diff --git "a/checkpoint-2000/trainer_state.json" "b/checkpoint-2000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2000/trainer_state.json"
@@ -0,0 +1,12019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9995002498750625,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.205,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.1741,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 2.3915,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.2188,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 2.2271,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 2.1674,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 2.0138,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.8433,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 2.0383,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 2.1195,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 2.1695,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-05,
+      "loss": 1.8903,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.729,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 2.0379,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-05,
+      "loss": 1.9539,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 2.0113,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.6799,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-05,
+      "loss": 2.0281,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6e-05,
+      "loss": 1.8322,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-05,
+      "loss": 1.8084,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4e-05,
+      "loss": 1.9343,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.9195,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.8495,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6e-05,
+      "loss": 1.8587,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001,
+      "loss": 1.814,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 1.824,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 1.7464,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 1.8182,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000116,
+      "loss": 1.6595,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012,
+      "loss": 1.8015,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000124,
+      "loss": 1.825,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.8051,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000132,
+      "loss": 1.7185,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 1.7766,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014,
+      "loss": 1.7529,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000144,
+      "loss": 1.6643,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000148,
+      "loss": 1.8265,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000152,
+      "loss": 1.6598,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 1.5992,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016,
+      "loss": 1.9037,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000164,
+      "loss": 1.7461,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000168,
+      "loss": 1.7369,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000172,
+      "loss": 1.78,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.8073,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018,
+      "loss": 1.6786,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 1.8868,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000188,
+      "loss": 1.9341,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000192,
+      "loss": 1.5922,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000196,
+      "loss": 1.8224,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.8133,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 1.8373,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00020800000000000001,
+      "loss": 1.5649,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00021200000000000003,
+      "loss": 1.7079,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00021600000000000002,
+      "loss": 1.7249,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00022000000000000003,
+      "loss": 1.7299,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00022400000000000002,
+      "loss": 1.7068,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 1.8436,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000232,
+      "loss": 1.7897,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000236,
+      "loss": 1.8142,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00024,
+      "loss": 1.8176,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000244,
+      "loss": 1.7742,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000248,
+      "loss": 1.6512,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000252,
+      "loss": 1.806,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00025600000000000004,
+      "loss": 1.8776,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 1.8309,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000264,
+      "loss": 1.7217,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000268,
+      "loss": 1.7578,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 1.7367,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000276,
+      "loss": 1.4613,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00028,
+      "loss": 1.85,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000284,
+      "loss": 1.7672,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000288,
+      "loss": 1.6758,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000292,
+      "loss": 1.6693,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000296,
+      "loss": 1.6831,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 1.7338,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000304,
+      "loss": 1.6858,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000308,
+      "loss": 1.7453,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00031200000000000005,
+      "loss": 1.6362,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00031600000000000004,
+      "loss": 1.875,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00032,
+      "loss": 1.7411,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000324,
+      "loss": 1.7638,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000328,
+      "loss": 1.7488,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000332,
+      "loss": 1.5624,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000336,
+      "loss": 1.8976,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00034,
+      "loss": 1.7016,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000344,
+      "loss": 1.6986,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000348,
+      "loss": 1.8508,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00035200000000000005,
+      "loss": 1.7049,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00035600000000000003,
+      "loss": 1.7912,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00036,
+      "loss": 1.6776,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000364,
+      "loss": 1.7725,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00036800000000000005,
+      "loss": 1.734,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00037200000000000004,
+      "loss": 1.8326,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000376,
+      "loss": 1.7755,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00038,
+      "loss": 1.5972,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000384,
+      "loss": 1.7307,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000388,
+      "loss": 1.7856,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000392,
+      "loss": 1.7529,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 1.5194,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0004,
+      "loss": 1.7358,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039999972689137685,
+      "loss": 1.8458,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039999890756625326,
+      "loss": 1.7214,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003999975420268669,
+      "loss": 1.9563,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003999956302769471,
+      "loss": 1.9295,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003999931723217151,
+      "loss": 1.7522,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003999901681678838,
+      "loss": 1.6478,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039998661782365765,
+      "loss": 1.7303,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039998252129873314,
+      "loss": 1.7452,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039997787860429813,
+      "loss": 1.6586,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003999726897530322,
+      "loss": 1.5898,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003999669547591067,
+      "loss": 1.5924,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003999606736381842,
+      "loss": 1.5336,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000399953846407419,
+      "loss": 1.6407,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003999464730854571,
+      "loss": 1.5072,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039993855369243534,
+      "loss": 1.8297,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039993008824998246,
+      "loss": 1.8099,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003999210767812183,
+      "loss": 1.5861,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003999115193107539,
+      "loss": 1.7694,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003999014158646916,
+      "loss": 1.8015,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039989076647062473,
+      "loss": 1.7514,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003998795711576378,
+      "loss": 1.8225,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039986782995630603,
+      "loss": 1.7047,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039985554289869574,
+      "loss": 1.5168,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039984271001836395,
+      "loss": 1.8128,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003998293313503583,
+      "loss": 1.8507,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039981540693121716,
+      "loss": 1.6675,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003998009367989693,
+      "loss": 1.6252,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039978592099313386,
+      "loss": 1.75,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039977035955472034,
+      "loss": 1.4685,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003997542525262284,
+      "loss": 1.6629,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003997375999516476,
+      "loss": 1.8121,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003997204018764577,
+      "loss": 1.8056,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000399702658347628,
+      "loss": 1.6775,
+      "step": 133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039968436941361773,
+      "loss": 1.6369,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003996655351243755,
+      "loss": 1.6055,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003996461555313393,
+      "loss": 1.636,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003996262306874366,
+      "loss": 1.739,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003996057606470837,
+      "loss": 1.561,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039958474546618626,
+      "loss": 1.6892,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039956318520213837,
+      "loss": 1.8357,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003995410799138231,
+      "loss": 1.8147,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039951842966161176,
+      "loss": 1.6856,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003994952345073643,
+      "loss": 1.6191,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003994714945144286,
+      "loss": 1.5254,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003994472097476406,
+      "loss": 1.5864,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003994223802733241,
+      "loss": 1.7604,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003993970061592906,
+      "loss": 1.7338,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039937108747483893,
+      "loss": 1.8078,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003993446242907553,
+      "loss": 1.6927,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039931761667931287,
+      "loss": 1.6732,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039929006471427187,
+      "loss": 1.5369,
+      "step": 151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039926196847087905,
+      "loss": 1.5529,
+      "step": 152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003992333280258676,
+      "loss": 1.6886,
+      "step": 153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003992041434574572,
+      "loss": 1.7383,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003991744148453532,
+      "loss": 1.7943,
+      "step": 155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003991441422707472,
+      "loss": 1.6811,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039911332581631613,
+      "loss": 1.7442,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003990819655662224,
+      "loss": 1.5712,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039905006160611357,
+      "loss": 1.5422,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039901761402312205,
+      "loss": 1.6689,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000398984622905865,
+      "loss": 1.8412,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039895108834444405,
+      "loss": 1.7318,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039891701043044496,
+      "loss": 1.4819,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003988823892569375,
+      "loss": 1.4612,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039884722491847504,
+      "loss": 1.5737,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003988115175110943,
+      "loss": 1.7666,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003987752671323155,
+      "loss": 1.4879,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003987384738811414,
+      "loss": 1.7151,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003987011378580576,
+      "loss": 1.751,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039866325916503205,
+      "loss": 1.7216,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003986248379055146,
+      "loss": 1.5863,
+      "step": 171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039858587418443715,
+      "loss": 1.793,
+      "step": 172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039854636810821286,
+      "loss": 1.6011,
+      "step": 173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003985063197847363,
+      "loss": 1.7418,
+      "step": 174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003984657293233829,
+      "loss": 1.6862,
+      "step": 175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003984245968350087,
+      "loss": 1.6857,
+      "step": 176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039838292243195013,
+      "loss": 1.7442,
+      "step": 177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003983407062280234,
+      "loss": 1.9183,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003982979483385249,
+      "loss": 1.592,
+      "step": 179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003982546488802298,
+      "loss": 1.777,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039821080797139283,
+      "loss": 1.7078,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003981664257317472,
+      "loss": 1.7803,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039812150228250474,
+      "loss": 1.7864,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003980760377463552,
+      "loss": 1.7185,
+      "step": 184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003980300322474662,
+      "loss": 1.7885,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039798348591148263,
+      "loss": 1.8994,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039793639886552665,
+      "loss": 1.713,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000397888771238197,
+      "loss": 1.5899,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003978406031595688,
+      "loss": 1.6652,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003977918947611932,
+      "loss": 1.5935,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003977426461760972,
+      "loss": 1.6655,
+      "step": 191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039769285753878277,
+      "loss": 1.8054,
+      "step": 192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000397642528985227,
+      "loss": 1.6275,
+      "step": 193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039759166065288167,
+      "loss": 1.7446,
+      "step": 194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039754025268067233,
+      "loss": 1.6488,
+      "step": 195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039748830520899874,
