551859099804305,
+      "loss": 3.2997,
+      "step": 75
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028532289628180037,
+      "loss": 3.2983,
+      "step": 76
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002851272015655577,
+      "loss": 3.2836,
+      "step": 77
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028493150684931505,
+      "loss": 3.2893,
+      "step": 78
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028473581213307237,
+      "loss": 3.2929,
+      "step": 79
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002845401174168297,
+      "loss": 3.2806,
+      "step": 80
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028434442270058706,
+      "loss": 3.2935,
+      "step": 81
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002841487279843444,
+      "loss": 3.3072,
+      "step": 82
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028395303326810174,
+      "loss": 3.3108,
+      "step": 83
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028375733855185906,
+      "loss": 3.2932,
+      "step": 84
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028356164383561643,
+      "loss": 3.2835,
+      "step": 85
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028336594911937375,
+      "loss": 3.3251,
+      "step": 86
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002831702544031311,
+      "loss": 3.2801,
+      "step": 87
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028297455968688843,
+      "loss": 3.2857,
+      "step": 88
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028277886497064575,
+      "loss": 3.3075,
+      "step": 89
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00028258317025440307,
+      "loss": 3.2917,
+      "step": 90
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00028238747553816044,
+      "loss": 3.2721,
+      "step": 91
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00028219178082191775,
+      "loss": 3.2839,
+      "step": 92
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002819960861056751,
+      "loss": 3.3037,
+      "step": 93
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00028180039138943244,
+      "loss": 3.2903,
+      "step": 94
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002816046966731898,
+      "loss": 3.283,
+      "step": 95
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002814090019569471,
+      "loss": 3.2963,
+      "step": 96
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002812133072407045,
+      "loss": 3.3007,
+      "step": 97
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002810176125244618,
+      "loss": 3.2914,
+      "step": 98
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00028082191780821913,
+      "loss": 3.3052,
+      "step": 99
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002806262230919765,
+      "loss": 3.308,
+      "step": 100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002804305283757338,
+      "loss": 3.3233,
+      "step": 101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00028023483365949113,
+      "loss": 3.2813,
+      "step": 102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002800391389432485,
+      "loss": 3.2777,
+      "step": 103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002798434442270058,
+      "loss": 3.29,
+      "step": 104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002796477495107632,
+      "loss": 3.2916,
+      "step": 105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002794520547945205,
+      "loss": 3.2815,
+      "step": 106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002792563600782779,
+      "loss": 3.2872,
+      "step": 107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002790606653620352,
+      "loss": 3.3095,
+      "step": 108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00027886497064579256,
+      "loss": 3.3155,
+      "step": 109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002786692759295499,
+      "loss": 3.3012,
+      "step": 110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002784735812133072,
+      "loss": 3.2934,
+      "step": 111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00027827788649706456,
+      "loss": 3.3186,
+      "step": 112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002780821917808219,
+      "loss": 3.2776,
+      "step": 113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002778864970645792,
+      "loss": 3.3155,
+      "step": 114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00027769080234833657,
+      "loss": 3.3051,
+      "step": 115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002774951076320939,
+      "loss": 3.3012,
+      "step": 116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00027729941291585125,
+      "loss": 3.2965,
+      "step": 117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00027710371819960857,
+      "loss": 3.31,
+      "step": 118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00027690802348336594,
+      "loss": 3.2916,
+      "step": 119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00027671232876712326,
+      "loss": 3.2975,
+      "step": 120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002765166340508806,
+      "loss": 3.2767,
+      "step": 121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00027632093933463794,
+      "loss": 3.2795,
+      "step": 122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00027612524461839526,
+      "loss": 3.2931,
+      "step": 123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00027592954990215263,
+      "loss": 3.2788,
+      "step": 124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00027573385518590995,
+      "loss": 3.2713,
+      "step": 125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00027553816046966726,
+      "loss": 3.2781,
+      "step": 126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00027534246575342463,
+      "loss": 3.2881,
+      "step": 127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00027514677103718195,
+      "loss": 3.2577,
+      "step": 128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002749510763209393,
+      "loss": 3.2739,
+      "step": 129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00027475538160469663,
+      "loss": 3.277,
+      "step": 130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000274559686888454,
+      "loss": 3.2712,
+      "step": 131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002743639921722113,
+      "loss": 3.2791,
+      "step": 132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002741682974559687,
+      "loss": 3.2689,
+      "step": 133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000273972602739726,
+      "loss": 3.266,
+      "step": 134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002737769080234833,
+      "loss": 3.2763,
+      "step": 135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00027358121330724064,
+      "loss": 3.2733,
+      "step": 136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000273385518590998,
+      "loss": 3.2581,
+      "step": 137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002731898238747553,
+      "loss": 3.2686,
+      "step": 138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002729941291585127,
+      "loss": 3.262,
+      "step": 139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00027279843444227,
+      "loss": 3.2614,
+      "step": 140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002726027397260274,
+      "loss": 3.2648,
+      "step": 141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002724070450097847,
+      "loss": 3.264,
+      "step": 142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00027221135029354207,
+      "loss": 3.2752,
+      "step": 143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002720156555772994,
+      "loss": 3.2459,
+      "step": 144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002718199608610567,
+      "loss": 3.2722,
+      "step": 145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002716242661448141,
+      "loss": 3.2584,
+      "step": 146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002714285714285714,
+      "loss": 3.2534,
+      "step": 147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002712328767123287,
+      "loss": 3.2674,
+      "step": 148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002710371819960861,
+      "loss": 3.2833,
+      "step": 149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002708414872798434,
+      "loss": 3.2681,
+      "step": 150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00027064579256360076,
+      "loss": 3.2882,
+      "step": 151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002704500978473581,
+      "loss": 3.2487,
+      "step": 152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00027025440313111545,
+      "loss": 3.2686,
+      "step": 153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00027005870841487277,
+      "loss": 3.2715,
+      "step": 154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00026986301369863014,
+      "loss": 3.3318,
+      "step": 155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00026966731898238745,
+      "loss": 3.2894,
+      "step": 156
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00026947162426614477,
+      "loss": 3.2855,
+      "step": 157
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00026927592954990214,
+      "loss": 3.2884,
+      "step": 158
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00026908023483365945,
+      "loss": 3.2857,
+      "step": 159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00026888454011741677,
+      "loss": 3.2585,
+      "step": 160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00026868884540117414,
+      "loss": 3.2767,
+      "step": 161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00026849315068493146,
+      "loss": 3.2683,
+      "step": 162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00026829745596868883,
+      "loss": 3.278,
+      "step": 163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00026810176125244614,
+      "loss": 3.2791,
+      "step": 164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002679060665362035,
+      "loss": 3.2493,
+      "step": 165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00026771037181996083,
+      "loss": 3.2582,
+      "step": 166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002675146771037182,
+      "loss": 3.2674,
+      "step": 167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002673189823874755,
+      "loss": 3.2644,
+      "step": 168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00026712328767123283,
+      "loss": 3.275,
+      "step": 169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002669275929549902,
+      "loss": 3.261,
+      "step": 170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002667318982387475,
+      "loss": 3.2672,
+      "step": 171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00026653620352250484,
+      "loss": 3.2667,
+      "step": 172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002663405088062622,
+      "loss": 3.2546,
+      "step": 173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002661448140900195,
+      "loss": 3.2809,
+      "step": 174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002659491193737769,
+      "loss": 3.2547,
+      "step": 175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002657534246575342,
+      "loss": 3.267,
+      "step": 176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002655577299412916,
+      "loss": 3.2613,
+      "step": 177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002653620352250489,
+      "loss": 3.2727,
+      "step": 178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00026516634050880627,
+      "loss": 3.2651,
+      "step": 179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002649706457925636,
+      "loss": 3.2562,
+      "step": 180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002647749510763209,
+      "loss": 3.2423,
+      "step": 181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002645792563600782,
+      "loss": 3.2899,
+      "step": 182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002643835616438356,
+      "loss": 3.2654,
+      "step": 183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002641878669275929,
+      "loss": 3.2669,
+      "step": 184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026399217221135027,
+      "loss": 3.2416,
+      "step": 185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002637964774951076,
+      "loss": 3.2704,
+      "step": 186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026360078277886496,
+      "loss": 3.2871,
+      "step": 187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002634050880626223,
+      "loss": 3.2606,
+      "step": 188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026320939334637964,
+      "loss": 3.2691,
+      "step": 189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026301369863013696,
+      "loss": 3.2654,
+      "step": 190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002628180039138943,
+      "loss": 3.2508,
+      "step": 191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026262230919765165,
+      "loss": 3.2768,
+      "step": 192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026242661448140896,
+      "loss": 3.2809,
+      "step": 193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002622309197651663,
+      "loss": 3.2623,
+      "step": 194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026203522504892365,
+      "loss": 3.2536,
+      "step": 195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026183953033268097,
+      "loss": 3.2727,
+      "step": 196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00026164383561643834,
+      "loss": 3.2631,
+      "step": 197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00026144814090019565,
+      "loss": 3.2742,
+      "step": 198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000261252446183953,
+      "loss": 3.2635,
+      "step": 199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00026105675146771034,
+      "loss": 3.2664,
+      "step": 200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002608610567514677,
+      "loss": 3.2646,
+      "step": 201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000260665362035225,
+      "loss": 3.2695,
+      "step": 202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026046966731898234,
+      "loss": 3.2475,
+      "step": 203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002602739726027397,
+      "loss": 3.2742,
+      "step": 204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026007827788649703,
+      "loss": 3.2562,
+      "step": 205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00025988258317025435,
+      "loss": 3.2738,
+      "step": 206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002596868884540117,
+      "loss": 3.241,
+      "step": 207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00025949119373776903,
+      "loss": 3.2863,
+      "step": 208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002592954990215264,
+      "loss": 3.2579,
+      "step": 209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002590998043052837,
+      "loss": 3.2495,
+      "step": 210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002589041095890411,
+      "loss": 3.2385,
+      "step": 211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002587084148727984,
+      "loss": 3.2534,
+      "step": 212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002585127201565558,
+      "loss": 3.272,
+      "step": 213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002583170254403131,
+      "loss": 3.2582,
+      "step": 214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002581213307240704,
+      "loss": 3.2588,
+      "step": 215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002579256360078278,
+      "loss": 3.2652,
+      "step": 216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002577299412915851,
+      "loss": 3.2526,
+      "step": 217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002575342465753424,
+      "loss": 3.2574,
+      "step": 218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002573385518590998,
+      "loss": 3.2686,
+      "step": 219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002571428571428571,
+      "loss": 3.2537,
+      "step": 220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025694716242661447,
+      "loss": 3.2528,
+      "step": 221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002567514677103718,
+      "loss": 3.2461,
+      "step": 222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025655577299412915,
+      "loss": 3.2619,
+      "step": 223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025636007827788647,
+      "loss": 3.2562,
+      "step": 224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025616438356164384,
+      "loss": 3.2308,
+      "step": 225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025596868884540116,
+      "loss": 3.2509,
+      "step": 226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002557729941291585,
+      "loss": 3.245,
+      "step": 227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002555772994129158,
+      "loss": 3.2481,
+      "step": 228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00025538160469667316,
+      "loss": 3.2704,
+      "step": 229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002551859099804305,
+      "loss": 3.2524,
+      "step": 230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00025499021526418785,
+      "loss": 3.2566,
+      "step": 231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00025479452054794516,
+      "loss": 3.2582,
+      "step": 232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00025459882583170253,
+      "loss": 3.2517,
+      "step": 233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00025440313111545985,
+      "loss": 3.2534,
+      "step": 234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002542074363992172,
+      "loss": 3.2599,
+      "step": 235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00025401174168297454,
+      "loss": 3.2448,
+      "step": 236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00025381604696673185,
+      "loss": 3.2577,
+      "step": 237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002536203522504892,
+      "loss": 3.2621,
+      "step": 238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00025342465753424654,
+      "loss": 3.2591,
+      "step": 239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00025322896281800385,
+      "loss": 3.2394,
+      "step": 240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002530332681017612,
+      "loss": 3.2334,
+      "step": 241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00025283757338551854,
+      "loss": 3.2331,
+      "step": 242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002526418786692759,
+      "loss": 3.2516,
+      "step": 243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025244618395303323,
+      "loss": 3.2424,
+      "step": 244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002522504892367906,
+      "loss": 3.2745,
+      "step": 245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002520547945205479,
+      "loss": 3.2458,
+      "step": 246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002518590998043053,
+      "loss": 3.2672,
+      "step": 247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002516634050880626,
+      "loss": 3.2674,
+      "step": 248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002514677103718199,
+      "loss": 3.2412,
+      "step": 249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002512720156555773,
+      "loss": 3.2619,
+      "step": 250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002510763209393346,
+      "loss": 3.2398,
+      "step": 251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002508806262230919,
+      "loss": 3.3514,
+      "step": 252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002506849315068493,
+      "loss": 3.241,
+      "step": 253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002504892367906066,
+      "loss": 3.2831,
+      "step": 254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000250293542074364,
+      "loss": 3.2722,
+      "step": 255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002500978473581213,
+      "loss": 3.2489,
+      "step": 256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00024990215264187866,
+      "loss": 3.2612,
+      "step": 257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000249706457925636,
+      "loss": 3.2677,
+      "step": 258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024951076320939335,
+      "loss": 3.2593,
+      "step": 259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024931506849315067,
+      "loss": 3.2472,
+      "step": 260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000249119373776908,