+      "loss": 1.7123,
+      "step": 196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003974358183797339,
+      "loss": 1.6823,
+      "step": 197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039738279233622386,
+      "loss": 1.4115,
+      "step": 198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039732922722328725,
+      "loss": 1.3705,
+      "step": 199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039727512318721514,
+      "loss": 1.776,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039722048037577024,
+      "loss": 1.7096,
+      "step": 201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003971652989381868,
+      "loss": 1.5852,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003971095790251701,
+      "loss": 1.8677,
+      "step": 203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039705332078889596,
+      "loss": 1.4359,
+      "step": 204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039699652438301053,
+      "loss": 1.7743,
+      "step": 205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039693918996262974,
+      "loss": 1.6298,
+      "step": 206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003968813176843387,
+      "loss": 1.7088,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003968229077061917,
+      "loss": 1.4698,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039676396018771147,
+      "loss": 1.6022,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003967044752898886,
+      "loss": 1.6729,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003966444531751816,
+      "loss": 1.5742,
+      "step": 211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000396583894007516,
+      "loss": 1.437,
+      "step": 212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003965227979522842,
+      "loss": 1.615,
+      "step": 213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039646116517634463,
+      "loss": 1.6582,
+      "step": 214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039639899584802184,
+      "loss": 1.5987,
+      "step": 215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039633629013710554,
+      "loss": 1.8452,
+      "step": 216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039627304821485056,
+      "loss": 1.6249,
+      "step": 217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003962092702539759,
+      "loss": 1.7399,
+      "step": 218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003961449564286648,
+      "loss": 1.5969,
+      "step": 219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039608010691456367,
+      "loss": 1.5766,
+      "step": 220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039601472188878235,
+      "loss": 1.6199,
+      "step": 221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003959488015298929,
+      "loss": 1.773,
+      "step": 222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039588234601792944,
+      "loss": 1.8759,
+      "step": 223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003958153555343878,
+      "loss": 1.7133,
+      "step": 224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039574783026222475,
+      "loss": 1.5905,
+      "step": 225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039567977038585756,
+      "loss": 1.7087,
+      "step": 226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003956111760911637,
+      "loss": 1.5135,
+      "step": 227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039554204756548,
+      "loss": 1.4835,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039547238499760255,
+      "loss": 1.7274,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039540218857778576,
+      "loss": 1.568,
+      "step": 230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003953314584977421,
+      "loss": 1.5526,
+      "step": 231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039526019495064155,
+      "loss": 1.7895,
+      "step": 232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000395188398131111,
+      "loss": 1.6583,
+      "step": 233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039511606823523375,
+      "loss": 1.7053,
+      "step": 234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039504320546054894,
+      "loss": 1.709,
+      "step": 235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039496981000605117,
+      "loss": 1.6779,
+      "step": 236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003948958820721897,
+      "loss": 1.6184,
+      "step": 237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003948214218608681,
+      "loss": 1.6131,
+      "step": 238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039474642957544365,
+      "loss": 1.6341,
+      "step": 239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003946709054207267,
+      "loss": 1.6489,
+      "step": 240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039459484960298026,
+      "loss": 1.6761,
+      "step": 241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039451826232991935,
+      "loss": 1.5096,
+      "step": 242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003944411438107104,
+      "loss": 1.736,
+      "step": 243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003943634942559708,
+      "loss": 1.7403,
+      "step": 244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039428531387776804,
+      "loss": 1.554,
+      "step": 245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003942066028896195,
+      "loss": 1.8338,
+      "step": 246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003941273615064918,
+      "loss": 1.7811,
+      "step": 247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039404758994479984,
+      "loss": 1.6816,
+      "step": 248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039396728842240673,
+      "loss": 1.6188,
+      "step": 249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003938864571586229,
+      "loss": 1.6212,
+      "step": 250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039380509637420533,
+      "loss": 1.8131,
+      "step": 251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003937232062913575,
+      "loss": 1.7621,
+      "step": 252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039364078713372816,
+      "loss": 1.7201,
+      "step": 253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039355783912641126,
+      "loss": 1.6273,
+      "step": 254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003934743624959449,
+      "loss": 1.812,
+      "step": 255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003933903574703109,
+      "loss": 1.7049,
+      "step": 256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003933058242789344,
+      "loss": 1.7132,
+      "step": 257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039322076315268266,
+      "loss": 1.5422,
+      "step": 258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000393135174323865,
+      "loss": 1.6279,
+      "step": 259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003930490580262319,
+      "loss": 1.5107,
+      "step": 260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039296241449497443,
+      "loss": 1.6808,
+      "step": 261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039287524396672345,
+      "loss": 1.8189,
+      "step": 262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039278754667954936,
+      "loss": 1.6716,
+      "step": 263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039269932287296083,
+      "loss": 1.7371,
+      "step": 264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039261057278790483,
+      "loss": 1.5354,
+      "step": 265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003925212966667654,
+      "loss": 1.5848,
+      "step": 266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003924314947533633,
+      "loss": 1.6857,
+      "step": 267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039234116729295536,
+      "loss": 1.8029,
+      "step": 268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039225031453223367,
+      "loss": 1.6986,
+      "step": 269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039215893671932497,
+      "loss": 1.5044,
+      "step": 270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003920670341037899,
+      "loss": 1.6712,
+      "step": 271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039197460693662245,
+      "loss": 1.5513,
+      "step": 272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039188165547024916,
+      "loss": 1.4925,
+      "step": 273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039178817995852856,
+      "loss": 1.6916,
+      "step": 274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039169418065675024,
+      "loss": 1.7875,
+      "step": 275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039159965782163453,
+      "loss": 1.5011,
+      "step": 276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039150461171133126,
+      "loss": 1.6323,
+      "step": 277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003914090425854197,
+      "loss": 1.6161,
+      "step": 278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039131295070490727,
+      "loss": 1.7276,
+      "step": 279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003912163363322293,
+      "loss": 1.6082,
+      "step": 280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000391119199731248,
+      "loss": 1.732,
+      "step": 281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003910215411672516,
+      "loss": 1.6163,
+      "step": 282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003909233609069542,
+      "loss": 1.8471,
+      "step": 283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003908246592184946,
+      "loss": 1.6584,
+      "step": 284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003907254363714355,
+      "loss": 1.6216,
+      "step": 285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039062569263676307,
+      "loss": 1.6343,
+      "step": 286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003905254282868861,
+      "loss": 1.682,
+      "step": 287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039042464359563523,
+      "loss": 1.7296,
+      "step": 288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039032333883826206,
+      "loss": 1.5631,
+      "step": 289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00039022151429143865,
+      "loss": 1.7181,
+      "step": 290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00039011917023325655,
+      "loss": 1.7264,
+      "step": 291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003900163069432263,
+      "loss": 1.815,
+      "step": 292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038991292470227636,
+      "loss": 1.5588,
+      "step": 293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038980902379275257,
+      "loss": 1.7129,
+      "step": 294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038970460449841725,
+      "loss": 1.6191,
+      "step": 295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003895996671044485,
+      "loss": 1.6827,
+      "step": 296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003894942118974394,
+      "loss": 1.5777,
+      "step": 297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003893882391653973,
+      "loss": 1.664,
+      "step": 298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003892817491977427,
+      "loss": 1.7806,
+      "step": 299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000389174742285309,
+      "loss": 1.6249,
+      "step": 300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003890672187203413,
+      "loss": 1.7223,
+      "step": 301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003889591787964957,
+      "loss": 1.6471,
+      "step": 302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003888506228088385,
+      "loss": 1.5887,
+      "step": 303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003887415510538456,
+      "loss": 1.7015,
+      "step": 304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038863196382940123,
+      "loss": 1.6102,
+      "step": 305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038852186143479764,
+      "loss": 1.5874,
+      "step": 306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003884112441707339,
+      "loss": 1.6151,
+      "step": 307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038830011233931526,
+      "loss": 1.739,
+      "step": 308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003881884662440525,
+      "loss": 1.7366,
+      "step": 309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038807630618986063,
+      "loss": 1.6522,
+      "step": 310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003879636324830584,
+      "loss": 1.71,
+      "step": 311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003878504454313675,
+      "loss": 1.7334,
+      "step": 312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038773674534391144,
+      "loss": 1.613,
+      "step": 313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000387622532531215,
+      "loss": 1.7322,
+      "step": 314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038750780730520325,
+      "loss": 1.5234,
+      "step": 315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038739256997920063,
+      "loss": 1.5423,
+      "step": 316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003872768208679302,
+      "loss": 1.6155,
+      "step": 317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038716056028751284,
+      "loss": 1.6176,
+      "step": 318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038704378855546615,
+      "loss": 1.696,
+      "step": 319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038692650599070393,
+      "loss": 1.5135,
+      "step": 320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003868087129135348,
+      "loss": 1.53,
+      "step": 321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003866904096456619,
+      "loss": 1.6036,
+      "step": 322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038657159651018163,
+      "loss": 1.8304,
+      "step": 323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003864522738315829,
+      "loss": 1.6235,
+      "step": 324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003863324419357463,
+      "loss": 1.5371,
+      "step": 325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000386212101149943,
+      "loss": 1.6708,
+      "step": 326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038609125180283414,
+      "loss": 1.6137,
+      "step": 327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038596989422446954,
+      "loss": 1.546,
+      "step": 328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003858480287462874,
+      "loss": 1.5953,
+      "step": 329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038572565570111283,
+      "loss": 1.7084,
+      "step": 330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003856027754231571,
+      "loss": 1.6816,
+      "step": 331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038547938824801684,
+      "loss": 1.5167,
+      "step": 332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038535549451267315,
+      "loss": 1.4679,
+      "step": 333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003852310945554904,
+      "loss": 1.6548,
+      "step": 334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003851061887162156,
+      "loss": 1.9461,
+      "step": 335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003849807773359774,
+      "loss": 1.782,
+      "step": 336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003848548607572852,
+      "loss": 1.7077,
+      "step": 337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003847284393240279,
+      "loss": 1.5953,
+      "step": 338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038460151338147333,
+      "loss": 1.5885,
+      "step": 339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038447408327626733,
+      "loss": 1.5709,
+      "step": 340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003843461493564323,
+      "loss": 1.7463,
+      "step": 341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038421771197136696,
+      "loss": 1.7452,
+      "step": 342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038408877147184483,
+      "loss": 1.4542,
+      "step": 343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038395932821001354,
+      "loss": 1.4835,
+      "step": 344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038382938253939385,
+      "loss": 1.6311,
+      "step": 345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038369893481487847,
+      "loss": 1.6245,
+      "step": 346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038356798539273146,
+      "loss": 1.6264,
+      "step": 347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038343653463058705,
+      "loss": 1.5778,
+      "step": 348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003833045828874485,
+      "loss": 1.6097,
+      "step": 349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038317213052368744,