+      "loss": 3.2613,
+      "step": 261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024892367906066535,
+      "loss": 3.2534,
+      "step": 262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00024872798434442267,
+      "loss": 3.2596,
+      "step": 263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00024853228962818,
+      "loss": 3.2533,
+      "step": 264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00024833659491193736,
+      "loss": 3.2543,
+      "step": 265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00024814090019569467,
+      "loss": 3.2532,
+      "step": 266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00024794520547945204,
+      "loss": 3.2553,
+      "step": 267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00024774951076320936,
+      "loss": 3.2615,
+      "step": 268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024755381604696673,
+      "loss": 3.2719,
+      "step": 269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024735812133072404,
+      "loss": 3.2568,
+      "step": 270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002471624266144814,
+      "loss": 3.2807,
+      "step": 271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024696673189823873,
+      "loss": 3.2515,
+      "step": 272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024677103718199605,
+      "loss": 3.252,
+      "step": 273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00024657534246575336,
+      "loss": 3.2526,
+      "step": 274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00024637964774951073,
+      "loss": 3.2635,
+      "step": 275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00024618395303326805,
+      "loss": 3.2602,
+      "step": 276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002459882583170254,
+      "loss": 3.2795,
+      "step": 277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00024579256360078274,
+      "loss": 3.2548,
+      "step": 278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002455968688845401,
+      "loss": 3.2573,
+      "step": 279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002454011741682974,
+      "loss": 3.2685,
+      "step": 280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002452054794520548,
+      "loss": 3.2407,
+      "step": 281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002450097847358121,
+      "loss": 3.2391,
+      "step": 282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002448140900195694,
+      "loss": 3.261,
+      "step": 283
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002446183953033268,
+      "loss": 3.2429,
+      "step": 284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002444227005870841,
+      "loss": 3.2453,
+      "step": 285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024422700587084143,
+      "loss": 3.2567,
+      "step": 286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002440313111545988,
+      "loss": 3.2556,
+      "step": 287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024383561643835614,
+      "loss": 3.2319,
+      "step": 288
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024363992172211349,
+      "loss": 3.2526,
+      "step": 289
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002434442270058708,
+      "loss": 3.2126,
+      "step": 290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024324853228962817,
+      "loss": 3.2325,
+      "step": 291
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002430528375733855,
+      "loss": 3.2457,
+      "step": 292
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024285714285714283,
+      "loss": 3.2599,
+      "step": 293
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024266144814090015,
+      "loss": 3.2293,
+      "step": 294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00024246575342465752,
+      "loss": 3.2366,
+      "step": 295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00024227005870841483,
+      "loss": 3.2429,
+      "step": 296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002420743639921722,
+      "loss": 3.2486,
+      "step": 297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00024187866927592952,
+      "loss": 3.2709,
+      "step": 298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00024168297455968686,
+      "loss": 3.2296,
+      "step": 299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00024148727984344418,
+      "loss": 3.2423,
+      "step": 300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00024129158512720155,
+      "loss": 3.2379,
+      "step": 301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00024109589041095887,
+      "loss": 3.2257,
+      "step": 302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00024090019569471624,
+      "loss": 3.2138,
+      "step": 303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00024070450097847355,
+      "loss": 3.2305,
+      "step": 304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002405088062622309,
+      "loss": 3.2308,
+      "step": 305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002403131115459882,
+      "loss": 3.2321,
+      "step": 306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00024011741682974558,
+      "loss": 3.235,
+      "step": 307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002399217221135029,
+      "loss": 3.2562,
+      "step": 308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023972602739726024,
+      "loss": 3.2388,
+      "step": 309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002395303326810176,
+      "loss": 3.2424,
+      "step": 310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023933463796477493,
+      "loss": 3.2445,
+      "step": 311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023913894324853225,
+      "loss": 3.2203,
+      "step": 312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023894324853228962,
+      "loss": 3.2501,
+      "step": 313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023874755381604693,
+      "loss": 3.2524,
+      "step": 314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00023855185909980428,
+      "loss": 3.2162,
+      "step": 315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00023835616438356162,
+      "loss": 3.2458,
+      "step": 316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00023816046966731896,
+      "loss": 3.2323,
+      "step": 317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00023796477495107628,
+      "loss": 3.2198,
+      "step": 318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00023776908023483365,
+      "loss": 3.2279,
+      "step": 319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00023757338551859097,
+      "loss": 3.2285,
+      "step": 320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002373776908023483,
+      "loss": 3.2181,
+      "step": 321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00023718199608610565,
+      "loss": 3.2484,
+      "step": 322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000236986301369863,
+      "loss": 3.2226,
+      "step": 323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002367906066536203,
+      "loss": 3.2449,
+      "step": 324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00023659491193737768,
+      "loss": 3.214,
+      "step": 325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000236399217221135,
+      "loss": 3.2303,
+      "step": 326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00023620352250489234,
+      "loss": 3.2398,
+      "step": 327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00023600782778864968,
+      "loss": 3.23,
+      "step": 328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00023581213307240703,
+      "loss": 3.2279,
+      "step": 329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00023561643835616434,
+      "loss": 3.2459,
+      "step": 330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00023542074363992171,
+      "loss": 3.231,
+      "step": 331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00023522504892367903,
+      "loss": 3.231,
+      "step": 332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00023502935420743637,
+      "loss": 3.2127,
+      "step": 333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00023483365949119372,
+      "loss": 3.2426,
+      "step": 334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00023463796477495106,
+      "loss": 3.2341,
+      "step": 335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023444227005870838,
+      "loss": 3.2154,
+      "step": 336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023424657534246575,
+      "loss": 3.2525,
+      "step": 337
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023405088062622306,
+      "loss": 3.226,
+      "step": 338
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002338551859099804,
+      "loss": 3.2392,
+      "step": 339
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023365949119373772,
+      "loss": 3.2265,
+      "step": 340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002334637964774951,
+      "loss": 3.2227,
+      "step": 341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002332681017612524,
+      "loss": 3.2645,
+      "step": 342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00023307240704500978,
+      "loss": 3.2254,
+      "step": 343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002328767123287671,
+      "loss": 3.2208,
+      "step": 344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00023268101761252444,
+      "loss": 3.2552,
+      "step": 345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023248532289628176,
+      "loss": 3.2221,
+      "step": 346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023228962818003913,
+      "loss": 3.2311,
+      "step": 347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023209393346379644,
+      "loss": 3.2324,
+      "step": 348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002318982387475538,
+      "loss": 3.231,
+      "step": 349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023170254403131113,
+      "loss": 3.2104,
+      "step": 350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023150684931506847,
+      "loss": 3.2188,
+      "step": 351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002313111545988258,
+      "loss": 3.2336,
+      "step": 352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023111545988258316,
+      "loss": 3.2231,
+      "step": 353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023091976516634047,
+      "loss": 3.2291,
+      "step": 354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023072407045009782,
+      "loss": 3.2345,
+      "step": 355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023052837573385516,
+      "loss": 3.2295,
+      "step": 356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002303326810176125,
+      "loss": 3.2328,
+      "step": 357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023013698630136982,
+      "loss": 3.2326,
+      "step": 358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002299412915851272,
+      "loss": 3.2417,
+      "step": 359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002297455968688845,
+      "loss": 3.2101,
+      "step": 360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022954990215264185,
+      "loss": 3.2424,
+      "step": 361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002293542074363992,
+      "loss": 3.2306,
+      "step": 362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022915851272015654,
+      "loss": 3.2162,
+      "step": 363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022896281800391385,
+      "loss": 3.2224,
+      "step": 364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00022876712328767122,
+      "loss": 3.2394,
+      "step": 365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022857142857142854,
+      "loss": 3.2171,
+      "step": 366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022837573385518588,
+      "loss": 3.231,
+      "step": 367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022818003913894323,
+      "loss": 3.2285,
+      "step": 368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022798434442270057,
+      "loss": 3.2165,
+      "step": 369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022778864970645789,
+      "loss": 3.2229,
+      "step": 370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022759295499021526,
+      "loss": 3.2362,
+      "step": 371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022739726027397257,
+      "loss": 3.2053,
+      "step": 372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022720156555772992,
+      "loss": 3.225,
+      "step": 373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022700587084148726,
+      "loss": 3.2213,
+      "step": 374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002268101761252446,
+      "loss": 3.2369,
+      "step": 375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00022661448140900192,
+      "loss": 3.2185,
+      "step": 376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002264187866927593,
+      "loss": 3.2339,
+      "step": 377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002262230919765166,
+      "loss": 3.2277,
+      "step": 378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00022602739726027395,
+      "loss": 3.2195,
+      "step": 379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002258317025440313,
+      "loss": 3.2055,
+      "step": 380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00022563600782778863,
+      "loss": 3.2305,
+      "step": 381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00022544031311154595,
+      "loss": 3.2258,
+      "step": 382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00022524461839530332,
+      "loss": 3.2236,
+      "step": 383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00022504892367906064,
+      "loss": 3.2307,
+      "step": 384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00022485322896281798,
+      "loss": 3.227,
+      "step": 385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002246575342465753,
+      "loss": 3.2127,
+      "step": 386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00022446183953033267,
+      "loss": 3.2243,
+      "step": 387
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00022426614481408998,
+      "loss": 3.2343,
+      "step": 388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00022407045009784735,
+      "loss": 3.2228,
+      "step": 389
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00022387475538160467,
+      "loss": 3.214,
+      "step": 390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00022367906066536201,
+      "loss": 3.2134,
+      "step": 391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00022348336594911933,
+      "loss": 3.2233,
+      "step": 392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002232876712328767,
+      "loss": 3.2006,
+      "step": 393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00022309197651663402,
+      "loss": 3.2294,
+      "step": 394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002228962818003914,
+      "loss": 3.2246,
+      "step": 395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002227005870841487,
+      "loss": 3.2249,
+      "step": 396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00022250489236790605,
+      "loss": 3.2249,
+      "step": 397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00022230919765166336,
+      "loss": 3.2321,
+      "step": 398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00022211350293542073,
+      "loss": 3.2351,
+      "step": 399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00022191780821917805,
+      "loss": 3.2277,
+      "step": 400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002217221135029354,
+      "loss": 3.2224,
+      "step": 401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022152641878669274,
+      "loss": 3.2254,
+      "step": 402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022133072407045008,
+      "loss": 3.2131,
+      "step": 403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002211350293542074,
+      "loss": 3.2256,
+      "step": 404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022093933463796477,
+      "loss": 3.2157,
+      "step": 405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022074363992172208,
+      "loss": 3.2321,
+      "step": 406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00022054794520547942,
+      "loss": 3.2162,
+      "step": 407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00022035225048923677,
+      "loss": 3.2012,
+      "step": 408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002201565557729941,
+      "loss": 3.2206,
+      "step": 409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00021996086105675143,
+      "loss": 3.2201,
+      "step": 410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002197651663405088,
+      "loss": 3.2085,
+      "step": 411
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00021956947162426611,
+      "loss": 3.2232,
+      "step": 412
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00021937377690802346,
+      "loss": 3.2345,
+      "step": 413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002191780821917808,
+      "loss": 3.2241,
+      "step": 414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00021898238747553814,
+      "loss": 3.2347,
+      "step": 415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00021878669275929546,
+      "loss": 3.2207,
+      "step": 416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00021859099804305283,
+      "loss": 3.2173,
+      "step": 417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00021839530332681015,
+      "loss": 3.2202,
+      "step": 418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002181996086105675,
+      "loss": 3.218,
+      "step": 419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00021800391389432483,
+      "loss": 3.211,
+      "step": 420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00021780821917808218,
+      "loss": 3.2094,
+      "step": 421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002176125244618395,
+      "loss": 3.2192,
+      "step": 422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00021741682974559686,
+      "loss": 3.2214,
+      "step": 423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00021722113502935418,
+      "loss": 3.2346,
+      "step": 424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00021702544031311152,
+      "loss": 3.2299,
+      "step": 425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00021682974559686887,
+      "loss": 3.2275,
+      "step": 426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002166340508806262,
+      "loss": 3.2017,
+      "step": 427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00021643835616438353,
+      "loss": 3.2083,
+      "step": 428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002162426614481409,
+      "loss": 3.2247,
+      "step": 429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002160469667318982,
+      "loss": 3.211,
+      "step": 430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00021585127201565556,
+      "loss": 3.2229,
+      "step": 431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00021565557729941287,
+      "loss": 3.197,
+      "step": 432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00021545988258317024,
+      "loss": 3.2035,
+      "step": 433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00021526418786692756,
+      "loss": 3.2118,
+      "step": 434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00021506849315068493,
+      "loss": 3.2077,
+      "step": 435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00021487279843444224,
+      "loss": 3.215,
+      "step": 436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002146771037181996,
+      "loss": 3.2145,
+      "step": 437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002144814090019569,
+      "loss": 3.1987,
+      "step": 438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00021428571428571427,
+      "loss": 3.2123,
+      "step": 439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002140900195694716,
+      "loss": 3.1901,
+      "step": 440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00021389432485322896,
+      "loss": 3.1919,
+      "step": 441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00021369863013698628,
+      "loss": 3.2068,
+      "step": 442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00021350293542074362,
+      "loss": 3.2125,
+      "step": 443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00021330724070450094,
+      "loss": 3.1952,
+      "step": 444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002131115459882583,
+      "loss": 3.2172,
+      "step": 445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00021291585127201562,