+      "loss": 1.67,
+      "step": 350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00038303917790104264,
+      "loss": 1.5424,
+      "step": 351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00038290572538261927,
+      "loss": 1.607,
+      "step": 352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00038277177333288765,
+      "loss": 1.6334,
+      "step": 353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003826373221176823,
+      "loss": 1.5178,
+      "step": 354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003825023721042012,
+      "loss": 1.5779,
+      "step": 355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003823669236610044,
+      "loss": 1.5685,
+      "step": 356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003822309771580132,
+      "loss": 1.6973,
+      "step": 357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00038209453296650944,
+      "loss": 1.8553,
+      "step": 358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003819575914591338,
+      "loss": 1.6619,
+      "step": 359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003818201530098853,
+      "loss": 1.7002,
+      "step": 360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003816822179941204,
+      "loss": 1.7676,
+      "step": 361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003815437867885514,
+      "loss": 1.5902,
+      "step": 362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003814048597712458,
+      "loss": 1.7975,
+      "step": 363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003812654373216254,
+      "loss": 1.5581,
+      "step": 364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00038112551982046484,
+      "loss": 1.7062,
+      "step": 365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00038098510764989087,
+      "loss": 1.7171,
+      "step": 366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003808442011933814,
+      "loss": 1.5299,
+      "step": 367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003807028008357638,
+      "loss": 1.5733,
+      "step": 368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000380560906963215,
+      "loss": 1.8065,
+      "step": 369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003804185199632591,
+      "loss": 1.4458,
+      "step": 370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003802756402247674,
+      "loss": 1.5408,
+      "step": 371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00038013226813795686,
+      "loss": 1.476,
+      "step": 372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003799884040943889,
+      "loss": 1.5259,
+      "step": 373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037984404848696873,
+      "loss": 1.6524,
+      "step": 374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003796992017099438,
+      "loss": 1.6925,
+      "step": 375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003795538641589033,
+      "loss": 1.5929,
+      "step": 376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003794080362307766,
+      "loss": 1.5452,
+      "step": 377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037926171832383226,
+      "loss": 1.6099,
+      "step": 378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037911491083767715,
+      "loss": 1.5971,
+      "step": 379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037896761417325524,
+      "loss": 1.8214,
+      "step": 380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003788198287328463,
+      "loss": 1.4613,
+      "step": 381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037867155492006516,
+      "loss": 1.6455,
+      "step": 382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037852279313986044,
+      "loss": 1.4235,
+      "step": 383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003783735437985133,
+      "loss": 1.6516,
+      "step": 384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003782238073036367,
+      "loss": 1.5719,
+      "step": 385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037807358406417374,
+      "loss": 1.5233,
+      "step": 386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037792287449039713,
+      "loss": 1.8381,
+      "step": 387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037777167899390776,
+      "loss": 1.5918,
+      "step": 388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003776199979876335,
+      "loss": 1.7216,
+      "step": 389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037746783188582827,
+      "loss": 1.8514,
+      "step": 390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037731518110407084,
+      "loss": 1.5811,
+      "step": 391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037716204605926367,
+      "loss": 1.6584,
+      "step": 392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003770084271696317,
+      "loss": 1.5818,
+      "step": 393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037685432485472145,
+      "loss": 1.7268,
+      "step": 394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003766997395353995,
+      "loss": 1.6397,
+      "step": 395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003765446716338518,
+      "loss": 1.7954,
+      "step": 396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037638912157358223,
+      "loss": 1.6724,
+      "step": 397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037623308977941124,
+      "loss": 1.5138,
+      "step": 398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037607657667747523,
+      "loss": 1.6065,
+      "step": 399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000375919582695225,
+      "loss": 1.6297,
+      "step": 400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003757621082614245,
+      "loss": 1.5817,
+      "step": 401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037560415380615014,
+      "loss": 1.5774,
+      "step": 402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037544571976078913,
+      "loss": 1.6676,
+      "step": 403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003752868065580384,
+      "loss": 1.7233,
+      "step": 404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037512741463190374,
+      "loss": 1.6582,
+      "step": 405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003749675444176983,
+      "loss": 1.648,
+      "step": 406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003748071963520412,
+      "loss": 1.581,
+      "step": 407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000374646370872857,
+      "loss": 1.7211,
+      "step": 408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037448506841937393,
+      "loss": 1.567,
+      "step": 409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003743232894321229,
+      "loss": 1.4671,
+      "step": 410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00037416103435293616,
+      "loss": 1.6647,
+      "step": 411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003739983036249465,
+      "loss": 1.5772,
+      "step": 412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003738350976925854,
+      "loss": 1.5484,
+      "step": 413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00037367141700158247,
+      "loss": 1.5484,
+      "step": 414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00037350726199896384,
+      "loss": 1.6113,
+      "step": 415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00037334263313305074,
+      "loss": 1.5764,
+      "step": 416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000373177530853459,
+      "loss": 1.6048,
+      "step": 417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003730119556110971,
+      "loss": 1.6771,
+      "step": 418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00037284590785816534,
+      "loss": 1.6426,
+      "step": 419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00037267938804815443,
+      "loss": 1.6309,
+      "step": 420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003725123966358444,
+      "loss": 1.6995,
+      "step": 421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00037234493407730307,
+      "loss": 1.8452,
+      "step": 422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003721770008298852,
+      "loss": 1.6109,
+      "step": 423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000372008597352231,
+      "loss": 1.6494,
+      "step": 424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00037183972410426483,
+      "loss": 1.7364,
+      "step": 425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003716703815471942,
+      "loss": 1.6014,
+      "step": 426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00037150057014350796,
+      "loss": 1.6395,
+      "step": 427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000371330290356976,
+      "loss": 1.6497,
+      "step": 428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000371159542652647,
+      "loss": 1.4915,
+      "step": 429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00037098832749684767,
+      "loss": 1.4835,
+      "step": 430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003708166453571813,
+      "loss": 1.8236,
+      "step": 431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003706444967025267,
+      "loss": 1.5515,
+      "step": 432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003704718820030366,
+      "loss": 1.4684,
+      "step": 433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003702988017301368,
+      "loss": 1.8934,
+      "step": 434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00037012525635652424,
+      "loss": 1.7834,
+      "step": 435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003699512463561664,
+      "loss": 1.6533,
+      "step": 436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036977677220429963,
+      "loss": 1.5941,
+      "step": 437
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036960183437742783,
+      "loss": 1.6756,
+      "step": 438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036942643335332134,
+      "loss": 1.5123,
+      "step": 439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036925056961101537,
+      "loss": 1.5916,
+      "step": 440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003690742436308091,
+      "loss": 1.4561,
+      "step": 441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003688974558942639,
+      "loss": 1.7021,
+      "step": 442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003687202068842024,
+      "loss": 1.4809,
+      "step": 443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036854249708470686,
+      "loss": 1.4896,
+      "step": 444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036836432698111806,
+      "loss": 1.5072,
+      "step": 445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000368185697060034,
+      "loss": 1.5797,
+      "step": 446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036800660780930835,
+      "loss": 1.5629,
+      "step": 447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036782705971804923,
+      "loss": 1.64,
+      "step": 448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036764705327661806,
+      "loss": 1.6466,
+      "step": 449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036746658897662793,
+      "loss": 1.6695,
+      "step": 450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036728566731094236,
+      "loss": 1.8699,
+      "step": 451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003671042887736741,
+      "loss": 1.7128,
+      "step": 452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036692245386018353,
+      "loss": 1.6109,
+      "step": 453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003667401630670774,
+      "loss": 1.5786,
+      "step": 454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003665574168922077,
+      "loss": 1.8162,
+      "step": 455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036637421583466995,
+      "loss": 1.7225,
+      "step": 456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003661905603948021,
+      "loss": 1.7744,
+      "step": 457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003660064510741829,
+      "loss": 1.7345,
+      "step": 458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003658218883756308,
+      "loss": 1.6193,
+      "step": 459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036563687280320245,
+      "loss": 1.5931,
+      "step": 460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036545140486219133,
+      "loss": 1.6595,
+      "step": 461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003652654850591264,
+      "loss": 1.7491,
+      "step": 462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003650791139017707,
+      "loss": 1.4703,
+      "step": 463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036489229189911985,
+      "loss": 1.615,
+      "step": 464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000364705019561401,
+      "loss": 1.6676,
+      "step": 465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036451729740007084,
+      "loss": 1.5476,
+      "step": 466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003643291259278149,
+      "loss": 1.4557,
+      "step": 467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036414050565854574,
+      "loss": 1.7099,
+      "step": 468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036395143710740143,
+      "loss": 1.7883,
+      "step": 469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003637619207907447,
+      "loss": 1.4658,
+      "step": 470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003635719572261608,
+      "loss": 1.6454,
+      "step": 471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003633815469324566,
+      "loss": 1.6372,
+      "step": 472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003631906904296591,
+      "loss": 1.4782,
+      "step": 473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003629993882390139,
+      "loss": 1.6954,
+      "step": 474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003628076408829836,
+      "loss": 1.7706,
+      "step": 475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00036261544888524695,
+      "loss": 1.7135,
+      "step": 476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003624228127706968,
+      "loss": 1.6762,
+      "step": 477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000362229733065439,
+      "loss": 1.697,
+      "step": 478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003620362102967909,
+      "loss": 1.5524,
+      "step": 479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00036184224499327976,
+      "loss": 1.488,
+      "step": 480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003616478376846417,
+      "loss": 1.5274,
+      "step": 481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003614529889018197,
+      "loss": 1.437,
+      "step": 482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003612576991769627,
+      "loss": 1.6498,
+      "step": 483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00036106196904342377,
+      "loss": 1.6727,
+      "step": 484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00036086579903575866,
+      "loss": 1.5969,
+      "step": 485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003606691896897248,
+      "loss": 1.6462,
+      "step": 486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000360472141542279,
+      "loss": 1.5607,
+      "step": 487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003602746551315769,
+      "loss": 1.687,
+      "step": 488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000360076730996971,
+      "loss": 1.3288,
+      "step": 489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000359878369679009,
+      "loss": 1.6054,
+      "step": 490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003596795717194328,
+      "loss": 1.6685,
+      "step": 491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035948033766117687,
+      "loss": 1.5,
+      "step": 492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035928066804836653,
+      "loss": 1.7705,
+      "step": 493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003590805634263167,
+      "loss": 1.4625,
+      "step": 494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003588800243415304,
+      "loss": 1.438,
+      "step": 495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035867905134169716,
+      "loss": 1.7041,
+      "step": 496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003584776449756915,
+      "loss": 1.5696,
+      "step": 497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003582758057935717,
+      "loss": 1.571,
+      "step": 498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003580735343465778,
+      "loss": 1.66,
+      "step": 499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003578708311871308,
+      "loss": 1.4366,
+      "step": 500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003576676968688303,
+      "loss": 1.5916,
+      "step": 501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003574641319464537,
+      "loss": 1.5619,
+      "step": 502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003572601369759544,
+      "loss": 1.7142,
+      "step": 503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003570557125144602,
+      "loss": 1.7435,
+      "step": 504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035685085912027197,
+      "loss": 1.5314,
+      "step": 505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035664557735286197,
+      "loss": 1.8234,
+      "step": 506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003564398677728724,
+      "loss": 1.7898,
+      "step": 507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003562337309421139,
+      "loss": 1.5523,
+      "step": 508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035602716742356397,
+      "loss": 1.5225,
+      "step": 509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003558201777813653,
+      "loss": 1.784,
+      "step": 510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00035561276258082444,
+      "loss": 1.5623,
+      "step": 511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00035540492238841025,
+      "loss": 1.6344,
+      "step": 512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003551966577717522,
+      "loss": 1.7194,
+      "step": 513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00035498796929963895,
+      "loss": 1.4878,
+      "step": 514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00035477885754201666,
+      "loss": 1.5972,
+      "step": 515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00035456932306998765,
+      "loss": 1.4085,
+      "step": 516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00035435936645580846,
+      "loss": 1.4897,
+      "step": 517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003541489882728889,
+      "loss": 1.5127,
+      "step": 518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00035393818909578985,
+      "loss": 1.6159,
+      "step": 519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003537269695002221,
+      "loss": 1.6481,
+      "step": 520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003535153300630444,
+      "loss": 1.6057,
+      "step": 521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003533032713622625,
+      "loss": 1.6588,
+      "step": 522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003530907939770269,
+      "loss": 1.6406,
+      "step": 523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00035287789848763166,
+      "loss": 1.4532,
+      "step": 524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003526645854755128,
+      "loss": 1.4354,
+      "step": 525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003524508555232464,
+      "loss": 1.6292,
+      "step": 526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00035223670921454757,
+      "loss": 1.7245,
+      "step": 527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003520221471342682,
+      "loss": 1.5872,
+      "step": 528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000351807169868396,
+      "loss": 1.5689,
+      "step": 529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003515917780040522,
+      "loss": 1.719,
+      "step": 530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003513759721294907,
+      "loss": 1.6705,
+      "step": 531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00035115975283409593,
+      "loss": 1.6617,
+      "step": 532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003509431207083814,
+      "loss": 1.6712,
+      "step": 533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003507260763439882,
+      "loss": 1.5884,
+      "step": 534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003505086203336831,
+      "loss": 1.7172,
+      "step": 535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003502907532713573,
+      "loss": 1.6537,
+      "step": 536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00035007247575202446,
+      "loss": 1.5581,
+      "step": 537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003498537883718194,
+      "loss": 1.632,
+      "step": 538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034963469172799615,
+      "loss": 1.5551,
+      "step": 539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003494151864189266,
+      "loss": 1.7205,
+      "step": 540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034919527304409857,
+      "loss": 1.5633,
+      "step": 541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003489749522041145,
+      "loss": 1.5343,
+      "step": 542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034875422450068963,
+      "loss": 1.6079,
+      "step": 543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003485330905366503,
+      "loss": 1.4977,
+      "step": 544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003483115509159325,
+      "loss": 1.6487,
+      "step": 545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034808960624358,
+      "loss": 1.6665,
+      "step": 546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034786725712574287,
+      "loss": 1.6199,
+      "step": 547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003476445041696757,
+      "loss": 1.5882,
+      "step": 548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000347421347983736,
+      "loss": 1.6632,
+      "step": 549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034719778917738256,
+      "loss": 1.4553,
+      "step": 550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003469738283611738,
+      "loss": 1.5907,
+      "step": 551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034674946614676597,
+      "loss": 1.7215,
+      "step": 552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003465247031469117,
+      "loss": 1.6233,
+      "step": 553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000346299539975458,
+      "loss": 1.6185,
+      "step": 554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000346073977247345,
+      "loss": 1.4927,
+      "step": 555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000345848015578604,
+      "loss": 1.6307,
+      "step": 556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034562165558635577,
+      "loss": 1.6981,
+      "step": 557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034539489788880883,
+      "loss": 1.5469,
+      "step": 558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003451677431052582,
+      "loss": 1.6071,
+      "step": 559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003449401918560831,
+      "loss": 1.658,
+      "step": 560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003447122447627456,
+      "loss": 1.6857,
+      "step": 561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003444839024477889,
+      "loss": 1.5205,
+      "step": 562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003442551655348355,
+      "loss": 1.5646,
+      "step": 563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034402603464858564,
+      "loss": 1.7056,
+      "step": 564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003437965104148156,
+      "loss": 1.6276,
+      "step": 565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034356659346037585,
+      "loss": 1.6262,
+      "step": 566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034333628441318936,
+      "loss": 1.6572,
+      "step": 567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034310558390225,
+      "loss": 1.6792,
+      "step": 568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003428744925576208,
+      "loss": 1.4858,
+      "step": 569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003426430110104321,
+      "loss": 1.318,
+      "step": 570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00034241113989288003,
+      "loss": 1.3952,
+      "step": 571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00034217887983822463,
+      "loss": 1.7068,
+      "step": 572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003419462314807879,
+      "loss": 1.7115,
+      "step": 573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003417131954559529,
+      "loss": 1.4274,
+      "step": 574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003414797724001609,
+      "loss": 1.4477,
+      "step": 575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003412459629509105,
+      "loss": 1.4958,
+      "step": 576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003410117677467553,
+      "loss": 1.513,
+      "step": 577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003407771874273028,
+      "loss": 1.6229,
+      "step": 578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00034054222263321194,
+      "loss": 1.5662,
+      "step": 579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000340306874006192,
+      "loss": 1.6515,
+      "step": 580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003400711421890001,
+      "loss": 1.762,
+      "step": 581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033983502782544044,
+      "loss": 1.4794,
+      "step": 582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003395985315603615,
+      "loss": 1.5187,
+      "step": 583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033936165403965516,
+      "loss": 1.5355,
+      "step": 584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003391243959102542,
+      "loss": 1.6659,
+      "step": 585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033888675782013113,
+      "loss": 1.5111,
+      "step": 586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000338648740418296,
+      "loss": 1.7099,
+      "step": 587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003384103443547948,
+      "loss": 1.6258,
+      "step": 588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003381715702807079,
+      "loss": 1.5812,
+      "step": 589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033793241884814783,
+      "loss": 1.5988,
+      "step": 590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003376928907102578,
+      "loss": 1.4556,
+      "step": 591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003374529865212097,
+      "loss": 1.4119,
+      "step": 592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033721270693620254,
+      "loss": 1.8002,
+      "step": 593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033697205261146076,
+      "loss": 1.6446,
+      "step": 594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033673102420423193,
+      "loss": 1.5731,
+      "step": 595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003364896223727855,
+      "loss": 1.5963,
+      "step": 596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033624784777641067,
+      "loss": 1.603,
+      "step": 597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033600570107541463,
+      "loss": 1.7701,
+      "step": 598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033576318293112103,
+      "loss": 1.4142,
+      "step": 599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033552029400586773,
+      "loss": 1.6056,
+      "step": 600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033527703496300535,
+      "loss": 1.6508,
+      "step": 601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033503340646689534,
+      "loss": 1.6636,
+      "step": 602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033478940918290815,
+      "loss": 1.6291,
+      "step": 603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033454504377742136,
+      "loss": 1.4496,
+      "step": 604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000334300310917818,
+      "loss": 1.5698,
+      "step": 605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003340552112724845,
+      "loss": 1.6541,
+      "step": 606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003338097455108093,
+      "loss": 1.5061,
+      "step": 607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033356391430318047,
+      "loss": 1.4854,
+      "step": 608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003333177183209842,
+      "loss": 1.6627,
+      "step": 609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003330711582366031,
+      "loss": 1.535,
+      "step": 610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033282423472341384,
+      "loss": 1.6511,
+      "step": 611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003325769484557859,
+      "loss": 1.5739,
+      "step": 612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003323293001090795,
+      "loss": 1.3858,
+      "step": 613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003320812903596434,
+      "loss": 1.646,
+      "step": 614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003318329198848138,
+      "loss": 1.7639,
+      "step": 615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003315841893629118,
+      "loss": 1.447,
+      "step": 616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000331335099473242,
+      "loss": 1.6622,
+      "step": 617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033108565089609034,
+      "loss": 1.5356,
+      "step": 618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033083584431272225,
+      "loss": 1.4241,
+      "step": 619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003305856804053812,
+      "loss": 1.8722,
+      "step": 620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033033515985728633,
+      "loss": 1.4449,
+      "step": 621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003300842833526309,
+      "loss": 1.8373,
+      "step": 622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032983305157658027,
+      "loss": 1.6631,
+      "step": 623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032958146521527,
+      "loss": 1.771,
+      "step": 624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032932952495580425,
+      "loss": 1.6418,
+      "step": 625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032907723148625355,
+      "loss": 1.4662,
+      "step": 626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003288245854956531,
+      "loss": 1.655,
+      "step": 627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032857158767400104,
+      "loss": 1.4315,
+      "step": 628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000328318238712256,
+      "loss": 1.716,
+      "step": 629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032806453930233595,
+      "loss": 1.6393,
+      "step": 630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003278104901371159,
+      "loss": 1.7138,
+      "step": 631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003275560919104259,
+      "loss": 1.6158,
+      "step": 632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003273013453170496,
+      "loss": 1.5414,
+      "step": 633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003270462510527218,
+      "loss": 1.5223,
+      "step": 634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003267908098141271,
+      "loss": 1.6761,
+      "step": 635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003265350222988972,
+      "loss": 1.443,
+      "step": 636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032627888920561024,
+      "loss": 1.756,
+      "step": 637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003260224112337876,
+      "loss": 1.506,
+      "step": 638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003257655890838927,
+      "loss": 1.6524,
+      "step": 639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032550842345732917,
+      "loss": 1.6236,
+      "step": 640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032525091505643825,
+      "loss": 1.578,
+      "step": 641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003249930645844978,
+      "loss": 1.2744,
+      "step": 642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032473487274571963,
+      "loss": 1.7205,
+      "step": 643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032447634024524786,
+      "loss": 1.6393,
+      "step": 644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003242174677891571,
+      "loss": 1.5814,
+      "step": 645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003239582560844503,
+      "loss": 1.4664,
+      "step": 646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003236987058390571,
+      "loss": 1.6065,
+      "step": 647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003234388177618314,
+      "loss": 1.5789,
+      "step": 648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032317859256255016,
+      "loss": 1.4482,
+      "step": 649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032291803095191074,
+      "loss": 1.5858,
+      "step": 650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00032265713364152933,
+      "loss": 1.6917,
+      "step": 651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000322395901343939,
+      "loss": 1.5473,
+      "step": 652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00032213433477258776,
+      "loss": 1.574,
+      "step": 653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003218724346418364,
+      "loss": 1.5813,
+      "step": 654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003216102016669568,