+      "loss": 3.2029,
+      "step": 446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00021272015655577297,
+      "loss": 3.197,
+      "step": 447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002125244618395303,
+      "loss": 3.1921,
+      "step": 448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021232876712328765,
+      "loss": 3.1898,
+      "step": 449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021213307240704497,
+      "loss": 3.201,
+      "step": 450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021193737769080234,
+      "loss": 3.2142,
+      "step": 451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021174168297455966,
+      "loss": 3.2015,
+      "step": 452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000211545988258317,
+      "loss": 3.1952,
+      "step": 453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00021135029354207434,
+      "loss": 3.2191,
+      "step": 454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00021115459882583169,
+      "loss": 3.2111,
+      "step": 455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000210958904109589,
+      "loss": 3.1956,
+      "step": 456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00021076320939334637,
+      "loss": 3.1995,
+      "step": 457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002105675146771037,
+      "loss": 3.1866,
+      "step": 458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00021037181996086103,
+      "loss": 3.2022,
+      "step": 459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00021017612524461838,
+      "loss": 3.197,
+      "step": 460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00020998043052837572,
+      "loss": 3.1843,
+      "step": 461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00020978473581213303,
+      "loss": 3.202,
+      "step": 462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002095890410958904,
+      "loss": 3.1863,
+      "step": 463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00020939334637964772,
+      "loss": 3.2066,
+      "step": 464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00020919765166340506,
+      "loss": 3.217,
+      "step": 465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002090019569471624,
+      "loss": 3.204,
+      "step": 466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00020880626223091975,
+      "loss": 3.2038,
+      "step": 467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00020861056751467707,
+      "loss": 3.2124,
+      "step": 468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00020841487279843444,
+      "loss": 3.1845,
+      "step": 469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00020821917808219175,
+      "loss": 3.2088,
+      "step": 470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002080234833659491,
+      "loss": 3.2019,
+      "step": 471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00020782778864970644,
+      "loss": 3.1932,
+      "step": 472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00020763209393346378,
+      "loss": 3.1972,
+      "step": 473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002074363992172211,
+      "loss": 3.1964,
+      "step": 474
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00020724070450097847,
+      "loss": 3.197,
+      "step": 475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002070450097847358,
+      "loss": 3.1952,
+      "step": 476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00020684931506849313,
+      "loss": 3.1906,
+      "step": 477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00020665362035225045,
+      "loss": 3.1882,
+      "step": 478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00020645792563600782,
+      "loss": 3.2031,
+      "step": 479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00020626223091976513,
+      "loss": 3.1884,
+      "step": 480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002060665362035225,
+      "loss": 3.1871,
+      "step": 481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00020587084148727982,
+      "loss": 3.1905,
+      "step": 482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00020567514677103716,
+      "loss": 3.1984,
+      "step": 483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00020547945205479448,
+      "loss": 3.1861,
+      "step": 484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00020528375733855185,
+      "loss": 3.2009,
+      "step": 485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00020508806262230917,
+      "loss": 3.1891,
+      "step": 486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00020489236790606654,
+      "loss": 3.1748,
+      "step": 487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00020469667318982385,
+      "loss": 3.2082,
+      "step": 488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002045009784735812,
+      "loss": 3.2025,
+      "step": 489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002043052837573385,
+      "loss": 3.186,
+      "step": 490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00020410958904109588,
+      "loss": 3.1866,
+      "step": 491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002039138943248532,
+      "loss": 3.1825,
+      "step": 492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00020371819960861057,
+      "loss": 3.1905,
+      "step": 493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00020352250489236788,
+      "loss": 3.1809,
+      "step": 494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00020332681017612523,
+      "loss": 3.1974,
+      "step": 495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00020313111545988254,
+      "loss": 3.2088,
+      "step": 496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00020293542074363991,
+      "loss": 3.1773,
+      "step": 497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00020273972602739723,
+      "loss": 3.2,
+      "step": 498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00020254403131115457,
+      "loss": 3.1715,
+      "step": 499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00020234833659491192,
+      "loss": 3.1883,
+      "step": 500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00020215264187866926,
+      "loss": 3.1853,
+      "step": 501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00020195694716242658,
+      "loss": 3.1999,
+      "step": 502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00020176125244618395,
+      "loss": 3.1896,
+      "step": 503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00020156555772994126,
+      "loss": 3.1796,
+      "step": 504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002013698630136986,
+      "loss": 3.1942,
+      "step": 505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00020117416829745595,
+      "loss": 3.2231,
+      "step": 506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002009784735812133,
+      "loss": 3.1886,
+      "step": 507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002007827788649706,
+      "loss": 3.1945,
+      "step": 508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00020058708414872798,
+      "loss": 3.1898,
+      "step": 509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002003913894324853,
+      "loss": 3.1947,
+      "step": 510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00020019569471624264,
+      "loss": 3.1883,
+      "step": 511
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 3.519240379333496,
+      "eval_runtime": 508.4632,
+      "eval_samples_per_second": 31.135,
+      "eval_steps_per_second": 0.071,
+      "eval_wer": 1.0,
+      "step": 511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 4.7961,
+      "step": 512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019980430528375733,
+      "loss": 3.1885,
+      "step": 513
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019960861056751464,
+      "loss": 3.1757,
+      "step": 514
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000199412915851272,
+      "loss": 3.202,
+      "step": 515
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019921722113502933,
+      "loss": 3.1831,
+      "step": 516
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019902152641878667,
+      "loss": 3.201,
+      "step": 517
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019882583170254402,
+      "loss": 3.175,
+      "step": 518
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00019863013698630136,
+      "loss": 3.1899,
+      "step": 519
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00019843444227005867,
+      "loss": 3.1941,
+      "step": 520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00019823874755381604,
+      "loss": 3.1991,
+      "step": 521
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00019804305283757336,
+      "loss": 3.1991,
+      "step": 522
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001978473581213307,
+      "loss": 3.1906,
+      "step": 523
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00019765166340508805,
+      "loss": 3.1671,
+      "step": 524
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001974559686888454,
+      "loss": 3.1912,
+      "step": 525
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001972602739726027,
+      "loss": 3.1836,
+      "step": 526
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00019706457925636008,
+      "loss": 3.1905,
+      "step": 527
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001968688845401174,
+      "loss": 3.1853,
+      "step": 528
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00019667318982387474,
+      "loss": 3.2102,
+      "step": 529
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00019647749510763205,
+      "loss": 3.1815,
+      "step": 530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00019628180039138942,
+      "loss": 3.1788,
+      "step": 531
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00019608610567514674,
+      "loss": 3.1824,
+      "step": 532
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001958904109589041,
+      "loss": 3.1827,
+      "step": 533
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00019569471624266143,
+      "loss": 3.1924,
+      "step": 534
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00019549902152641877,
+      "loss": 3.177,
+      "step": 535
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00019530332681017609,
+      "loss": 3.1616,
+      "step": 536
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00019510763209393346,
+      "loss": 3.1888,
+      "step": 537
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00019491193737769077,
+      "loss": 3.1857,
+      "step": 538
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00019471624266144814,
+      "loss": 3.1935,
+      "step": 539
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00019452054794520546,
+      "loss": 3.1786,
+      "step": 540
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001943248532289628,
+      "loss": 3.1816,
+      "step": 541
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00019412915851272012,
+      "loss": 3.1987,
+      "step": 542
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001939334637964775,
+      "loss": 3.1934,
+      "step": 543
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001937377690802348,
+      "loss": 3.1769,
+      "step": 544
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00019354207436399215,
+      "loss": 3.1844,
+      "step": 545
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001933463796477495,
+      "loss": 3.2072,
+      "step": 546
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00019315068493150683,
+      "loss": 3.2114,
+      "step": 547
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00019295499021526415,
+      "loss": 3.2147,
+      "step": 548
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00019275929549902152,
+      "loss": 3.2072,
+      "step": 549
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00019256360078277884,
+      "loss": 3.179,
+      "step": 550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00019236790606653618,
+      "loss": 3.1849,
+      "step": 551
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00019217221135029352,
+      "loss": 3.1866,
+      "step": 552
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00019197651663405087,
+      "loss": 3.1894,
+      "step": 553
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00019178082191780818,
+      "loss": 3.1924,
+      "step": 554
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00019158512720156555,
+      "loss": 3.1798,
+      "step": 555
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00019138943248532287,
+      "loss": 3.1698,
+      "step": 556
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00019119373776908021,
+      "loss": 3.1712,
+      "step": 557
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00019099804305283756,
+      "loss": 3.1818,
+      "step": 558
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001908023483365949,
+      "loss": 3.178,
+      "step": 559
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00019060665362035222,
+      "loss": 3.1869,
+      "step": 560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001904109589041096,
+      "loss": 3.1863,
+      "step": 561
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001902152641878669,
+      "loss": 3.1727,
+      "step": 562
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00019001956947162425,
+      "loss": 3.1704,
+      "step": 563
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001898238747553816,
+      "loss": 3.1992,
+      "step": 564
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018962818003913893,
+      "loss": 3.1775,
+      "step": 565
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018943248532289625,
+      "loss": 3.1862,
+      "step": 566
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018923679060665362,
+      "loss": 3.1678,
+      "step": 567
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018904109589041094,
+      "loss": 3.1795,
+      "step": 568
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018884540117416828,
+      "loss": 3.1739,
+      "step": 569
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018864970645792562,
+      "loss": 3.1721,
+      "step": 570
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018845401174168297,
+      "loss": 3.1792,
+      "step": 571
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018825831702544028,
+      "loss": 3.1794,
+      "step": 572
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018806262230919765,
+      "loss": 3.1725,
+      "step": 573
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018786692759295497,
+      "loss": 3.183,
+      "step": 574
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001876712328767123,
+      "loss": 3.1853,
+      "step": 575
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018747553816046963,
+      "loss": 3.1672,
+      "step": 576
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000187279843444227,
+      "loss": 3.159,
+      "step": 577
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018708414872798431,
+      "loss": 3.1886,
+      "step": 578
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018688845401174168,
+      "loss": 3.1689,
+      "step": 579
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000186692759295499,
+      "loss": 3.1761,
+      "step": 580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018649706457925634,
+      "loss": 3.1779,
+      "step": 581
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018630136986301366,
+      "loss": 3.1857,
+      "step": 582
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018610567514677103,
+      "loss": 3.1752,
+      "step": 583
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018590998043052835,
+      "loss": 3.1783,
+      "step": 584
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018571428571428572,
+      "loss": 3.1647,
+      "step": 585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018551859099804303,
+      "loss": 3.1541,
+      "step": 586
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018532289628180038,
+      "loss": 3.1607,
+      "step": 587
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001851272015655577,
+      "loss": 3.1729,
+      "step": 588
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018493150684931506,
+      "loss": 3.172,
+      "step": 589
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018473581213307238,
+      "loss": 3.1645,
+      "step": 590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018454011741682972,
+      "loss": 3.1598,
+      "step": 591
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018434442270058707,
+      "loss": 3.1866,
+      "step": 592
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001841487279843444,
+      "loss": 3.179,
+      "step": 593
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018395303326810173,
+      "loss": 3.1719,
+      "step": 594
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001837573385518591,
+      "loss": 3.1766,
+      "step": 595
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001835616438356164,
+      "loss": 3.1781,
+      "step": 596
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018336594911937376,
+      "loss": 3.1735,
+      "step": 597
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001831702544031311,
+      "loss": 3.1646,
+      "step": 598
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018297455968688844,
+      "loss": 3.1759,
+      "step": 599
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018277886497064576,
+      "loss": 3.1729,
+      "step": 600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018258317025440313,
+      "loss": 3.1831,
+      "step": 601
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018238747553816044,
+      "loss": 3.1864,
+      "step": 602
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001821917808219178,
+      "loss": 3.1831,
+      "step": 603
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018199608610567513,
+      "loss": 3.1806,
+      "step": 604
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018180039138943247,
+      "loss": 3.1729,
+      "step": 605
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001816046966731898,
+      "loss": 3.1798,
+      "step": 606
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018140900195694716,
+      "loss": 3.1612,
+      "step": 607
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018121330724070448,
+      "loss": 3.1533,
+      "step": 608
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018101761252446182,
+      "loss": 3.1838,
+      "step": 609
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018082191780821916,
+      "loss": 3.1636,
+      "step": 610
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001806262230919765,
+      "loss": 3.1589,
+      "step": 611
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018043052837573382,
+      "loss": 3.1738,
+      "step": 612
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001802348336594912,
+      "loss": 3.2014,
+      "step": 613
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001800391389432485,
+      "loss": 3.1662,
+      "step": 614
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017984344422700585,
+      "loss": 3.1626,
+      "step": 615
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001796477495107632,
+      "loss": 3.1654,
+      "step": 616
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017945205479452054,
+      "loss": 3.1632,
+      "step": 617
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017925636007827786,
+      "loss": 3.1679,
+      "step": 618
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017906066536203523,
+      "loss": 3.1821,
+      "step": 619
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017886497064579254,
+      "loss": 3.1765,
+      "step": 620
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017866927592954989,
+      "loss": 3.185,
+      "step": 621
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001784735812133072,
+      "loss": 3.1564,
+      "step": 622
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017827788649706457,
+      "loss": 3.1747,
+      "step": 623
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001780821917808219,
+      "loss": 3.1589,
+      "step": 624
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017788649706457926,
+      "loss": 3.1813,
+      "step": 625
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017769080234833658,
+      "loss": 3.1688,
+      "step": 626
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017749510763209392,
+      "loss": 3.1721,
+      "step": 627
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017729941291585123,
+      "loss": 3.1574,
+      "step": 628
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001771037181996086,