+      "loss": 1.4462,
+      "step": 655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003213476365641298,
+      "loss": 1.5667,
+      "step": 656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00032108474005044325,
+      "loss": 1.6149,
+      "step": 657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003208215128438904,
+      "loss": 1.5657,
+      "step": 658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003205579556633673,
+      "loss": 1.5153,
+      "step": 659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003202940692286714,
+      "loss": 1.5941,
+      "step": 660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00032002985426049925,
+      "loss": 1.6282,
+      "step": 661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031976531148044475,
+      "loss": 1.478,
+      "step": 662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003195004416109971,
+      "loss": 1.4623,
+      "step": 663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031923524537553864,
+      "loss": 1.5935,
+      "step": 664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003189697234983432,
+      "loss": 1.5293,
+      "step": 665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000318703876704574,
+      "loss": 1.4841,
+      "step": 666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031843770572028145,
+      "loss": 1.4985,
+      "step": 667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003181712112724015,
+      "loss": 1.604,
+      "step": 668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003179043940887535,
+      "loss": 1.5311,
+      "step": 669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003176372548980381,
+      "loss": 1.4669,
+      "step": 670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031736979442983557,
+      "loss": 1.4565,
+      "step": 671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003171020134146035,
+      "loss": 1.6435,
+      "step": 672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031683391258367484,
+      "loss": 1.6284,
+      "step": 673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031656549266925613,
+      "loss": 1.4404,
+      "step": 674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031629675440442536,
+      "loss": 1.5517,
+      "step": 675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031602769852312983,
+      "loss": 1.6271,
+      "step": 676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031575832576018437,
+      "loss": 1.6055,
+      "step": 677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031548863685126926,
+      "loss": 1.7491,
+      "step": 678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031521863253292814,
+      "loss": 1.6137,
+      "step": 679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031494831354256605,
+      "loss": 1.5847,
+      "step": 680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031467768061844753,
+      "loss": 1.3082,
+      "step": 681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003144067344996944,
+      "loss": 1.6252,
+      "step": 682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003141354759262839,
+      "loss": 1.5987,
+      "step": 683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003138639056390465,
+      "loss": 1.4824,
+      "step": 684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003135920243796641,
+      "loss": 1.3903,
+      "step": 685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000313319832890668,
+      "loss": 1.5826,
+      "step": 686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003130473319154365,
+      "loss": 1.5808,
+      "step": 687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031277452219819325,
+      "loss": 1.6244,
+      "step": 688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003125014044840051,
+      "loss": 1.5069,
+      "step": 689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031222797951878026,
+      "loss": 1.6237,
+      "step": 690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00031195424804926567,
+      "loss": 1.4301,
+      "step": 691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00031168021082304565,
+      "loss": 1.5819,
+      "step": 692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003114058685885396,
+      "loss": 1.8096,
+      "step": 693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003111312220949996,
+      "loss": 1.4657,
+      "step": 694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00031085627209250915,
+      "loss": 1.6379,
+      "step": 695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00031058101933198023,
+      "loss": 1.4865,
+      "step": 696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00031030546456515195,
+      "loss": 1.3837,
+      "step": 697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003100296085445881,
+      "loss": 1.6494,
+      "step": 698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003097534520236754,
+      "loss": 1.7428,
+      "step": 699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030947699575662087,
+      "loss": 1.5992,
+      "step": 700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003092002404984506,
+      "loss": 1.6187,
+      "step": 701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030892318700500703,
+      "loss": 1.7533,
+      "step": 702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003086458360329471,
+      "loss": 1.6341,
+      "step": 703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003083681883397403,
+      "loss": 1.6211,
+      "step": 704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030809024468366635,
+      "loss": 1.5643,
+      "step": 705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030781200582381336,
+      "loss": 1.5116,
+      "step": 706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003075334725200757,
+      "loss": 1.627,
+      "step": 707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030725464553315186,
+      "loss": 1.8024,
+      "step": 708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030697552562454223,
+      "loss": 1.4348,
+      "step": 709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030669611355654743,
+      "loss": 1.4995,
+      "step": 710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000306416410092266,
+      "loss": 1.6125,
+      "step": 711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003061364159955921,
+      "loss": 1.6559,
+      "step": 712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003058561320312139,
+      "loss": 1.469,
+      "step": 713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00030557555896461086,
+      "loss": 1.404,
+      "step": 714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003052946975620524,
+      "loss": 1.6919,
+      "step": 715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003050135485905951,
+      "loss": 1.5993,
+      "step": 716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003047321128180812,
+      "loss": 1.5386,
+      "step": 717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00030445039101313593,
+      "loss": 1.472,
+      "step": 718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00030416838394516587,
+      "loss": 1.413,
+      "step": 719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003038860923843567,
+      "loss": 1.6552,
+      "step": 720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00030360351710167094,
+      "loss": 1.4909,
+      "step": 721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003033206588688461,
+      "loss": 1.4127,
+      "step": 722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003030375184583923,
+      "loss": 1.4671,
+      "step": 723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00030275409664359056,
+      "loss": 1.6795,
+      "step": 724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00030247039419849025,
+      "loss": 1.6729,
+      "step": 725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003021864118979071,
+      "loss": 1.4749,
+      "step": 726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003019021505174215,
+      "loss": 1.6014,
+      "step": 727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003016176108333756,
+      "loss": 1.6341,
+      "step": 728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00030133279362287187,
+      "loss": 1.823,
+      "step": 729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0003010476996637706,
+      "loss": 1.5576,
+      "step": 730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003007623297346881,
+      "loss": 1.729,
+      "step": 731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00030047668461499413,
+      "loss": 1.4666,
+      "step": 732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0003001907650848103,
+      "loss": 1.5468,
+      "step": 733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002999045719250074,
+      "loss": 1.6001,
+      "step": 734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029961810591720364,
+      "loss": 1.6967,
+      "step": 735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029933136784376235,
+      "loss": 1.5668,
+      "step": 736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029904435848779,
+      "loss": 1.5286,
+      "step": 737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002987570786331339,
+      "loss": 1.6096,
+      "step": 738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029846952906438,
+      "loss": 1.5456,
+      "step": 739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029818171056685103,
+      "loss": 1.6902,
+      "step": 740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002978936239266042,
+      "loss": 1.7138,
+      "step": 741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029760526993042886,
+      "loss": 1.5856,
+      "step": 742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029731664936584473,
+      "loss": 1.464,
+      "step": 743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029702776302109943,
+      "loss": 1.4777,
+      "step": 744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029673861168516634,
+      "loss": 1.6905,
+      "step": 745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002964491961477429,
+      "loss": 1.54,
+      "step": 746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029615951719924783,
+      "loss": 1.6324,
+      "step": 747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029586957563081925,
+      "loss": 1.5705,
+      "step": 748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002955793722343127,
+      "loss": 1.7081,
+      "step": 749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002952889078022985,
+      "loss": 1.5256,
+      "step": 750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002949981831280602,
+      "loss": 1.4953,
+      "step": 751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000294707199005592,
+      "loss": 1.51,
+      "step": 752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029441595622959656,
+      "loss": 1.5811,
+      "step": 753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000294124455595483,
+      "loss": 1.5375,
+      "step": 754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029383269789936466,
+      "loss": 1.7053,
+      "step": 755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002935406839380571,
+      "loss": 1.4679,
+      "step": 756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002932484145090755,
+      "loss": 1.6812,
+      "step": 757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002929558904106329,
+      "loss": 1.5965,
+      "step": 758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029266311244163784,
+      "loss": 1.7395,
+      "step": 759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029237008140169227,
+      "loss": 1.5782,
+      "step": 760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002920767980910891,
+      "loss": 1.6442,
+      "step": 761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029178326331081043,
+      "loss": 1.5455,
+      "step": 762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002914894778625251,
+      "loss": 1.3559,
+      "step": 763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029119544254858643,
+      "loss": 1.4975,
+      "step": 764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002909011581720302,
+      "loss": 1.6059,
+      "step": 765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002906066255365724,
+      "loss": 1.6428,
+      "step": 766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029031184544660717,
+      "loss": 1.6501,
+      "step": 767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029001681870720434,
+      "loss": 1.4806,
+      "step": 768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002897215461241072,
+      "loss": 1.4095,
+      "step": 769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028942602850373086,
+      "loss": 1.6663,
+      "step": 770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002891302666531592,
+      "loss": 1.8152,
+      "step": 771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002888342613801436,
+      "loss": 1.7046,
+      "step": 772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028853801349309983,
+      "loss": 1.6848,
+      "step": 773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028824152380110645,
+      "loss": 1.6858,
+      "step": 774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002879447931139023,
+      "loss": 1.5137,
+      "step": 775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028764782224188453,
+      "loss": 1.6079,
+      "step": 776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028735061199610626,
+      "loss": 1.7438,
+      "step": 777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002870531631882742,
+      "loss": 1.6403,
+      "step": 778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002867554766307468,
+      "loss": 1.6697,
+      "step": 779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002864575531365316,
+      "loss": 1.513,
+      "step": 780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028615939351928337,
+      "loss": 1.5807,
+      "step": 781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028586099859330183,
+      "loss": 1.5632,
+      "step": 782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028556236917352926,
+      "loss": 1.7058,
+      "step": 783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028526350607554823,
+      "loss": 1.5961,
+      "step": 784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002849644101155797,
+      "loss": 1.6368,
+      "step": 785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002846650821104805,
+      "loss": 1.5342,
+      "step": 786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002843655228777413,
+      "loss": 1.6549,
+      "step": 787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002840657332354841,
+      "loss": 1.5317,
+      "step": 788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028376571400246035,
+      "loss": 1.687,
+      "step": 789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002834654659980484,
+      "loss": 1.421,
+      "step": 790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002831649900422514,
+      "loss": 1.639,
+      "step": 791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028286428695569523,
+      "loss": 1.62,
+      "step": 792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028256335755962584,
+      "loss": 1.5643,
+      "step": 793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002822622026759074,
+      "loss": 1.6884,
+      "step": 794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028196082312701977,
+      "loss": 1.7462,
+      "step": 795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002816592197360566,
+      "loss": 1.6372,
+      "step": 796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028135739332672274,
+      "loss": 1.5019,
+      "step": 797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002810553447233321,
+      "loss": 1.5365,
+      "step": 798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002807530747508056,
+      "loss": 1.6709,
+      "step": 799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002804505842346684,
+      "loss": 1.5888,
+      "step": 800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00028014787400104825,
+      "loss": 1.6876,
+      "step": 801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027984494487667296,
+      "loss": 1.5555,
+      "step": 802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000279541797688868,
+      "loss": 1.762,
+      "step": 803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027923843326555463,
+      "loss": 1.5098,
+      "step": 804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027893485243524706,
+      "loss": 1.7013,
+      "step": 805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002786310560270509,
+      "loss": 1.5521,
+      "step": 806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002783270448706601,
+      "loss": 1.6102,
+      "step": 807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027802281979635564,
+      "loss": 1.5667,
+      "step": 808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027771838163500223,
+      "loss": 1.3938,
+      "step": 809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027741373121804684,
+      "loss": 1.5571,
+      "step": 810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002771088693775159,
+      "loss": 1.4687,
+      "step": 811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002768037969460135,
+      "loss": 1.6164,
+      "step": 812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002764985147567187,
+      "loss": 1.5342,
+      "step": 813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002761930236433836,
+      "loss": 1.675,
+      "step": 814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027588732444033066,
+      "loss": 1.5154,
+      "step": 815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027558141798245074,
+      "loss": 1.6015,
+      "step": 816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002752753051052007,
+      "loss": 1.6213,
+      "step": 817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002749689866446015,
+      "loss": 1.7362,
+      "step": 818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000274662463437235,
+      "loss": 1.5406,
+      "step": 819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002743557363202427,
+      "loss": 1.5625,
+      "step": 820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002740488061313225,
+      "loss": 1.7225,
+      "step": 821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027374167370872746,
+      "loss": 1.5959,
+      "step": 822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027343433989126273,
+      "loss": 1.5027,
+      "step": 823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027312680551828337,
+      "loss": 1.5802,
+      "step": 824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002728190714296923,
+      "loss": 1.6822,
+      "step": 825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027251113846593785,
+      "loss": 1.5289,
+      "step": 826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002722030074680114,
+      "loss": 1.6296,
+      "step": 827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002718946792774455,
+      "loss": 1.6557,
+      "step": 828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002715861547363109,
+      "loss": 1.5834,
+      "step": 829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027127743468721466,
+      "loss": 1.4457,
+      "step": 830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00027096851997329794,
+      "loss": 1.5919,
+      "step": 831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002706594114382335,
+      "loss": 1.7794,
+      "step": 832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002703501099262233,
+      "loss": 1.5044,
+      "step": 833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00027004061628199645,
+      "loss": 1.4793,
+      "step": 834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026973093135080684,
+      "loss": 1.4537,
+      "step": 835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026942105597843076,
+      "loss": 1.5793,
+      "step": 836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026911099101116444,
+      "loss": 1.5592,
+      "step": 837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026880073729582213,
+      "loss": 1.4737,
+      "step": 838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002684902956797335,
+      "loss": 1.5163,
+      "step": 839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002681796670107413,
+      "loss": 1.3649,
+      "step": 840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002678688521371993,
+      "loss": 1.6616,
+      "step": 841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026755785190796965,
+      "loss": 1.5122,
+      "step": 842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002672466671724208,
+      "loss": 1.6713,
+      "step": 843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002669352987804251,
+      "loss": 1.594,
+      "step": 844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026662374758235655,
+      "loss": 1.5373,
+      "step": 845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002663120144290883,
+      "loss": 1.3893,
+      "step": 846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002660001001719904,
+      "loss": 1.6124,
+      "step": 847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026568800566292763,
+      "loss": 1.5687,
+      "step": 848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000265375731754257,
+      "loss": 1.2931,
+      "step": 849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002650632792988255,
+      "loss": 1.5906,
+      "step": 850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026475064914996773,
+      "loss": 1.5324,
+      "step": 851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002644378421615036,
+      "loss": 1.6167,
+      "step": 852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026412485918773595,
+      "loss": 1.6344,
+      "step": 853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026381170108344827,
+      "loss": 1.5958,
+      "step": 854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026349836870390235,
+      "loss": 1.4789,
+      "step": 855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026318486290483593,
+      "loss": 1.7105,
+      "step": 856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026287118454246033,
+      "loss": 1.4852,
+      "step": 857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026255733447345833,
+      "loss": 1.6808,
+      "step": 858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002622433135549814,
+      "loss": 1.5825,
+      "step": 859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026192912264464785,
+      "loss": 1.6492,
+      "step": 860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026161476260054014,
+      "loss": 1.5621,
+      "step": 861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002613002342812026,
+      "loss": 1.6808,
+      "step": 862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026098553854563916,
+      "loss": 1.5637,
+      "step": 863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026067067625331117,
+      "loss": 1.6629,
+      "step": 864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002603556482641345,
+      "loss": 1.3583,
+      "step": 865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00026004045543847796,
+      "loss": 1.6123,
+      "step": 866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025972509863716016,
+      "loss": 1.4788,
+      "step": 867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002594095787214478,
+      "loss": 1.5373,
+      "step": 868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025909389655305305,
+      "loss": 1.5852,
+      "step": 869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000258778052994131,
+      "loss": 1.5569,
+      "step": 870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002584620489072777,
+      "loss": 1.54,
+      "step": 871
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025814588515552753,
+      "loss": 1.4899,
+      "step": 872
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000257829562602351,
+      "loss": 1.4598,
+      "step": 873
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025751308211165223,
+      "loss": 1.5688,
+      "step": 874
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002571964445477668,
+      "loss": 1.4195,
+      "step": 875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002568796507754592,
+      "loss": 1.6566,
+      "step": 876
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002565627016599205,
+      "loss": 1.406,
+      "step": 877
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025624559806676603,
+      "loss": 1.827,
+      "step": 878
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025592834086203315,
+      "loss": 1.4952,
+      "step": 879
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002556109309121786,
+      "loss": 1.4737,
+      "step": 880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002552933690840762,
+      "loss": 1.4676,
+      "step": 881
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002549756562450149,
+      "loss": 1.5686,
+      "step": 882
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002546577932626957,
+      "loss": 1.639,
+      "step": 883
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002543397810052299,
+      "loss": 1.6325,
+      "step": 884
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025402162034113637,
+      "loss": 1.4458,
+      "step": 885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025370331213933926,
+      "loss": 1.4646,
+      "step": 886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002533848572691658,
+      "loss": 1.7105,
+      "step": 887
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025306625660034365,
+      "loss": 1.6741,
+      "step": 888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002527475110029988,
+      "loss": 1.6431,
+      "step": 889
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002524286213476529,
+      "loss": 1.7294,
+      "step": 890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00025210958850522104,
+      "loss": 1.5165,
+      "step": 891
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002517904133470095,
+      "loss": 1.5219,
+      "step": 892
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00025147109674471317,
+      "loss": 1.3967,
+      "step": 893
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002511516395704132,
+      "loss": 1.6214,
+      "step": 894
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00025083204269657467,
+      "loss": 1.6539,
+      "step": 895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002505123069960442,
+      "loss": 1.5878,
+      "step": 896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002501924333420475,
+      "loss": 1.6149,
+      "step": 897
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002498724226081872,
+      "loss": 1.6648,
+      "step": 898
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002495522756684402,
+      "loss": 1.4552,
+      "step": 899
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024923199339715543,
+      "loss": 1.531,
+      "step": 900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002489115766690513,
+      "loss": 1.6431,
+      "step": 901
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002485910263592135,
+      "loss": 1.5006,
+      "step": 902
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024827034334309265,
+      "loss": 1.4951,
+      "step": 903
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024794952849650174,
+      "loss": 1.3887,
+      "step": 904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002476285826956138,
+      "loss": 1.593,
+      "step": 905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002473075068169593,
+      "loss": 1.4254,
+      "step": 906
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024698630173742436,
+      "loss": 1.5903,
+      "step": 907
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002466649683342477,
+      "loss": 1.5224,
+      "step": 908
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002466649683342477,
+      "loss": 1.6739,
+      "step": 909
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002463435074850184,
+      "loss": 1.4916,
+      "step": 910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000246021920067674,
+      "loss": 1.3713,
+      "step": 911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002457002069604973,
+      "loss": 1.3699,
+      "step": 912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002453783690421146,
+      "loss": 1.3908,
+      "step": 913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000245056407191493,
+      "loss": 1.4784,
+      "step": 914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024473432228793807,
+      "loss": 1.5448,
+      "step": 915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002444121152110915,
+      "loss": 1.4938,
+      "step": 916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024408978684092847,
+      "loss": 1.5958,
+      "step": 917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024376733805775574,
+      "loss": 1.5453,
+      "step": 918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024344476974220855,
+      "loss": 1.4917,
+      "step": 919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024312208277524892,
+      "loss": 1.4726,
+      "step": 920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024279927803816276,
+      "loss": 1.7581,
+      "step": 921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024247635641255766,
+      "loss": 1.517,
+      "step": 922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024215331878036037,
+      "loss": 1.5761,
+      "step": 923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024183016602381447,
+      "loss": 1.3919,
+      "step": 924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024150689902547811,
+      "loss": 1.6481,
+      "step": 925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024118351866822137,
+      "loss": 1.5396,
+      "step": 926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024086002583522382,
+      "loss": 1.6246,
+      "step": 927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024053642140997225,
+      "loss": 1.4913,
+      "step": 928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00024021270627625825,
+      "loss": 1.6974,
+      "step": 929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023988888131817583,
+      "loss": 1.727,
+      "step": 930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002395649474201189,
+      "loss": 1.4335,
+      "step": 931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002392409054667788,
+      "loss": 1.7088,
+      "step": 932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023891675634314202,
+      "loss": 1.6817,
+      "step": 933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023859250093448783,
+      "loss": 1.7091,
+      "step": 934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023826814012638568,
+      "loss": 1.5044,
+      "step": 935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023794367480469295,
+      "loss": 1.6067,
+      "step": 936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002376191058555524,
+      "loss": 1.6023,
+      "step": 937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023729443416538982,
+      "loss": 1.5024,
+      "step": 938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023696966062091148,
+      "loss": 1.6559,
+      "step": 939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023664478610910207,
+      "loss": 1.6554,
+      "step": 940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002363198115172219,
+      "loss": 1.5379,
+      "step": 941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023599473773280454,
+      "loss": 1.4307,
+      "step": 942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002356695656436546,
+      "loss": 1.6144,
+      "step": 943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023534429613784497,
+      "loss": 1.6197,
+      "step": 944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023501893010371476,
+      "loss": 1.6309,
+      "step": 945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023469346842986677,
+      "loss": 1.5075,
+      "step": 946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002343679120051648,
+      "loss": 1.6614,
+      "step": 947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023404226171873157,
+      "loss": 1.4659,
+      "step": 948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023371651845994603,
+      "loss": 1.3864,
+      "step": 949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023339068311844114,
+      "loss": 1.3556,
+      "step": 950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002330647565841013,
+      "loss": 1.5859,
+      "step": 951
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002327387397470601,
+      "loss": 1.6599,
+      "step": 952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00023241263349769748,
+      "loss": 1.5984,
+      "step": 953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002320864387266378,
+      "loss": 1.5987,
+      "step": 954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00023176015632474703,
+      "loss": 1.727,
+      "step": 955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00023143378718313066,
+      "loss": 1.4925,
+      "step": 956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00023110733219313087,
+      "loss": 1.5061,
+      "step": 957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002307807922463245,
+      "loss": 1.6809,
+      "step": 958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00023045416823452023,