+      "loss": 3.1615,
+      "step": 629
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017690802348336592,
+      "loss": 3.1536,
+      "step": 630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001767123287671233,
+      "loss": 3.1602,
+      "step": 631
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001765166340508806,
+      "loss": 3.18,
+      "step": 632
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017632093933463795,
+      "loss": 3.1716,
+      "step": 633
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017612524461839527,
+      "loss": 3.1635,
+      "step": 634
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017592954990215264,
+      "loss": 3.1576,
+      "step": 635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017573385518590995,
+      "loss": 3.1521,
+      "step": 636
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001755381604696673,
+      "loss": 3.1763,
+      "step": 637
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017534246575342464,
+      "loss": 3.1736,
+      "step": 638
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017514677103718198,
+      "loss": 3.1842,
+      "step": 639
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001749510763209393,
+      "loss": 3.155,
+      "step": 640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017475538160469667,
+      "loss": 3.1671,
+      "step": 641
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000174559686888454,
+      "loss": 3.1791,
+      "step": 642
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017436399217221133,
+      "loss": 3.1427,
+      "step": 643
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017416829745596867,
+      "loss": 3.1859,
+      "step": 644
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017397260273972602,
+      "loss": 3.183,
+      "step": 645
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017377690802348333,
+      "loss": 3.1634,
+      "step": 646
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001735812133072407,
+      "loss": 3.1799,
+      "step": 647
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017338551859099802,
+      "loss": 3.1703,
+      "step": 648
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017318982387475536,
+      "loss": 3.1582,
+      "step": 649
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001729941291585127,
+      "loss": 3.1662,
+      "step": 650
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017279843444227005,
+      "loss": 3.1642,
+      "step": 651
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017260273972602737,
+      "loss": 3.173,
+      "step": 652
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017240704500978474,
+      "loss": 3.168,
+      "step": 653
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017221135029354205,
+      "loss": 3.1578,
+      "step": 654
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001720156555772994,
+      "loss": 3.1685,
+      "step": 655
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017181996086105674,
+      "loss": 3.1567,
+      "step": 656
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017162426614481408,
+      "loss": 3.1639,
+      "step": 657
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001714285714285714,
+      "loss": 3.1586,
+      "step": 658
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017123287671232877,
+      "loss": 3.1552,
+      "step": 659
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017103718199608608,
+      "loss": 3.153,
+      "step": 660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017084148727984343,
+      "loss": 3.1729,
+      "step": 661
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017064579256360077,
+      "loss": 3.1465,
+      "step": 662
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017045009784735811,
+      "loss": 3.168,
+      "step": 663
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017025440313111543,
+      "loss": 3.1802,
+      "step": 664
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001700587084148728,
+      "loss": 3.1603,
+      "step": 665
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00016986301369863012,
+      "loss": 3.1563,
+      "step": 666
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00016966731898238746,
+      "loss": 3.1637,
+      "step": 667
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00016947162426614478,
+      "loss": 3.1618,
+      "step": 668
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00016927592954990215,
+      "loss": 3.1485,
+      "step": 669
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00016908023483365946,
+      "loss": 3.1536,
+      "step": 670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00016888454011741683,
+      "loss": 3.1656,
+      "step": 671
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00016868884540117415,
+      "loss": 3.1689,
+      "step": 672
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001684931506849315,
+      "loss": 3.1441,
+      "step": 673
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001682974559686888,
+      "loss": 3.1692,
+      "step": 674
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00016810176125244618,
+      "loss": 3.1777,
+      "step": 675
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001679060665362035,
+      "loss": 3.1608,
+      "step": 676
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00016771037181996087,
+      "loss": 3.1557,
+      "step": 677
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00016751467710371818,
+      "loss": 3.1526,
+      "step": 678
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00016731898238747553,
+      "loss": 3.1393,
+      "step": 679
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00016712328767123284,
+      "loss": 3.165,
+      "step": 680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001669275929549902,
+      "loss": 3.1819,
+      "step": 681
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00016673189823874753,
+      "loss": 3.1709,
+      "step": 682
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00016653620352250487,
+      "loss": 3.166,
+      "step": 683
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00016634050880626222,
+      "loss": 3.1599,
+      "step": 684
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00016614481409001956,
+      "loss": 3.1628,
+      "step": 685
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00016594911937377687,
+      "loss": 3.1642,
+      "step": 686
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00016575342465753425,
+      "loss": 3.1734,
+      "step": 687
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00016555772994129156,
+      "loss": 3.1581,
+      "step": 688
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001653620352250489,
+      "loss": 3.1575,
+      "step": 689
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00016516634050880625,
+      "loss": 3.1557,
+      "step": 690
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001649706457925636,
+      "loss": 3.15,
+      "step": 691
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001647749510763209,
+      "loss": 3.1475,
+      "step": 692
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00016457925636007828,
+      "loss": 3.1371,
+      "step": 693
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001643835616438356,
+      "loss": 3.1573,
+      "step": 694
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00016418786692759294,
+      "loss": 3.1629,
+      "step": 695
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00016399217221135028,
+      "loss": 3.1715,
+      "step": 696
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00016379647749510762,
+      "loss": 3.1557,
+      "step": 697
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00016360078277886494,
+      "loss": 3.1765,
+      "step": 698
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001634050880626223,
+      "loss": 3.1585,
+      "step": 699
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00016320939334637963,
+      "loss": 3.186,
+      "step": 700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00016301369863013697,
+      "loss": 3.1744,
+      "step": 701
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001628180039138943,
+      "loss": 3.1756,
+      "step": 702
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016262230919765166,
+      "loss": 3.1623,
+      "step": 703
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016242661448140897,
+      "loss": 3.1626,
+      "step": 704
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016223091976516634,
+      "loss": 3.1468,
+      "step": 705
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016203522504892366,
+      "loss": 3.1574,
+      "step": 706
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000161839530332681,
+      "loss": 3.1872,
+      "step": 707
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016164383561643835,
+      "loss": 3.1618,
+      "step": 708
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001614481409001957,
+      "loss": 3.1654,
+      "step": 709
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000161252446183953,
+      "loss": 3.1765,
+      "step": 710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016105675146771038,
+      "loss": 3.1534,
+      "step": 711
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001608610567514677,
+      "loss": 3.1434,
+      "step": 712
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016066536203522504,
+      "loss": 3.1704,
+      "step": 713
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016046966731898235,
+      "loss": 3.1692,
+      "step": 714
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016027397260273972,
+      "loss": 3.1699,
+      "step": 715
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016007827788649704,
+      "loss": 3.1624,
+      "step": 716
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001598825831702544,
+      "loss": 3.1482,
+      "step": 717
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00015968688845401172,
+      "loss": 3.1488,
+      "step": 718
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00015949119373776907,
+      "loss": 3.1766,
+      "step": 719
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00015929549902152638,
+      "loss": 3.1617,
+      "step": 720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00015909980430528375,
+      "loss": 3.1672,
+      "step": 721
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00015890410958904107,
+      "loss": 3.1481,
+      "step": 722
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00015870841487279844,
+      "loss": 3.1539,
+      "step": 723
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00015851272015655576,
+      "loss": 3.1513,
+      "step": 724
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001583170254403131,
+      "loss": 3.1677,
+      "step": 725
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00015812133072407042,
+      "loss": 3.1623,
+      "step": 726
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001579256360078278,
+      "loss": 3.1733,
+      "step": 727
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001577299412915851,
+      "loss": 3.1467,
+      "step": 728
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00015753424657534245,
+      "loss": 3.1513,
+      "step": 729
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001573385518590998,
+      "loss": 3.1497,
+      "step": 730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00015714285714285713,
+      "loss": 3.159,
+      "step": 731
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00015694716242661445,
+      "loss": 3.1563,
+      "step": 732
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00015675146771037182,
+      "loss": 3.1581,
+      "step": 733
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00015655577299412914,
+      "loss": 3.133,
+      "step": 734
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00015636007827788648,
+      "loss": 3.145,
+      "step": 735
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00015616438356164382,
+      "loss": 3.1565,
+      "step": 736
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00015596868884540117,
+      "loss": 3.1405,
+      "step": 737
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00015577299412915848,
+      "loss": 3.1518,
+      "step": 738
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00015557729941291585,
+      "loss": 3.1659,
+      "step": 739
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00015538160469667317,
+      "loss": 3.1323,
+      "step": 740
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001551859099804305,
+      "loss": 3.1682,
+      "step": 741
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00015499021526418785,
+      "loss": 3.1342,
+      "step": 742
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001547945205479452,
+      "loss": 3.1752,
+      "step": 743
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00015459882583170251,
+      "loss": 3.156,
+      "step": 744
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00015440313111545988,
+      "loss": 3.1592,
+      "step": 745
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001542074363992172,
+      "loss": 3.1566,
+      "step": 746
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00015401174168297454,
+      "loss": 3.1442,
+      "step": 747
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001538160469667319,
+      "loss": 3.1806,
+      "step": 748
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00015362035225048923,
+      "loss": 3.1548,
+      "step": 749
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00015342465753424655,
+      "loss": 3.1635,
+      "step": 750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00015322896281800392,
+      "loss": 3.1544,
+      "step": 751
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00015303326810176123,
+      "loss": 3.1547,
+      "step": 752
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00015283757338551858,
+      "loss": 3.1348,
+      "step": 753
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015264187866927592,
+      "loss": 3.1474,
+      "step": 754
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015244618395303326,
+      "loss": 3.16,
+      "step": 755
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015225048923679058,
+      "loss": 3.1442,
+      "step": 756
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015205479452054795,
+      "loss": 3.1437,
+      "step": 757
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015185909980430527,
+      "loss": 3.1491,
+      "step": 758
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001516634050880626,
+      "loss": 3.1727,
+      "step": 759
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015146771037181993,
+      "loss": 3.1545,
+      "step": 760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001512720156555773,
+      "loss": 3.1386,
+      "step": 761
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001510763209393346,
+      "loss": 3.128,
+      "step": 762
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015088062622309198,
+      "loss": 3.1668,
+      "step": 763
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001506849315068493,
+      "loss": 3.1344,
+      "step": 764
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015048923679060664,
+      "loss": 3.1301,
+      "step": 765
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015029354207436396,
+      "loss": 3.1607,
+      "step": 766
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015009784735812133,
+      "loss": 3.1406,
+      "step": 767
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014990215264187864,
+      "loss": 3.141,
+      "step": 768
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000149706457925636,
+      "loss": 3.1406,
+      "step": 769
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014951076320939333,
+      "loss": 3.1468,
+      "step": 770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014931506849315067,
+      "loss": 3.1326,
+      "step": 771
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000149119373776908,
+      "loss": 3.1391,
+      "step": 772
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014892367906066533,
+      "loss": 3.1678,
+      "step": 773
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014872798434442268,
+      "loss": 3.1543,
+      "step": 774
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014853228962818002,
+      "loss": 3.1469,
+      "step": 775
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014833659491193736,
+      "loss": 3.1584,
+      "step": 776
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001481409001956947,
+      "loss": 3.1335,
+      "step": 777
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014794520547945202,
+      "loss": 3.1468,
+      "step": 778
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014774951076320937,
+      "loss": 3.1406,
+      "step": 779
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001475538160469667,
+      "loss": 3.1297,
+      "step": 780
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014735812133072405,
+      "loss": 3.1374,
+      "step": 781
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001471624266144814,
+      "loss": 3.1366,
+      "step": 782
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001469667318982387,
+      "loss": 3.1248,
+      "step": 783
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014677103718199606,
+      "loss": 3.1382,
+      "step": 784
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001465753424657534,
+      "loss": 3.1226,
+      "step": 785
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00014637964774951074,
+      "loss": 3.1489,
+      "step": 786
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00014618395303326809,
+      "loss": 3.1356,
+      "step": 787
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00014598825831702543,
+      "loss": 3.129,
+      "step": 788
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00014579256360078275,
+      "loss": 3.1357,
+      "step": 789
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001455968688845401,
+      "loss": 3.1262,
+      "step": 790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00014540117416829743,
+      "loss": 3.1393,
+      "step": 791
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00014520547945205478,
+      "loss": 3.1368,
+      "step": 792
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00014500978473581212,
+      "loss": 3.1179,
+      "step": 793
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00014481409001956946,
+      "loss": 3.1203,
+      "step": 794
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00014461839530332678,
+      "loss": 3.1377,
+      "step": 795
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00014442270058708412,
+      "loss": 3.1352,
+      "step": 796
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00014422700587084146,
+      "loss": 3.1483,
+      "step": 797
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001440313111545988,
+      "loss": 3.1367,
+      "step": 798
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00014383561643835615,
+      "loss": 3.1331,
+      "step": 799
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001436399217221135,
+      "loss": 3.1125,
+      "step": 800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001434442270058708,
+      "loss": 3.1554,
+      "step": 801
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00014324853228962815,
+      "loss": 3.1296,
+      "step": 802
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001430528375733855,
+      "loss": 3.1257,
+      "step": 803
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00014285714285714284,
+      "loss": 3.1232,
+      "step": 804
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00014266144814090018,
+      "loss": 3.1271,
+      "step": 805
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00014246575342465753,
+      "loss": 3.1358,
+      "step": 806
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00014227005870841484,
+      "loss": 3.1306,
+      "step": 807
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001420743639921722,
+      "loss": 3.1541,
+      "step": 808
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00014187866927592953,
+      "loss": 3.1495,
+      "step": 809
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00014168297455968687,
+      "loss": 3.1358,
+      "step": 810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00014148727984344422,
+      "loss": 3.1495,
+      "step": 811
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00014129158512720153,
+      "loss": 3.1224,
+      "step": 812
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00014109589041095888,
+      "loss": 3.1549,
+      "step": 813
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00014090019569471622,
+      "loss": 3.1578,
+      "step": 814
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00014070450097847356,
+      "loss": 3.1453,
+      "step": 815
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001405088062622309,
+      "loss": 3.1538,
+      "step": 816
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00014031311154598825,
+      "loss": 3.139,
+      "step": 817
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00014011741682974557,
+      "loss": 3.1142,
+      "step": 818
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001399217221135029,
+      "loss": 3.1151,
+      "step": 819
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013972602739726025,
+      "loss": 3.1302,
+      "step": 820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001395303326810176,
+      "loss": 3.1555,
+      "step": 821
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013933463796477494,
+      "loss": 3.1429,
+      "step": 822
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013913894324853228,
+      "loss": 3.1318,
+      "step": 823
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001389432485322896,
+      "loss": 3.1529,
+      "step": 824
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013874755381604694,
+      "loss": 3.1258,
+      "step": 825
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013855185909980428,
+      "loss": 3.1296,
+      "step": 826
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013835616438356163,
+      "loss": 3.1461,
+      "step": 827
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013816046966731897,
+      "loss": 3.1311,
+      "step": 828
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013796477495107631,
+      "loss": 3.1395,
+      "step": 829
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013776908023483363,
+      "loss": 3.1253,
+      "step": 830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013757338551859097,
+      "loss": 3.1341,
+      "step": 831
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013737769080234832,
+      "loss": 3.1389,
+      "step": 832
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013718199608610566,
+      "loss": 3.1301,
+      "step": 833
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000136986301369863,
+      "loss": 3.125,
+      "step": 834
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013679060665362032,
+      "loss": 3.127,
+      "step": 835
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013659491193737766,
+      "loss": 3.1212,
+      "step": 836
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000136399217221135,
+      "loss": 3.1257,
+      "step": 837
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013620352250489235,
+      "loss": 3.1402,
+      "step": 838
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001360078277886497,
+      "loss": 3.1311,
+      "step": 839
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013581213307240704,
+      "loss": 3.1327,
+      "step": 840
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013561643835616435,
+      "loss": 3.1322,
+      "step": 841
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001354207436399217,
+      "loss": 3.1303,
+      "step": 842
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013522504892367904,
+      "loss": 3.1196,
+      "step": 843
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013502935420743638,
+      "loss": 3.1346,
+      "step": 844
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013483365949119373,
+      "loss": 3.1205,
+      "step": 845
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013463796477495107,
+      "loss": 3.1428,
+      "step": 846
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013444227005870839,
+      "loss": 3.1473,
+      "step": 847
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013424657534246573,
+      "loss": 3.1177,
+      "step": 848
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013405088062622307,
+      "loss": 3.128,
+      "step": 849
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013385518590998042,
+      "loss": 3.162,
+      "step": 850
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013365949119373776,
+      "loss": 3.1314,
+      "step": 851
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001334637964774951,
+      "loss": 3.1364,
+      "step": 852
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013326810176125242,
+      "loss": 3.1359,
+      "step": 853
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013307240704500976,
+      "loss": 3.1367,
+      "step": 854
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001328767123287671,
+      "loss": 3.1389,
+      "step": 855
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013268101761252445,
+      "loss": 3.1369,
+      "step": 856
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001324853228962818,
+      "loss": 3.1663,
+      "step": 857
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001322896281800391,
+      "loss": 3.1757,
+      "step": 858
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013209393346379645,
+      "loss": 3.1359,
+      "step": 859
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001318982387475538,
+      "loss": 3.1262,
+      "step": 860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013170254403131114,
+      "loss": 3.145,
+      "step": 861
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013150684931506848,
+      "loss": 3.1185,
+      "step": 862
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013131115459882582,
+      "loss": 3.1311,
+      "step": 863
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013111545988258314,
+      "loss": 3.1238,
+      "step": 864
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013091976516634048,
+      "loss": 3.1349,
+      "step": 865
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013072407045009783,
+      "loss": 3.1431,
+      "step": 866
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013052837573385517,
+      "loss": 3.1335,
+      "step": 867
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001303326810176125,
+      "loss": 3.1396,
+      "step": 868
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013013698630136986,
+      "loss": 3.1351,
+      "step": 869
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012994129158512717,
+      "loss": 3.1457,
+      "step": 870
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012974559686888452,
+      "loss": 3.124,
+      "step": 871
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012954990215264186,
+      "loss": 3.127,
+      "step": 872
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001293542074363992,
+      "loss": 3.1555,
+      "step": 873
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012915851272015655,
+      "loss": 3.1196,
+      "step": 874
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001289628180039139,
+      "loss": 3.1233,
+      "step": 875
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001287671232876712,
+      "loss": 3.1462,
+      "step": 876
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012857142857142855,
+      "loss": 3.1208,
+      "step": 877
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001283757338551859,
+      "loss": 3.1295,
+      "step": 878
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012818003913894324,
+      "loss": 3.1144,
+      "step": 879
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012798434442270058,
+      "loss": 3.1134,
+      "step": 880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001277886497064579,
+      "loss": 3.1273,
+      "step": 881
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012759295499021524,
+      "loss": 3.1219,
+      "step": 882
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012739726027397258,
+      "loss": 3.124,
+      "step": 883
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012720156555772992,
+      "loss": 3.1346,
+      "step": 884
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012700587084148727,
+      "loss": 3.1182,
+      "step": 885
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001268101761252446,
+      "loss": 3.1364,
+      "step": 886
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012661448140900193,
+      "loss": 3.1302,
+      "step": 887
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012641878669275927,
+      "loss": 3.1325,
+      "step": 888
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012622309197651661,
+      "loss": 3.1439,
+      "step": 889
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012602739726027396,
+      "loss": 3.1263,
+      "step": 890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001258317025440313,
+      "loss": 3.1286,
+      "step": 891
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012563600782778864,
+      "loss": 3.1035,
+      "step": 892
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012544031311154596,
+      "loss": 3.1113,
+      "step": 893
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001252446183953033,
+      "loss": 3.1172,
+      "step": 894
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012504892367906065,
+      "loss": 3.147,
+      "step": 895
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000124853228962818,
+      "loss": 3.1284,
+      "step": 896
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012465753424657533,
+      "loss": 3.1333,
+      "step": 897
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012446183953033268,
+      "loss": 3.1376,
+      "step": 898
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012426614481409,
+      "loss": 3.1115,
+      "step": 899
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012407045009784734,
+      "loss": 3.1494,
+      "step": 900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012387475538160468,
+      "loss": 3.1442,
+      "step": 901
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012367906066536202,
+      "loss": 3.1194,
+      "step": 902
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012348336594911937,
+      "loss": 3.1343,
+      "step": 903
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012328767123287668,
+      "loss": 3.1355,
+      "step": 904
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012309197651663403,
+      "loss": 3.1238,
+      "step": 905
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012289628180039137,
+      "loss": 3.111,
+      "step": 906
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001227005870841487,
+      "loss": 3.1149,
+      "step": 907
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012250489236790606,
+      "loss": 3.1266,
+      "step": 908
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001223091976516634,
+      "loss": 3.1254,
+      "step": 909
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012211350293542071,
+      "loss": 3.1104,
+      "step": 910
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012191780821917807,
+      "loss": 3.1247,
+      "step": 911
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001217221135029354,
+      "loss": 3.1289,
+      "step": 912
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012152641878669274,
+      "loss": 3.1161,
+      "step": 913
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012133072407045007,
+      "loss": 3.1296,
+      "step": 914
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012113502935420742,
+      "loss": 3.1252,
+      "step": 915
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012093933463796476,
+      "loss": 3.1422,
+      "step": 916
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012074363992172209,
+      "loss": 3.1013,
+      "step": 917
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012054794520547943,
+      "loss": 3.1349,
+      "step": 918
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012035225048923678,
+      "loss": 3.1256,
+      "step": 919
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001201565557729941,
+      "loss": 3.1272,
+      "step": 920
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011996086105675145,
+      "loss": 3.1389,
+      "step": 921
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001197651663405088,
+      "loss": 3.1222,
+      "step": 922
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011956947162426612,
+      "loss": 3.122,
+      "step": 923
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011937377690802347,
+      "loss": 3.1322,
+      "step": 924
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011917808219178081,
+      "loss": 3.1271,
+      "step": 925
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011898238747553814,
+      "loss": 3.126,
+      "step": 926
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011878669275929548,
+      "loss": 3.1209,
+      "step": 927
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011859099804305283,
+      "loss": 3.1359,
+      "step": 928
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011839530332681016,
+      "loss": 3.1275,
+      "step": 929
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001181996086105675,
+      "loss": 3.1173,
+      "step": 930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011800391389432484,
+      "loss": 3.1109,
+      "step": 931
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011780821917808217,
+      "loss": 3.1367,
+      "step": 932
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011761252446183952,
+      "loss": 3.1124,
+      "step": 933
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011741682974559686,
+      "loss": 3.1189,
+      "step": 934
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011722113502935419,
+      "loss": 3.1114,
+      "step": 935
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011702544031311153,
+      "loss": 3.1296,
+      "step": 936
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011682974559686886,
+      "loss": 3.1141,
+      "step": 937
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001166340508806262,
+      "loss": 3.1284,
+      "step": 938
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011643835616438355,
+      "loss": 3.1284,
+      "step": 939
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011624266144814088,
+      "loss": 3.1153,
+      "step": 940
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011604696673189822,
+      "loss": 3.1182,
+      "step": 941
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011585127201565556,
+      "loss": 3.1171,
+      "step": 942
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001156555772994129,
+      "loss": 3.1289,
+      "step": 943
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011545988258317024,
+      "loss": 3.1143,
+      "step": 944
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011526418786692758,
+      "loss": 3.1236,
+      "step": 945
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011506849315068491,
+      "loss": 3.1242,
+      "step": 946
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011487279843444225,
+      "loss": 3.1404,
+      "step": 947
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001146771037181996,
+      "loss": 3.1173,
+      "step": 948
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011448140900195693,
+      "loss": 3.1406,
+      "step": 949
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011428571428571427,
+      "loss": 3.1177,
+      "step": 950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011409001956947161,
+      "loss": 3.1103,
+      "step": 951
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011389432485322894,
+      "loss": 3.128,
+      "step": 952
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011369863013698629,
+      "loss": 3.1282,
+      "step": 953
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011350293542074363,
+      "loss": 3.1315,
+      "step": 954
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011330724070450096,
+      "loss": 3.1205,
+      "step": 955
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001131115459882583,
+      "loss": 3.1464,
+      "step": 956
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011291585127201565,
+      "loss": 3.135,
+      "step": 957
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011272015655577298,
+      "loss": 3.1292,
+      "step": 958
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011252446183953032,
+      "loss": 3.1267,
+      "step": 959
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011232876712328765,
+      "loss": 3.1131,
+      "step": 960
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011213307240704499,
+      "loss": 3.128,
+      "step": 961
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011193737769080234,
+      "loss": 3.1223,
+      "step": 962
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011174168297455966,
+      "loss": 3.1283,
+      "step": 963
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011154598825831701,
+      "loss": 3.1163,
+      "step": 964
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011135029354207435,
+      "loss": 3.1357,
+      "step": 965
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011115459882583168,
+      "loss": 3.1144,
+      "step": 966
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011095890410958902,
+      "loss": 3.1232,
+      "step": 967
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011076320939334637,
+      "loss": 3.1239,
+      "step": 968
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001105675146771037,
+      "loss": 3.1255,
+      "step": 969
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011037181996086104,
+      "loss": 3.0973,
+      "step": 970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011017612524461838,
+      "loss": 3.1072,
+      "step": 971
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010998043052837571,
+      "loss": 3.1245,
+      "step": 972
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010978473581213306,
+      "loss": 3.1314,
+      "step": 973
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001095890410958904,
+      "loss": 3.1118,
+      "step": 974
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010939334637964773,
+      "loss": 3.1211,
+      "step": 975
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010919765166340507,
+      "loss": 3.1143,
+      "step": 976
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010900195694716242,
+      "loss": 3.1095,
+      "step": 977
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010880626223091975,
+      "loss": 3.1202,
+      "step": 978
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010861056751467709,
+      "loss": 3.1172,
+      "step": 979
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010841487279843443,
+      "loss": 3.1,
+      "step": 980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010821917808219176,
+      "loss": 3.1008,
+      "step": 981
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001080234833659491,
+      "loss": 3.1302,
+      "step": 982
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010782778864970644,
+      "loss": 3.1089,
+      "step": 983
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010763209393346378,
+      "loss": 3.0935,
+      "step": 984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010743639921722112,
+      "loss": 3.1132,
+      "step": 985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010724070450097845,
+      "loss": 3.1217,
+      "step": 986
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001070450097847358,
+      "loss": 3.1082,
+      "step": 987
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010684931506849314,
+      "loss": 3.1156,
+      "step": 988
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010665362035225047,
+      "loss": 3.1278,
+      "step": 989
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010645792563600781,
+      "loss": 3.1095,
+      "step": 990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010626223091976516,
+      "loss": 3.1043,
+      "step": 991
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010606653620352248,
+      "loss": 3.1249,
+      "step": 992
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010587084148727983,
+      "loss": 3.1214,
+      "step": 993
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010567514677103717,
+      "loss": 3.1249,
+      "step": 994
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001054794520547945,
+      "loss": 3.1,
+      "step": 995
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010528375733855184,
+      "loss": 3.1339,
+      "step": 996
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010508806262230919,
+      "loss": 3.1316,
+      "step": 997
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010489236790606652,
+      "loss": 3.1167,
+      "step": 998