+      "loss": 1.5185,
+      "step": 959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00023012746104975632,
+      "loss": 1.5064,
+      "step": 960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022980067158429832,
+      "loss": 1.4743,
+      "step": 961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022947380073063656,
+      "loss": 1.5773,
+      "step": 962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022914684938148342,
+      "loss": 1.3633,
+      "step": 963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022881981842977117,
+      "loss": 1.6052,
+      "step": 964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022849270876864965,
+      "loss": 1.5255,
+      "step": 965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022816552129148354,
+      "loss": 1.7143,
+      "step": 966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022783825689184998,
+      "loss": 1.5166,
+      "step": 967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022751091646353632,
+      "loss": 1.5318,
+      "step": 968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022718350090053752,
+      "loss": 1.512,
+      "step": 969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022685601109705364,
+      "loss": 1.5758,
+      "step": 970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022652844794748765,
+      "loss": 1.4,
+      "step": 971
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002262008123464427,
+      "loss": 1.5296,
+      "step": 972
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002258731051887199,
+      "loss": 1.7131,
+      "step": 973
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022554532736931577,
+      "loss": 1.5665,
+      "step": 974
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022521747978341972,
+      "loss": 1.4738,
+      "step": 975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022488956332641192,
+      "loss": 1.6389,
+      "step": 976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022456157889386033,
+      "loss": 1.6043,
+      "step": 977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022423352738151886,
+      "loss": 1.5159,
+      "step": 978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022390540968532442,
+      "loss": 1.5425,
+      "step": 979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002235772267013947,
+      "loss": 1.5404,
+      "step": 980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022324897932602574,
+      "loss": 1.6195,
+      "step": 981
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002229206684556895,
+      "loss": 1.5286,
+      "step": 982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002225922949870311,
+      "loss": 1.4833,
+      "step": 983
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022226385981686706,
+      "loss": 1.5749,
+      "step": 984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022193536384218195,
+      "loss": 1.718,
+      "step": 985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022160680796012665,
+      "loss": 1.4774,
+      "step": 986
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022127819306801567,
+      "loss": 1.4944,
+      "step": 987
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022094952006332453,
+      "loss": 1.6471,
+      "step": 988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022062078984368756,
+      "loss": 1.6706,
+      "step": 989
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022029200330689545,
+      "loss": 1.5164,
+      "step": 990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021996316135089239,
+      "loss": 1.6634,
+      "step": 991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021963426487377433,
+      "loss": 1.7244,
+      "step": 992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021930531477378572,
+      "loss": 1.4838,
+      "step": 993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002189763119493178,
+      "loss": 1.6398,
+      "step": 994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021864725729890555,
+      "loss": 1.3957,
+      "step": 995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002183181517212256,
+      "loss": 1.6115,
+      "step": 996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021798899611509377,
+      "loss": 1.7738,
+      "step": 997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021765979137946233,
+      "loss": 1.6396,
+      "step": 998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021733053841341775,
+      "loss": 1.5969,
+      "step": 999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021700123811617834,
+      "loss": 1.4756,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002166718913870916,
+      "loss": 1.5806,
+      "step": 1001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002163424991256318,
+      "loss": 1.5961,
+      "step": 1002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002160130622313977,
+      "loss": 1.5734,
+      "step": 1003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002156835816041098,
+      "loss": 1.4678,
+      "step": 1004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002153540581436081,
+      "loss": 1.587,
+      "step": 1005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021502449274984957,
+      "loss": 1.6823,
+      "step": 1006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021469488632290583,
+      "loss": 1.5939,
+      "step": 1007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021436523976296038,
+      "loss": 1.6331,
+      "step": 1008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021403555397030632,
+      "loss": 1.4015,
+      "step": 1009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021370582984534412,
+      "loss": 1.3894,
+      "step": 1010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021337606828857872,
+      "loss": 1.5989,
+      "step": 1011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021304627020061738,
+      "loss": 1.444,
+      "step": 1012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021271643648216717,
+      "loss": 1.5686,
+      "step": 1013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002123865680340323,
+      "loss": 1.6826,
+      "step": 1014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021205666575711205,
+      "loss": 1.4933,
+      "step": 1015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021172673055239795,
+      "loss": 1.503,
+      "step": 1016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021139676332097146,
+      "loss": 1.6633,
+      "step": 1017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021106676496400164,
+      "loss": 1.6478,
+      "step": 1018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021073673638274238,
+      "loss": 1.4642,
+      "step": 1019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021040667847853017,
+      "loss": 1.6525,
+      "step": 1020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00021007659215278174,
+      "loss": 1.4527,
+      "step": 1021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020974647830699112,
+      "loss": 1.4418,
+      "step": 1022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020941633784272786,
+      "loss": 1.3624,
+      "step": 1023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020908617166163392,
+      "loss": 1.4555,
+      "step": 1024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020875598066542165,
+      "loss": 1.6097,
+      "step": 1025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002084257657558712,
+      "loss": 1.4663,
+      "step": 1026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002080955278348278,
+      "loss": 1.5117,
+      "step": 1027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002077652678041999,
+      "loss": 1.4675,
+      "step": 1028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020743498656595598,
+      "loss": 1.5064,
+      "step": 1029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020710468502212263,
+      "loss": 1.5412,
+      "step": 1030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002067743640747818,
+      "loss": 1.4103,
+      "step": 1031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002064440246260685,
+      "loss": 1.4971,
+      "step": 1032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020611366757816835,
+      "loss": 1.5548,
+      "step": 1033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002057832938333148,
+      "loss": 1.4879,
+      "step": 1034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002054529042937871,
+      "loss": 1.441,
+      "step": 1035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020512249986190755,
+      "loss": 1.5259,
+      "step": 1036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020479208144003912,
+      "loss": 1.6095,
+      "step": 1037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020446164993058309,
+      "loss": 1.3883,
+      "step": 1038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002041312062359764,
+      "loss": 1.5599,
+      "step": 1039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020380075125868926,
+      "loss": 1.5393,
+      "step": 1040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020347028590122257,
+      "loss": 1.5833,
+      "step": 1041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002031398110661059,
+      "loss": 1.5744,
+      "step": 1042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020280932765589444,
+      "loss": 1.5294,
+      "step": 1043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020247883657316696,
+      "loss": 1.6453,
+      "step": 1044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020214833872052301,
+      "loss": 1.5872,
+      "step": 1045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002018178350005807,
+      "loss": 1.5812,
+      "step": 1046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020148732631597427,
+      "loss": 1.3474,
+      "step": 1047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020115681356935148,
+      "loss": 1.6486,
+      "step": 1048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020082629766337103,
+      "loss": 1.4995,
+      "step": 1049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020049577950070045,
+      "loss": 1.6424,
+      "step": 1050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020016525998401327,
+      "loss": 1.4996,
+      "step": 1051
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019983474001598677,
+      "loss": 1.6066,
+      "step": 1052
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019950422049929965,
+      "loss": 1.5163,
+      "step": 1053
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019917370233662898,
+      "loss": 1.4323,
+      "step": 1054
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019884318643064854,
+      "loss": 1.5404,
+      "step": 1055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019851267368402577,
+      "loss": 1.5302,
+      "step": 1056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019818216499941935,
+      "loss": 1.7073,
+      "step": 1057
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019785166127947709,
+      "loss": 1.6111,
+      "step": 1058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001975211634268331,
+      "loss": 1.4342,
+      "step": 1059
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019719067234410555,
+      "loss": 1.5892,
+      "step": 1060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019686018893389417,
+      "loss": 1.3523,
+      "step": 1061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019652971409877748,
+      "loss": 1.4861,
+      "step": 1062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019619924874131087,
+      "loss": 1.5304,
+      "step": 1063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019586879376402362,
+      "loss": 1.7805,
+      "step": 1064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019553835006941688,
+      "loss": 1.4986,
+      "step": 1065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001952079185599609,
+      "loss": 1.5767,
+      "step": 1066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001948775001380925,
+      "loss": 1.6053,
+      "step": 1067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019454709570621297,
+      "loss": 1.6215,
+      "step": 1068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019421670616668522,
+      "loss": 1.5673,
+      "step": 1069
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019388633242183175,
+      "loss": 1.5943,
+      "step": 1070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019355597537393154,
+      "loss": 1.5343,
+      "step": 1071
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019322563592521824,
+      "loss": 1.5878,
+      "step": 1072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019289531497787744,
+      "loss": 1.4234,
+      "step": 1073
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019256501343404406,
+      "loss": 1.4744,
+      "step": 1074
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001922347321958002,
+      "loss": 1.5477,
+      "step": 1075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019190447216517223,
+      "loss": 1.421,
+      "step": 1076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019157423424412886,
+      "loss": 1.709,
+      "step": 1077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019124401933457837,
+      "loss": 1.6079,
+      "step": 1078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001909138283383661,
+      "loss": 1.4921,
+      "step": 1079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019058366215727227,
+      "loss": 1.6009,
+      "step": 1080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019025352169300895,
+      "loss": 1.4151,
+      "step": 1081
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018992340784721834,
+      "loss": 1.5145,
+      "step": 1082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018959332152146987,
+      "loss": 1.7787,
+      "step": 1083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018926326361725767,
+      "loss": 1.6729,
+      "step": 1084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018893323503599844,
+      "loss": 1.8148,
+      "step": 1085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018860323667902856,
+      "loss": 1.6266,
+      "step": 1086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001882732694476021,
+      "loss": 1.5139,
+      "step": 1087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000187943334242888,
+      "loss": 1.5239,
+      "step": 1088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001876134319659677,
+      "loss": 1.6728,
+      "step": 1089
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001872835635178329,
+      "loss": 1.7159,
+      "step": 1090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018695372979938264,
+      "loss": 1.5613,
+      "step": 1091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001866239317114213,
+      "loss": 1.7827,
+      "step": 1092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018629417015465592,
+      "loss": 1.4987,
+      "step": 1093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018596444602969367,
+      "loss": 1.599,
+      "step": 1094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018563476023703972,
+      "loss": 1.3636,
+      "step": 1095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001853051136770942,
+      "loss": 1.6464,
+      "step": 1096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001849755072501504,
+      "loss": 1.451,
+      "step": 1097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018464594185639194,
+      "loss": 1.4646,
+      "step": 1098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018431641839589021,
+      "loss": 1.6525,
+      "step": 1099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018398693776860235,
+      "loss": 1.4996,
+      "step": 1100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001836575008743682,
+      "loss": 1.483,
+      "step": 1101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018332810861290848,
+      "loss": 1.4097,
+      "step": 1102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001829987618838217,
+      "loss": 1.5803,
+      "step": 1103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018266946158658227,
+      "loss": 1.6121,
+      "step": 1104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018234020862053775,
+      "loss": 1.6006,
+      "step": 1105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018201100388490627,
+      "loss": 1.627,
+      "step": 1106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018168184827877442,
+      "loss": 1.7058,
+      "step": 1107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018135274270109447,
+      "loss": 1.673,
+      "step": 1108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001810236880506822,
+      "loss": 1.9456,
+      "step": 1109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018069468522621433,
+      "loss": 1.6382,
+      "step": 1110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018036573512622572,