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010469667318982386,
+      "loss": 3.1114,
+      "step": 999
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001045009784735812,
+      "loss": 3.1149,
+      "step": 1000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010430528375733853,
+      "loss": 3.124,
+      "step": 1001
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010410958904109588,
+      "loss": 3.1216,
+      "step": 1002
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010391389432485322,
+      "loss": 3.1094,
+      "step": 1003
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010371819960861055,
+      "loss": 3.1256,
+      "step": 1004
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001035225048923679,
+      "loss": 3.1146,
+      "step": 1005
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010332681017612522,
+      "loss": 3.1392,
+      "step": 1006
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010313111545988257,
+      "loss": 3.106,
+      "step": 1007
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010293542074363991,
+      "loss": 3.1158,
+      "step": 1008
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010273972602739724,
+      "loss": 3.1201,
+      "step": 1009
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010254403131115458,
+      "loss": 3.131,
+      "step": 1010
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010234833659491193,
+      "loss": 3.1173,
+      "step": 1011
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010215264187866926,
+      "loss": 3.1042,
+      "step": 1012
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001019569471624266,
+      "loss": 3.1043,
+      "step": 1013
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010176125244618394,
+      "loss": 3.1047,
+      "step": 1014
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010156555772994127,
+      "loss": 3.1142,
+      "step": 1015
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010136986301369862,
+      "loss": 3.1218,
+      "step": 1016
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010117416829745596,
+      "loss": 3.1069,
+      "step": 1017
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010097847358121329,
+      "loss": 3.1048,
+      "step": 1018
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010078277886497063,
+      "loss": 3.1196,
+      "step": 1019
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010058708414872797,
+      "loss": 3.1062,
+      "step": 1020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001003913894324853,
+      "loss": 3.109,
+      "step": 1021
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010019569471624265,
+      "loss": 3.1097,
+      "step": 1022
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 11.071310043334961,
+      "eval_runtime": 512.791,
+      "eval_samples_per_second": 30.872,
+      "eval_steps_per_second": 0.07,
+      "eval_wer": 0.9999636220770339,
+      "step": 1022
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 4.6471,
+      "step": 1023
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.980430528375732e-05,
+      "loss": 3.1252,
+      "step": 1024
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.960861056751466e-05,
+      "loss": 3.0992,
+      "step": 1025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.941291585127201e-05,
+      "loss": 3.1093,
+      "step": 1026
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.921722113502934e-05,
+      "loss": 3.1178,
+      "step": 1027
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.902152641878668e-05,
+      "loss": 3.1203,
+      "step": 1028
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.882583170254402e-05,
+      "loss": 3.1096,
+      "step": 1029
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.863013698630135e-05,
+      "loss": 3.1118,
+      "step": 1030
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.84344422700587e-05,
+      "loss": 3.0968,
+      "step": 1031
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.823874755381603e-05,
+      "loss": 3.116,
+      "step": 1032
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.804305283757337e-05,
+      "loss": 3.0984,
+      "step": 1033
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.784735812133071e-05,
+      "loss": 3.1049,
+      "step": 1034
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.765166340508804e-05,
+      "loss": 3.0993,
+      "step": 1035
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.745596868884539e-05,
+      "loss": 3.1017,
+      "step": 1036
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.726027397260273e-05,
+      "loss": 3.1103,
+      "step": 1037
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.706457925636006e-05,
+      "loss": 3.1097,
+      "step": 1038
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.68688845401174e-05,
+      "loss": 3.1288,
+      "step": 1039
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.667318982387475e-05,
+      "loss": 3.118,
+      "step": 1040
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.647749510763208e-05,
+      "loss": 3.1166,
+      "step": 1041
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.628180039138942e-05,
+      "loss": 3.106,
+      "step": 1042
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.608610567514676e-05,
+      "loss": 3.0979,
+      "step": 1043
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.589041095890409e-05,
+      "loss": 3.1159,
+      "step": 1044
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.569471624266144e-05,
+      "loss": 3.1071,
+      "step": 1045
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.549902152641878e-05,
+      "loss": 3.0906,
+      "step": 1046
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.530332681017611e-05,
+      "loss": 3.1089,
+      "step": 1047
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.510763209393345e-05,
+      "loss": 3.1008,
+      "step": 1048
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.49119373776908e-05,
+      "loss": 3.1032,
+      "step": 1049
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.471624266144812e-05,
+      "loss": 3.0994,
+      "step": 1050
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.452054794520547e-05,
+      "loss": 3.093,
+      "step": 1051
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.432485322896281e-05,
+      "loss": 3.1114,
+      "step": 1052
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.412915851272014e-05,
+      "loss": 3.0827,
+      "step": 1053
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.393346379647748e-05,
+      "loss": 3.1041,
+      "step": 1054
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.373776908023481e-05,
+      "loss": 3.1068,
+      "step": 1055
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.354207436399216e-05,
+      "loss": 3.0941,
+      "step": 1056
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.33463796477495e-05,
+      "loss": 3.1066,
+      "step": 1057
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.315068493150683e-05,
+      "loss": 3.1036,
+      "step": 1058
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.295499021526417e-05,
+      "loss": 3.1035,
+      "step": 1059
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.275929549902152e-05,
+      "loss": 3.0722,
+      "step": 1060
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.256360078277885e-05,
+      "loss": 3.1077,
+      "step": 1061
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.236790606653619e-05,
+      "loss": 3.0914,
+      "step": 1062
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.217221135029353e-05,
+      "loss": 3.0951,
+      "step": 1063
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.197651663405086e-05,
+      "loss": 3.104,
+      "step": 1064
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.17808219178082e-05,
+      "loss": 3.1262,
+      "step": 1065
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.158512720156555e-05,
+      "loss": 3.0933,
+      "step": 1066
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.138943248532288e-05,
+      "loss": 3.1002,
+      "step": 1067
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.119373776908022e-05,
+      "loss": 3.1073,
+      "step": 1068
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.099804305283757e-05,
+      "loss": 3.0895,
+      "step": 1069
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.08023483365949e-05,
+      "loss": 3.0823,
+      "step": 1070
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.060665362035224e-05,
+      "loss": 3.0955,
+      "step": 1071
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.041095890410958e-05,
+      "loss": 3.1231,
+      "step": 1072
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.021526418786691e-05,
+      "loss": 3.1017,
+      "step": 1073
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.001956947162426e-05,
+      "loss": 3.103,
+      "step": 1074
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.98238747553816e-05,
+      "loss": 3.0829,
+      "step": 1075
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.962818003913893e-05,
+      "loss": 3.0942,
+      "step": 1076
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.943248532289627e-05,
+      "loss": 3.0983,
+      "step": 1077
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.92367906066536e-05,
+      "loss": 3.0937,
+      "step": 1078
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.904109589041094e-05,
+      "loss": 3.0885,
+      "step": 1079
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.884540117416829e-05,
+      "loss": 3.0858,
+      "step": 1080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.864970645792562e-05,
+      "loss": 3.121,
+      "step": 1081
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.845401174168296e-05,
+      "loss": 3.1038,
+      "step": 1082
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.82583170254403e-05,
+      "loss": 3.0919,
+      "step": 1083
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.806262230919763e-05,
+      "loss": 3.1125,
+      "step": 1084
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.786692759295498e-05,
+      "loss": 3.1011,
+      "step": 1085
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.767123287671232e-05,
+      "loss": 3.1193,
+      "step": 1086
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.747553816046965e-05,
+      "loss": 3.0789,
+      "step": 1087
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.7279843444227e-05,
+      "loss": 3.1006,
+      "step": 1088
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.708414872798434e-05,
+      "loss": 3.1077,
+      "step": 1089
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.688845401174167e-05,
+      "loss": 3.0825,
+      "step": 1090
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.669275929549901e-05,
+      "loss": 3.0824,
+      "step": 1091
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.649706457925635e-05,
+      "loss": 3.1055,
+      "step": 1092
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.630136986301368e-05,
+      "loss": 3.0984,
+      "step": 1093
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.610567514677103e-05,
+      "loss": 3.0951,
+      "step": 1094
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.590998043052837e-05,
+      "loss": 3.1181,
+      "step": 1095
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.57142857142857e-05,
+      "loss": 3.1022,
+      "step": 1096
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.551859099804304e-05,
+      "loss": 3.0843,
+      "step": 1097
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.532289628180039e-05,
+      "loss": 3.0949,
+      "step": 1098
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.512720156555772e-05,
+      "loss": 3.1284,
+      "step": 1099
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.493150684931506e-05,
+      "loss": 3.0959,
+      "step": 1100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.473581213307239e-05,
+      "loss": 3.0898,
+      "step": 1101
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.454011741682973e-05,
+      "loss": 3.0951,
+      "step": 1102
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.434442270058707e-05,
+      "loss": 3.1127,
+      "step": 1103
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.41487279843444e-05,
+      "loss": 3.0788,
+      "step": 1104
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.395303326810175e-05,
+      "loss": 3.0924,
+      "step": 1105
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.375733855185909e-05,
+      "loss": 3.0868,
+      "step": 1106
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.356164383561642e-05,
+      "loss": 3.0955,
+      "step": 1107
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.336594911937376e-05,
+      "loss": 3.0779,
+      "step": 1108
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.317025440313111e-05,
+      "loss": 3.0863,
+      "step": 1109
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.297455968688844e-05,
+      "loss": 3.0776,
+      "step": 1110
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.277886497064578e-05,
+      "loss": 3.0884,
+      "step": 1111
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.258317025440312e-05,
+      "loss": 3.0996,
+      "step": 1112
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.238747553816045e-05,
+      "loss": 3.0983,
+      "step": 1113
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.21917808219178e-05,
+      "loss": 3.078,
+      "step": 1114
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.199608610567514e-05,
+      "loss": 3.0982,
+      "step": 1115
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.180039138943247e-05,
+      "loss": 3.1144,
+      "step": 1116
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.160469667318981e-05,
+      "loss": 3.0973,
+      "step": 1117
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.140900195694716e-05,
+      "loss": 3.0844,
+      "step": 1118
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.121330724070449e-05,
+      "loss": 3.0951,
+      "step": 1119
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.101761252446183e-05,
+      "loss": 3.0945,
+      "step": 1120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.082191780821917e-05,
+      "loss": 3.0839,
+      "step": 1121
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.06262230919765e-05,
+      "loss": 3.0876,
+      "step": 1122
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.043052837573385e-05,
+      "loss": 3.0903,
+      "step": 1123
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.023483365949118e-05,
+      "loss": 3.0901,
+      "step": 1124
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.003913894324852e-05,
+      "loss": 3.0799,
+      "step": 1125
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.984344422700586e-05,
+      "loss": 3.0929,
+      "step": 1126
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.964774951076319e-05,
+      "loss": 3.1026,
+      "step": 1127
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.945205479452054e-05,
+      "loss": 3.0934,
+      "step": 1128
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.925636007827788e-05,
+      "loss": 3.0998,
+      "step": 1129
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.906066536203521e-05,
+      "loss": 3.0951,
+      "step": 1130
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.886497064579255e-05,
+      "loss": 3.1021,
+      "step": 1131
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.86692759295499e-05,
+      "loss": 3.0904,
+      "step": 1132
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.847358121330722e-05,
+      "loss": 3.0769,
+      "step": 1133
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.827788649706457e-05,
+      "loss": 3.0951,
+      "step": 1134
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.808219178082191e-05,
+      "loss": 3.0719,
+      "step": 1135
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.788649706457924e-05,
+      "loss": 3.0889,
+      "step": 1136
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.769080234833658e-05,
+      "loss": 3.0789,
+      "step": 1137
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.749510763209393e-05,
+      "loss": 3.0973,
+      "step": 1138
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.729941291585126e-05,
+      "loss": 3.0977,
+      "step": 1139
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.71037181996086e-05,
+      "loss": 3.0884,
+      "step": 1140
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.690802348336594e-05,
+      "loss": 3.0909,
+      "step": 1141
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.671232876712327e-05,
+      "loss": 3.0774,
+      "step": 1142
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.651663405088062e-05,
+      "loss": 3.0874,
+      "step": 1143
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.632093933463796e-05,
+      "loss": 3.0765,
+      "step": 1144
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.612524461839529e-05,
+      "loss": 3.0828,
+      "step": 1145
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.592954990215263e-05,
+      "loss": 3.0936,
+      "step": 1146
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.573385518590996e-05,
+      "loss": 3.0783,
+      "step": 1147
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.55381604696673e-05,
+      "loss": 3.0877,
+      "step": 1148
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.534246575342465e-05,
+      "loss": 3.0906,
+      "step": 1149
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.514677103718198e-05,
+      "loss": 3.0953,
+      "step": 1150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.495107632093932e-05,
+      "loss": 3.0831,
+      "step": 1151
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.475538160469667e-05,
+      "loss": 3.0823,
+      "step": 1152
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.4559686888454e-05,
+      "loss": 3.063,
+      "step": 1153
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.436399217221134e-05,
+      "loss": 3.1071,
+      "step": 1154
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.416829745596868e-05,
+      "loss": 3.0938,
+      "step": 1155
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.397260273972601e-05,
+      "loss": 3.1188,
+      "step": 1156
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.377690802348336e-05,
+      "loss": 3.0981,
+      "step": 1157
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.35812133072407e-05,
+      "loss": 3.0792,
+      "step": 1158
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.338551859099803e-05,
+      "loss": 3.0976,
+      "step": 1159
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.318982387475537e-05,
+      "loss": 3.0991,
+      "step": 1160
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.299412915851271e-05,
+      "loss": 3.1003,
+      "step": 1161
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.279843444227004e-05,
+      "loss": 3.0793,
+      "step": 1162
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.260273972602739e-05,
+      "loss": 3.0891,
+      "step": 1163
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.240704500978473e-05,
+      "loss": 3.0808,
+      "step": 1164
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.221135029354206e-05,
+      "loss": 3.08,
+      "step": 1165
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.20156555772994e-05,
+      "loss": 3.0784,
+      "step": 1166
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.181996086105675e-05,
+      "loss": 3.0652,
+      "step": 1167
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.162426614481408e-05,
+      "loss": 3.0888,
+      "step": 1168
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.142857142857142e-05,
+      "loss": 3.0916,
+      "step": 1169
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.123287671232876e-05,
+      "loss": 3.084,
+      "step": 1170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.10371819960861e-05,
+      "loss": 3.0874,
+      "step": 1171
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.084148727984344e-05,
+      "loss": 3.0831,
+      "step": 1172
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.064579256360077e-05,
+      "loss": 3.0965,
+      "step": 1173
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.045009784735811e-05,
+      "loss": 3.0963,
+      "step": 1174
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.025440313111545e-05,
+      "loss": 3.1057,
+      "step": 1175
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.005870841487278e-05,
+      "loss": 3.0931,
+      "step": 1176
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.986301369863013e-05,
+      "loss": 3.0869,
+      "step": 1177
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.966731898238747e-05,
+      "loss": 3.0825,
+      "step": 1178
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.94716242661448e-05,
+      "loss": 3.1187,
+      "step": 1179
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.927592954990214e-05,
+      "loss": 3.0852,
+      "step": 1180
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.908023483365949e-05,
+      "loss": 3.0945,
+      "step": 1181
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.888454011741682e-05,
+      "loss": 3.0845,
+      "step": 1182
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.868884540117416e-05,
+      "loss": 3.0884,
+      "step": 1183
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.84931506849315e-05,
+      "loss": 3.0952,
+      "step": 1184
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.829745596868883e-05,
+      "loss": 3.0746,
+      "step": 1185
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.810176125244617e-05,
+      "loss": 3.0781,
+      "step": 1186
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.790606653620352e-05,
+      "loss": 3.0805,
+      "step": 1187
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.771037181996085e-05,
+      "loss": 3.0887,
+      "step": 1188
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.751467710371819e-05,
+      "loss": 3.0738,
+      "step": 1189
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.731898238747553e-05,
+      "loss": 3.1072,
+      "step": 1190
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.712328767123286e-05,
+      "loss": 3.0933,
+      "step": 1191
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.692759295499021e-05,
+      "loss": 3.0806,
+      "step": 1192
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.673189823874755e-05,
+      "loss": 3.0914,
+      "step": 1193
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.653620352250488e-05,
+      "loss": 3.0833,
+      "step": 1194
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.634050880626222e-05,
+      "loss": 3.0953,
+      "step": 1195
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.614481409001955e-05,
+      "loss": 3.0787,
+      "step": 1196
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.59491193737769e-05,
+      "loss": 3.0927,
+      "step": 1197
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.575342465753424e-05,
+      "loss": 3.0778,
+      "step": 1198
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.555772994129157e-05,
+      "loss": 3.1037,
+      "step": 1199
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.536203522504891e-05,
+      "loss": 3.0948,
+      "step": 1200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.516634050880626e-05,
+      "loss": 3.0799,
+      "step": 1201
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.497064579256359e-05,
+      "loss": 3.0921,
+      "step": 1202
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.477495107632093e-05,
+      "loss": 3.0831,
+      "step": 1203
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.457925636007827e-05,
+      "loss": 3.0792,
+      "step": 1204
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.43835616438356e-05,
+      "loss": 3.0563,
+      "step": 1205
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.418786692759295e-05,
+      "loss": 3.0839,
+      "step": 1206
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.399217221135029e-05,
+      "loss": 3.074,
+      "step": 1207
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.379647749510762e-05,
+      "loss": 3.081,
+      "step": 1208
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.360078277886496e-05,
+      "loss": 3.094,
+      "step": 1209
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.34050880626223e-05,
+      "loss": 3.088,
+      "step": 1210
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.320939334637964e-05,
+      "loss": 3.0876,
+      "step": 1211
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.301369863013698e-05,
+      "loss": 3.0858,
+      "step": 1212
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.281800391389432e-05,
+      "loss": 3.0721,
+      "step": 1213
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.262230919765165e-05,
+      "loss": 3.0885,
+      "step": 1214
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.2426614481409e-05,
+      "loss": 3.0734,
+      "step": 1215
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.223091976516634e-05,
+      "loss": 3.0742,
+      "step": 1216
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.203522504892367e-05,
+      "loss": 3.0683,
+      "step": 1217
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.183953033268101e-05,
+      "loss": 3.0922,
+      "step": 1218
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.164383561643834e-05,
+      "loss": 3.0714,
+      "step": 1219
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.144814090019568e-05,
+      "loss": 3.0796,
+      "step": 1220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.125244618395303e-05,
+      "loss": 3.0699,
+      "step": 1221
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.105675146771036e-05,
+      "loss": 3.0954,
+      "step": 1222
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.08610567514677e-05,
+      "loss": 3.0947,
+      "step": 1223
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.066536203522504e-05,
+      "loss": 3.1089,
+      "step": 1224
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.046966731898238e-05,
+      "loss": 3.0642,
+      "step": 1225
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.027397260273972e-05,
+      "loss": 3.0729,
+      "step": 1226
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.007827788649705e-05,
+      "loss": 3.0939,
+      "step": 1227
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.98825831702544e-05,
+      "loss": 3.0843,
+      "step": 1228
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.968688845401173e-05,
+      "loss": 3.0808,
+      "step": 1229
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.949119373776907e-05,
+      "loss": 3.0813,
+      "step": 1230
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.929549902152641e-05,
+      "loss": 3.0939,
+      "step": 1231
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.909980430528375e-05,
+      "loss": 3.106,
+      "step": 1232
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.8904109589041086e-05,
+      "loss": 3.0975,
+      "step": 1233
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.870841487279843e-05,
+      "loss": 3.0838,
+      "step": 1234
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.8512720156555766e-05,
+      "loss": 3.0846,
+      "step": 1235
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.83170254403131e-05,
+      "loss": 3.0797,
+      "step": 1236
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.812133072407044e-05,
+      "loss": 3.0816,
+      "step": 1237
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.792563600782778e-05,
+      "loss": 3.0816,
+      "step": 1238
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.772994129158512e-05,
+      "loss": 3.0732,
+      "step": 1239
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.7534246575342455e-05,
+      "loss": 3.0737,
+      "step": 1240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.73385518590998e-05,
+      "loss": 3.065,
+      "step": 1241
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.7142857142857135e-05,
+      "loss": 3.0787,
+      "step": 1242
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.694716242661447e-05,
+      "loss": 3.0977,
+      "step": 1243
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.6751467710371815e-05,
+      "loss": 3.0921,
+      "step": 1244
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.655577299412915e-05,
+      "loss": 3.0748,
+      "step": 1245
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.636007827788649e-05,
+      "loss": 3.0966,
+      "step": 1246
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.6164383561643824e-05,
+      "loss": 3.0837,
+      "step": 1247
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.596868884540117e-05,
+      "loss": 3.087,
+      "step": 1248
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.5772994129158504e-05,
+      "loss": 3.082,
+      "step": 1249
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.557729941291584e-05,
+      "loss": 3.0773,
+      "step": 1250
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.5381604696673184e-05,
+      "loss": 3.0805,
+      "step": 1251
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.518590998043052e-05,
+      "loss": 3.0782,
+      "step": 1252
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.499021526418786e-05,
+      "loss": 3.0895,
+      "step": 1253
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.47945205479452e-05,
+      "loss": 3.0899,
+      "step": 1254
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.459882583170254e-05,
+      "loss": 3.0689,
+      "step": 1255
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.440313111545987e-05,
+      "loss": 3.0842,
+      "step": 1256
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.4207436399217217e-05,
+      "loss": 3.0907,
+      "step": 1257
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.401174168297455e-05,
+      "loss": 3.0873,
+      "step": 1258
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.381604696673189e-05,
+      "loss": 3.0894,
+      "step": 1259
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.3620352250489226e-05,
+      "loss": 3.0787,
+      "step": 1260
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.342465753424657e-05,
+      "loss": 3.0702,
+      "step": 1261
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.3228962818003906e-05,
+      "loss": 3.1048,
+      "step": 1262
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.303326810176124e-05,
+      "loss": 3.0777,
+      "step": 1263
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.2837573385518586e-05,
+      "loss": 3.0874,
+      "step": 1264
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.264187866927592e-05,
+      "loss": 3.0819,
+      "step": 1265
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.244618395303326e-05,
+      "loss": 3.0687,
+      "step": 1266
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.22504892367906e-05,
+      "loss": 3.0667,
+      "step": 1267
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.205479452054794e-05,
+      "loss": 3.0798,
+      "step": 1268
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.1859099804305275e-05,
+      "loss": 3.0666,
+      "step": 1269
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.166340508806261e-05,
+      "loss": 3.083,
+      "step": 1270
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.1467710371819955e-05,
+      "loss": 3.0885,
+      "step": 1271
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.127201565557729e-05,
+      "loss": 3.1091,
+      "step": 1272
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.107632093933463e-05,
+      "loss": 3.0811,
+      "step": 1273
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.088062622309197e-05,
+      "loss": 3.08,
+      "step": 1274
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.068493150684931e-05,
+      "loss": 3.0862,
+      "step": 1275
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.0489236790606644e-05,
+      "loss": 3.0712,
+      "step": 1276
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.029354207436399e-05,
+      "loss": 3.0791,
+      "step": 1277
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.0097847358121324e-05,
+      "loss": 3.0695,
+      "step": 1278
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.990215264187866e-05,
+      "loss": 3.0748,
+      "step": 1279
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.9706457925636004e-05,
+      "loss": 3.0672,
+      "step": 1280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.951076320939334e-05,
+      "loss": 3.0857,
+      "step": 1281
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.931506849315068e-05,
+      "loss": 3.0796,
+      "step": 1282
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.911937377690801e-05,
+      "loss": 3.0754,
+      "step": 1283
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.8923679060665357e-05,
+      "loss": 3.0985,
+      "step": 1284
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.872798434442269e-05,
+      "loss": 3.0685,
+      "step": 1285
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.853228962818003e-05,
+      "loss": 3.0901,
+      "step": 1286
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.833659491193737e-05,
+      "loss": 3.0864,
+      "step": 1287
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.814090019569471e-05,
+      "loss": 3.0777,
+      "step": 1288
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.7945205479452046e-05,
+      "loss": 3.0787,
+      "step": 1289
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.774951076320939e-05,
+      "loss": 3.0945,
+      "step": 1290
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.7553816046966726e-05,
+      "loss": 3.1029,
+      "step": 1291
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.735812133072406e-05,
+      "loss": 3.0755,
+      "step": 1292
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.7162426614481406e-05,
+      "loss": 3.0828,
+      "step": 1293
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.696673189823874e-05,
+      "loss": 3.0773,
+      "step": 1294
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.677103718199608e-05,
+      "loss": 3.0886,
+      "step": 1295
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.6575342465753415e-05,
+      "loss": 3.0616,
+      "step": 1296
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.637964774951076e-05,
+      "loss": 3.112,
+      "step": 1297
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.6183953033268095e-05,
+      "loss": 3.0566,
+      "step": 1298
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.598825831702543e-05,
+      "loss": 3.0837,
+      "step": 1299
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5792563600782775e-05,
+      "loss": 3.0935,
+      "step": 1300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.559686888454011e-05,
+      "loss": 3.0689,
+      "step": 1301
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.540117416829745e-05,
+      "loss": 3.0551,
+      "step": 1302
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.520547945205479e-05,
+      "loss": 3.076,
+      "step": 1303
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.500978473581213e-05,
+      "loss": 3.0757,
+      "step": 1304
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.4814090019569464e-05,
+      "loss": 3.0753,
+      "step": 1305
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.46183953033268e-05,
+      "loss": 3.0839,
+      "step": 1306
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.4422700587084144e-05,
+      "loss": 3.0787,
+      "step": 1307
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.422700587084148e-05,
+      "loss": 3.0667,
+      "step": 1308
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.403131115459882e-05,
+      "loss": 3.0868,
+      "step": 1309
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.383561643835616e-05,
+      "loss": 3.0668,
+      "step": 1310
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.36399217221135e-05,
+      "loss": 3.0727,
+      "step": 1311
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.344422700587083e-05,
+      "loss": 3.0775,
+      "step": 1312
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.3248532289628176e-05,
+      "loss": 3.1012,
+      "step": 1313
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.305283757338551e-05,
+      "loss": 3.0824,
+      "step": 1314
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.285714285714285e-05,
+      "loss": 3.081,
+      "step": 1315
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.266144814090019e-05,
+      "loss": 3.0902,
+      "step": 1316
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.246575342465753e-05,
+      "loss": 3.0717,
+      "step": 1317
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.2270058708414866e-05,
+      "loss": 3.1039,
+      "step": 1318
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.20743639921722e-05,
+      "loss": 3.0871,
+      "step": 1319
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.1878669275929546e-05,
+      "loss": 3.0765,
+      "step": 1320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.168297455968688e-05,
+      "loss": 3.0723,
+      "step": 1321
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.148727984344422e-05,
+      "loss": 3.0852,
+      "step": 1322
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.129158512720156e-05,
+      "loss": 3.0862,
+      "step": 1323
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.10958904109589e-05,
+      "loss": 3.0799,
+      "step": 1324
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.0900195694716235e-05,
+      "loss": 3.0794,
+      "step": 1325
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.070450097847358e-05,
+      "loss": 3.0801,
+      "step": 1326
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.0508806262230915e-05,
+      "loss": 3.0738,
+      "step": 1327
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.031311154598825e-05,
+      "loss": 3.0657,
+      "step": 1328
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.011741682974559e-05,
+      "loss": 3.0888,
+      "step": 1329
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.992172211350293e-05,
+      "loss": 3.0913,
+      "step": 1330
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.972602739726027e-05,
+      "loss": 3.0793,
+      "step": 1331
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.9530332681017604e-05,
+      "loss": 3.0803,
+      "step": 1332
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.933463796477495e-05,
+      "loss": 3.072,
+      "step": 1333
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.9138943248532284e-05,
+      "loss": 3.1062,
+      "step": 1334
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.894324853228962e-05,
+      "loss": 3.0677,
+      "step": 1335
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.8747553816046964e-05,
+      "loss": 3.0691,
+      "step": 1336
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.85518590998043e-05,
+      "loss": 3.0733,
+      "step": 1337
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.835616438356164e-05,
+      "loss": 3.0785,
+      "step": 1338
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.816046966731898e-05,
+      "loss": 3.0713,
+      "step": 1339
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7964774951076317e-05,
+      "loss": 3.0738,
+      "step": 1340
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.776908023483365e-05,
+      "loss": 3.0579,
+      "step": 1341
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.757338551859099e-05,
+      "loss": 3.0828,
+      "step": 1342
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.737769080234833e-05,
+      "loss": 3.0829,
+      "step": 1343
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.718199608610567e-05,
+      "loss": 3.0813,
+      "step": 1344
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6986301369863006e-05,
+      "loss": 3.0789,
+      "step": 1345
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.679060665362035e-05,
+      "loss": 3.0724,
+      "step": 1346
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6594911937377686e-05,
+      "loss": 3.0742,
+      "step": 1347
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.639921722113502e-05,
+      "loss": 3.07,
+      "step": 1348
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6203522504892366e-05,
+      "loss": 3.1018,
+      "step": 1349
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.60078277886497e-05,
+      "loss": 3.102,
+      "step": 1350
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.581213307240704e-05,
+      "loss": 3.0771,
+      "step": 1351
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.561643835616438e-05,
+      "loss": 3.0765,
+      "step": 1352
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.542074363992172e-05,
+      "loss": 3.086,
+      "step": 1353
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5225048923679055e-05,
+      "loss": 3.092,
+      "step": 1354
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.502935420743639e-05,
+      "loss": 3.0907,
+      "step": 1355
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4833659491193735e-05,
+      "loss": 3.0756,
+      "step": 1356
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.463796477495107e-05,
+      "loss": 3.0839,
+      "step": 1357
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.444227005870841e-05,
+      "loss": 3.0753,
+      "step": 1358
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.424657534246575e-05,
+      "loss": 3.0714,
+      "step": 1359
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.405088062622309e-05,
+      "loss": 3.0728,
+      "step": 1360
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.3855185909980424e-05,
+      "loss": 3.0777,
+      "step": 1361
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.365949119373777e-05,
+      "loss": 3.0667,
+      "step": 1362
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3463796477495104e-05,
+      "loss": 3.0796,
+      "step": 1363
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.326810176125244e-05,
+      "loss": 3.0698,
+      "step": 1364
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.307240704500978e-05,
+      "loss": 3.0719,
+      "step": 1365
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.287671232876712e-05,
+      "loss": 3.0711,
+      "step": 1366
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.268101761252446e-05,
+      "loss": 3.0869,
+      "step": 1367
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.248532289628179e-05,
+      "loss": 3.0937,
+      "step": 1368
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2289628180039136e-05,
+      "loss": 3.0653,
+      "step": 1369
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.209393346379647e-05,
+      "loss": 3.0683,
+      "step": 1370
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.189823874755381e-05,
+      "loss": 3.0893,
+      "step": 1371
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.170254403131115e-05,
+      "loss": 3.0647,
+      "step": 1372
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.150684931506849e-05,
+      "loss": 3.0788,
+      "step": 1373
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1311154598825826e-05,
+      "loss": 3.0813,
+      "step": 1374
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.111545988258317e-05,
+      "loss": 3.0601,
+      "step": 1375
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.0919765166340506e-05,
+      "loss": 3.0741,
+      "step": 1376
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.072407045009784e-05,
+      "loss": 3.0684,
+      "step": 1377
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.052837573385518e-05,
+      "loss": 3.0627,
+      "step": 1378
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.033268101761252e-05,
+      "loss": 3.0901,
+      "step": 1379
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.013698630136986e-05,
+      "loss": 3.0768,
+      "step": 1380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.99412915851272e-05,
+      "loss": 3.0755,
+      "step": 1381
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.9745596868884535e-05,
+      "loss": 3.0837,
+      "step": 1382
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9549902152641875e-05,
+      "loss": 3.0699,
+      "step": 1383
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9354207436399215e-05,
+      "loss": 3.0797,
+      "step": 1384
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.915851272015655e-05,
+      "loss": 3.0603,
+      "step": 1385
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.896281800391389e-05,
+      "loss": 3.0857,
+      "step": 1386
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.8767123287671228e-05,
+      "loss": 3.0896,
+      "step": 1387
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8571428571428567e-05,
+      "loss": 3.0757,
+      "step": 1388
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8375733855185907e-05,
+      "loss": 3.0854,
+      "step": 1389
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8180039138943244e-05,
+      "loss": 3.097,
+      "step": 1390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.7984344422700584e-05,
+      "loss": 3.0782,
+      "step": 1391
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.778864970645792e-05,
+      "loss": 3.0538,
+      "step": 1392
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.759295499021526e-05,
+      "loss": 3.061,
+      "step": 1393
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.73972602739726e-05,
+      "loss": 3.0633,
+      "step": 1394
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7201565557729937e-05,
+      "loss": 3.0649,
+      "step": 1395
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7005870841487277e-05,
+      "loss": 3.0662,
+      "step": 1396
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.6810176125244613e-05,
+      "loss": 3.0493,
+      "step": 1397
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6614481409001953e-05,
+      "loss": 3.0856,
+      "step": 1398
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6418786692759293e-05,
+      "loss": 3.0655,
+      "step": 1399
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.622309197651663e-05,
+      "loss": 3.0681,
+      "step": 1400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.602739726027397e-05,
+      "loss": 3.0607,
+      "step": 1401
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5831702544031306e-05,
+      "loss": 3.0934,
+      "step": 1402
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5636007827788646e-05,
+      "loss": 3.075,
+      "step": 1403
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5440313111545986e-05,
+      "loss": 3.0954,
+      "step": 1404
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5244618395303322e-05,
+      "loss": 3.0619,
+      "step": 1405
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5048923679060662e-05,
+      "loss": 3.0723,
+      "step": 1406
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4853228962818002e-05,
+      "loss": 3.0663,
+      "step": 1407
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.465753424657534e-05,
+      "loss": 3.0554,
+      "step": 1408
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4461839530332678e-05,
+      "loss": 3.0691,
+      "step": 1409
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4266144814090015e-05,
+      "loss": 3.0708,
+      "step": 1410
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4070450097847355e-05,
+      "loss": 3.0729,
+      "step": 1411
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.3874755381604695e-05,
+      "loss": 3.0599,
+      "step": 1412
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.367906066536203e-05,
+      "loss": 3.0671,
+      "step": 1413
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.348336594911937e-05,
+      "loss": 3.0763,
+      "step": 1414
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3287671232876708e-05,
+      "loss": 3.0725,
+      "step": 1415
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3091976516634047e-05,
+      "loss": 3.0855,
+      "step": 1416
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2896281800391387e-05,
+      "loss": 3.0451,
+      "step": 1417
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2700587084148724e-05,
+      "loss": 3.0603,
+      "step": 1418
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2504892367906064e-05,
+      "loss": 3.0644,
+      "step": 1419
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.23091976516634e-05,
+      "loss": 3.0869,
+      "step": 1420
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.211350293542074e-05,
+      "loss": 3.0638,
+      "step": 1421
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.191780821917808e-05,
+      "loss": 3.0608,
+      "step": 1422
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1722113502935417e-05,
+      "loss": 3.0477,
+      "step": 1423
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1526418786692757e-05,
+      "loss": 3.0687,
+      "step": 1424
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1330724070450096e-05,
+      "loss": 3.0685,
+      "step": 1425
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1135029354207433e-05,
+      "loss": 3.0634,
+      "step": 1426
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.0939334637964773e-05,
+      "loss": 3.0752,
+      "step": 1427
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.074363992172211e-05,
+      "loss": 3.0728,
+      "step": 1428
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.054794520547945e-05,
+      "loss": 3.0551,
+      "step": 1429
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.035225048923679e-05,
+      "loss": 3.063,
+      "step": 1430
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0156555772994126e-05,
+      "loss": 3.0545,
+      "step": 1431
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9960861056751466e-05,
+      "loss": 3.0742,
+      "step": 1432
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9765166340508802e-05,
+      "loss": 3.0602,
+      "step": 1433
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9569471624266142e-05,
+      "loss": 3.0536,
+      "step": 1434
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9373776908023482e-05,
+      "loss": 3.084,
+      "step": 1435
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.917808219178082e-05,
+      "loss": 3.0582,
+      "step": 1436
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8982387475538158e-05,
+      "loss": 3.0698,
+      "step": 1437
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8786692759295495e-05,
+      "loss": 3.053,
+      "step": 1438
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8590998043052835e-05,
+      "loss": 3.0486,
+      "step": 1439
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8395303326810175e-05,
+      "loss": 3.0623,
+      "step": 1440
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.819960861056751e-05,
+      "loss": 3.0892,
+      "step": 1441
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.800391389432485e-05,
+      "loss": 3.0567,
+      "step": 1442
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.780821917808219e-05,
+      "loss": 3.064,
+      "step": 1443
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7612524461839527e-05,
+      "loss": 3.0547,
+      "step": 1444
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7416829745596867e-05,
+      "loss": 3.0642,
+      "step": 1445
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7221135029354204e-05,
+      "loss": 3.0613,
+      "step": 1446
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7025440313111544e-05,
+      "loss": 3.0615,
+      "step": 1447
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6829745596868884e-05,
+      "loss": 3.0612,
+      "step": 1448
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.663405088062622e-05,
+      "loss": 3.0518,
+      "step": 1449
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.643835616438356e-05,
+      "loss": 3.0547,
+      "step": 1450
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6242661448140897e-05,
+      "loss": 3.049,
+      "step": 1451
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6046966731898236e-05,
+      "loss": 3.0518,
+      "step": 1452
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5851272015655576e-05,
+      "loss": 3.0503,
+      "step": 1453
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5655577299412913e-05,
+      "loss": 3.0655,
+      "step": 1454
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5459882583170253e-05,
+      "loss": 3.0711,
+      "step": 1455
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.526418786692759e-05,
+      "loss": 3.062,
+      "step": 1456
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.506849315068493e-05,
+      "loss": 3.0557,
+      "step": 1457
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4872798434442267e-05,
+      "loss": 3.0654,
+      "step": 1458
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4677103718199607e-05,
+      "loss": 3.0582,
+      "step": 1459
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4481409001956946e-05,
+      "loss": 3.0748,
+      "step": 1460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4285714285714284e-05,
+      "loss": 3.0422,
+      "step": 1461
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4090019569471622e-05,
+      "loss": 3.0655,
+      "step": 1462
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.389432485322896e-05,
+      "loss": 3.0531,
+      "step": 1463
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.36986301369863e-05,
+      "loss": 3.0632,
+      "step": 1464
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3502935420743638e-05,
+      "loss": 3.0606,
+      "step": 1465
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3307240704500976e-05,
+      "loss": 3.0665,
+      "step": 1466
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3111545988258315e-05,
+      "loss": 3.0723,
+      "step": 1467
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.2915851272015653e-05,
+      "loss": 3.0579,
+      "step": 1468
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.2720156555772993e-05,
+      "loss": 3.0634,
+      "step": 1469
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2524461839530331e-05,
+      "loss": 3.0561,
+      "step": 1470
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.232876712328767e-05,
+      "loss": 3.0639,
+      "step": 1471
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2133072407045007e-05,
+      "loss": 3.0578,
+      "step": 1472
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1937377690802347e-05,
+      "loss": 3.0509,
+      "step": 1473
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1741682974559686e-05,
+      "loss": 3.0622,
+      "step": 1474
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1545988258317024e-05,
+      "loss": 3.0732,
+      "step": 1475
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1350293542074362e-05,
+      "loss": 3.0526,
+      "step": 1476
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.11545988258317e-05,
+      "loss": 3.0411,
+      "step": 1477
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.095890410958904e-05,
+      "loss": 3.0522,
+      "step": 1478
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.0763209393346378e-05,
+      "loss": 3.0538,
+      "step": 1479
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0567514677103716e-05,
+      "loss": 3.0508,
+      "step": 1480
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0371819960861055e-05,
+      "loss": 3.0542,
+      "step": 1481
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0176125244618395e-05,
+      "loss": 3.0544,
+      "step": 1482
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.980430528375733e-06,
+      "loss": 3.0592,
+      "step": 1483
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.784735812133071e-06,
+      "loss": 3.0523,
+      "step": 1484
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.58904109589041e-06,
+      "loss": 3.0467,
+      "step": 1485
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.393346379647747e-06,
+      "loss": 3.0587,
+      "step": 1486
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.197651663405087e-06,
+      "loss": 3.0535,
+      "step": 1487
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.001956947162426e-06,
+      "loss": 3.0468,
+      "step": 1488
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.806262230919764e-06,
+      "loss": 3.0594,
+      "step": 1489
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.610567514677102e-06,
+      "loss": 3.0614,
+      "step": 1490
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.414872798434442e-06,
+      "loss": 3.0695,
+      "step": 1491
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.21917808219178e-06,
+      "loss": 3.0522,
+      "step": 1492
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.023483365949118e-06,
+      "loss": 3.0282,
+      "step": 1493
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.827788649706456e-06,
+      "loss": 3.0886,
+      "step": 1494
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.632093933463795e-06,
+      "loss": 3.0477,
+      "step": 1495
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.436399217221134e-06,
+      "loss": 3.0401,
+      "step": 1496
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.240704500978473e-06,
+      "loss": 3.0552,
+      "step": 1497
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.045009784735811e-06,
+      "loss": 3.0469,
+      "step": 1498
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.84931506849315e-06,
+      "loss": 3.043,
+      "step": 1499
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.653620352250488e-06,
+      "loss": 3.0467,
+      "step": 1500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.4579256360078264e-06,
+      "loss": 3.0634,
+      "step": 1501
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.2622309197651655e-06,
+      "loss": 3.0546,
+      "step": 1502
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.066536203522504e-06,
+      "loss": 3.0741,
+      "step": 1503
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.870841487279843e-06,
+      "loss": 3.0711,
+      "step": 1504
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.675146771037181e-06,
+      "loss": 3.0573,
+      "step": 1505
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.47945205479452e-06,
+      "loss": 3.0528,
+      "step": 1506
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.283757338551858e-06,
+      "loss": 3.0352,
+      "step": 1507
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.088062622309197e-06,
+      "loss": 3.0307,
+      "step": 1508
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.8923679060665355e-06,
+      "loss": 3.0397,
+      "step": 1509
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.696673189823874e-06,
+      "loss": 3.0631,
+      "step": 1510
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.500978473581213e-06,
+      "loss": 3.0502,
+      "step": 1511
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.305283757338551e-06,
+      "loss": 3.0575,
+      "step": 1512
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.10958904109589e-06,
+      "loss": 3.0654,
+      "step": 1513
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.913894324853228e-06,
+      "loss": 3.0512,
+      "step": 1514
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.718199608610567e-06,
+      "loss": 3.0493,
+      "step": 1515
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5225048923679055e-06,
+      "loss": 3.0607,
+      "step": 1516
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.326810176125244e-06,
+      "loss": 3.0561,
+      "step": 1517
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.1311154598825827e-06,
+      "loss": 3.0601,
+      "step": 1518
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9354207436399214e-06,
+      "loss": 3.0527,
+      "step": 1519
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.73972602739726e-06,
+      "loss": 3.0516,
+      "step": 1520
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5440313111545986e-06,
+      "loss": 3.0704,
+      "step": 1521
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.348336594911937e-06,
+      "loss": 3.0557,
+      "step": 1522
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1526418786692755e-06,
+      "loss": 3.0465,
+      "step": 1523
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.956947162426614e-06,
+      "loss": 3.06,
+      "step": 1524
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7612524461839527e-06,
+      "loss": 3.0631,
+      "step": 1525
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5655577299412914e-06,
+      "loss": 3.0715,
+      "step": 1526
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.36986301369863e-06,
+      "loss": 3.0547,
+      "step": 1527
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1741682974559684e-06,
+      "loss": 3.0642,
+      "step": 1528
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.78473581213307e-07,
+      "loss": 3.0562,
+      "step": 1529
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.827788649706457e-07,
+      "loss": 3.0795,
+      "step": 1530
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.870841487279842e-07,
+      "loss": 3.0513,
+      "step": 1531
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.9138943248532284e-07,
+      "loss": 3.0782,
+      "step": 1532
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9569471624266142e-07,
+      "loss": 3.0541,
+      "step": 1533
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 14.038721084594727,
+      "eval_runtime": 520.0539,
+      "eval_samples_per_second": 30.441,
+      "eval_steps_per_second": 0.069,
+      "eval_wer": 0.9999636220770339,
+      "step": 1533
+    },
+    {
+      "epoch": 3.0,
+      "step": 1533,
+      "total_flos": 4.180774836907213e+19,
+      "train_loss": 3.1717406544744464,
+      "train_runtime": 43485.0285,
+      "train_samples_per_second": 31.616,
+      "train_steps_per_second": 0.035
+    }
+  ],
+  "max_steps": 1533,
+  "num_train_epochs": 3,
+  "total_flos": 4.180774836907213e+19,
+  "trial_name": null,
+  "trial_params": null
+}