+      "loss": 1.4667,
+      "step": 1111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018003683864910763,
+      "loss": 1.4617,
+      "step": 1112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001797079966931046,
+      "loss": 1.4715,
+      "step": 1113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001793792101563124,
+      "loss": 1.5749,
+      "step": 1114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017905047993667554,
+      "loss": 1.6088,
+      "step": 1115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001787218069319844,
+      "loss": 1.6457,
+      "step": 1116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001783931920398734,
+      "loss": 1.6885,
+      "step": 1117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017806463615781807,
+      "loss": 1.4638,
+      "step": 1118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017773614018313293,
+      "loss": 1.5279,
+      "step": 1119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001774077050129689,
+      "loss": 1.3872,
+      "step": 1120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017707933154431055,
+      "loss": 1.6311,
+      "step": 1121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017675102067397436,
+      "loss": 1.4838,
+      "step": 1122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017642277329860532,
+      "loss": 1.5995,
+      "step": 1123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001760945903146756,
+      "loss": 1.6014,
+      "step": 1124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017576647261848119,
+      "loss": 1.55,
+      "step": 1125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017543842110613966,
+      "loss": 1.6189,
+      "step": 1126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017511043667358818,
+      "loss": 1.5004,
+      "step": 1127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001747825202165803,
+      "loss": 1.5344,
+      "step": 1128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017445467263068422,
+      "loss": 1.4693,
+      "step": 1129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017412689481128015,
+      "loss": 1.5528,
+      "step": 1130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001737991876535573,
+      "loss": 1.5298,
+      "step": 1131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017347155205251245,
+      "loss": 1.4865,
+      "step": 1132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001731439889029464,
+      "loss": 1.4381,
+      "step": 1133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017281649909946256,
+      "loss": 1.4918,
+      "step": 1134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017248908353646372,
+      "loss": 1.7483,
+      "step": 1135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017216174310815004,
+      "loss": 1.5986,
+      "step": 1136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017183447870851653,
+      "loss": 1.6334,
+      "step": 1137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001715072912313504,
+      "loss": 1.6603,
+      "step": 1138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017118018157022887,
+      "loss": 1.5052,
+      "step": 1139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017085315061851665,
+      "loss": 1.4917,
+      "step": 1140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017052619926936346,
+      "loss": 1.4266,
+      "step": 1141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017019932841570173,
+      "loss": 1.4824,
+      "step": 1142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016987253895024376,
+      "loss": 1.553,
+      "step": 1143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001695458317654799,
+      "loss": 1.5831,
+      "step": 1144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016921920775367556,
+      "loss": 1.5165,
+      "step": 1145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001688926678068691,
+      "loss": 1.6343,
+      "step": 1146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001685662128168694,
+      "loss": 1.4093,
+      "step": 1147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016823984367525302,
+      "loss": 1.5443,
+      "step": 1148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001679135612733623,
+      "loss": 1.5261,
+      "step": 1149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016758736650230257,
+      "loss": 1.7538,
+      "step": 1150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016726126025293992,
+      "loss": 1.5514,
+      "step": 1151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001669352434158987,
+      "loss": 1.3497,
+      "step": 1152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001666093168815589,
+      "loss": 1.4964,
+      "step": 1153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016628348154005404,
+      "loss": 1.4952,
+      "step": 1154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016595773828126847,
+      "loss": 1.4553,
+      "step": 1155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016563208799483522,
+      "loss": 1.4294,
+      "step": 1156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016530653157013328,
+      "loss": 1.5523,
+      "step": 1157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016498106989628526,
+      "loss": 1.3405,
+      "step": 1158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016465570386215513,
+      "loss": 1.5908,
+      "step": 1159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016433043435634548,
+      "loss": 1.4411,
+      "step": 1160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016400526226719543,
+      "loss": 1.3684,
+      "step": 1161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016368018848277816,
+      "loss": 1.4479,
+      "step": 1162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016335521389089792,
+      "loss": 1.5124,
+      "step": 1163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016303033937908854,
+      "loss": 1.6517,
+      "step": 1164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016270556583461023,
+      "loss": 1.4531,
+      "step": 1165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016238089414444758,
+      "loss": 1.3264,
+      "step": 1166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016205632519530707,
+      "loss": 1.5187,
+      "step": 1167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016173185987361432,
+      "loss": 1.5387,
+      "step": 1168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001614074990655122,
+      "loss": 1.4165,
+      "step": 1169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000161083243656858,
+      "loss": 1.4233,
+      "step": 1170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016075909453322132,
+      "loss": 1.5269,
+      "step": 1171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016043505257988115,
+      "loss": 1.5829,
+      "step": 1172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00016011111868182416,
+      "loss": 1.6682,
+      "step": 1173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015978729372374177,
+      "loss": 1.4563,
+      "step": 1174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001594635785900278,
+      "loss": 1.4737,
+      "step": 1175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015913997416477628,
+      "loss": 1.572,
+      "step": 1176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015881648133177868,
+      "loss": 1.5574,
+      "step": 1177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015849310097452185,
+      "loss": 1.4618,
+      "step": 1178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015816983397618555,
+      "loss": 1.6466,
+      "step": 1179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015784668121963968,
+      "loss": 1.4817,
+      "step": 1180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015752364358744244,
+      "loss": 1.5657,
+      "step": 1181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001572007219618373,
+      "loss": 1.7042,
+      "step": 1182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001568779172247511,
+      "loss": 1.5432,
+      "step": 1183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001565552302577915,
+      "loss": 1.7243,
+      "step": 1184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015623266194224434,
+      "loss": 1.6929,
+      "step": 1185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001559102131590716,
+      "loss": 1.58,
+      "step": 1186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001555878847889086,
+      "loss": 1.4482,
+      "step": 1187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015526567771206195,
+      "loss": 1.4163,
+      "step": 1188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015494359280850703,
+      "loss": 1.4849,
+      "step": 1189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001546216309578854,
+      "loss": 1.6257,
+      "step": 1190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015429979303950277,
+      "loss": 1.4019,
+      "step": 1191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015397807993232604,
+      "loss": 1.6258,
+      "step": 1192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015365649251498158,
+      "loss": 1.5102,
+      "step": 1193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015333503166575237,
+      "loss": 1.7245,
+      "step": 1194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015301369826257563,
+      "loss": 1.4299,
+      "step": 1195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015269249318304072,
+      "loss": 1.4183,
+      "step": 1196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015237141730438627,
+      "loss": 1.5347,
+      "step": 1197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015205047150349823,
+      "loss": 1.6355,
+      "step": 1198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015172965665690737,
+      "loss": 1.6188,
+      "step": 1199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015140897364078647,
+      "loss": 1.3618,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001510884233309488,
+      "loss": 1.5299,
+      "step": 1201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015076800660284462,
+      "loss": 1.5383,
+      "step": 1202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015044772433155985,
+      "loss": 1.5368,
+      "step": 1203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00015012757739181284,
+      "loss": 1.6522,
+      "step": 1204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014980756665795253,
+      "loss": 1.5328,
+      "step": 1205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014948769300395588,
+      "loss": 1.4371,
+      "step": 1206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014916795730342538,
+      "loss": 1.5072,
+      "step": 1207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001488483604295869,
+      "loss": 1.5613,
+      "step": 1208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014852890325528688,
+      "loss": 1.5922,
+      "step": 1209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014820958665299049,
+      "loss": 1.5502,
+      "step": 1210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014789041149477897,
+      "loss": 1.3874,
+      "step": 1211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014757137865234712,
+      "loss": 1.6309,
+      "step": 1212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014725248899700126,
+      "loss": 1.5276,
+      "step": 1213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014693374339965637,
+      "loss": 1.6013,
+      "step": 1214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014661514273083424,
+      "loss": 1.4085,
+      "step": 1215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001462966878606608,
+      "loss": 1.3623,
+      "step": 1216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001459783796588637,
+      "loss": 1.5691,
+      "step": 1217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014566021899477022,
+      "loss": 1.557,
+      "step": 1218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014534220673730438,
+      "loss": 1.6516,
+      "step": 1219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014502434375498514,
+      "loss": 1.4433,
+      "step": 1220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014470663091592381,
+      "loss": 1.5922,
+      "step": 1221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014438906908782149,
+      "loss": 1.4993,
+      "step": 1222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014407165913796693,
+      "loss": 1.5018,
+      "step": 1223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014375440193323402,
+      "loss": 1.4906,
+      "step": 1224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014343729834007953,
+      "loss": 1.5737,
+      "step": 1225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014312034922454086,
+      "loss": 1.4981,
+      "step": 1226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001428035554522332,
+      "loss": 1.6424,
+      "step": 1227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014248691788834784,
+      "loss": 1.4925,
+      "step": 1228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014217043739764908,
+      "loss": 1.5765,
+      "step": 1229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014185411484447252,
+      "loss": 1.5934,
+      "step": 1230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014153795109272238,
+      "loss": 1.4563,
+      "step": 1231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014122194700586904,
+      "loss": 1.5703,
+      "step": 1232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014090610344694702,
+      "loss": 1.5285,
+      "step": 1233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014059042127855217,
+      "loss": 1.4793,
+      "step": 1234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00014027490136283988,
+      "loss": 1.7095,
+      "step": 1235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013995954456152206,
+      "loss": 1.5333,
+      "step": 1236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013964435173586547,
+      "loss": 1.4531,
+      "step": 1237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013932932374668893,
+      "loss": 1.5855,
+      "step": 1238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013901446145436083,
+      "loss": 1.3752,
+      "step": 1239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013869976571879747,
+      "loss": 1.5541,
+      "step": 1240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001383852373994599,
+      "loss": 1.7656,
+      "step": 1241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013807087735535214,
+      "loss": 1.5334,
+      "step": 1242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013775668644501864,
+      "loss": 1.4591,
+      "step": 1243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013744266552654172,
+      "loss": 1.6527,
+      "step": 1244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013712881545753972,
+      "loss": 1.5402,
+      "step": 1245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013681513709516414,
+      "loss": 1.4721,
+      "step": 1246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013650163129609767,
+      "loss": 1.451,
+      "step": 1247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001361882989165518,
+      "loss": 1.5583,
+      "step": 1248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001358751408122641,
+      "loss": 1.535,
+      "step": 1249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013556215783849644,
+      "loss": 1.2767,
+      "step": 1250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001352493508500323,
+      "loss": 1.5292,
+      "step": 1251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001349367207011745,
+      "loss": 1.4486,
+      "step": 1252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013462426824574307,
+      "loss": 1.4923,
+      "step": 1253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013431199433707245,
+      "loss": 1.5831,
+      "step": 1254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001339998998280097,
+      "loss": 1.5556,
+      "step": 1255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013368798557091176,
+      "loss": 1.3624,
+      "step": 1256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013337625241764345,
+      "loss": 1.5453,
+      "step": 1257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013306470121957493,
+      "loss": 1.7471,
+      "step": 1258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001327533328275792,
+      "loss": 1.4559,
+      "step": 1259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013244214809203042,
+      "loss": 1.2641,
+      "step": 1260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013213114786280074,
+      "loss": 1.4312,
+      "step": 1261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001318203329892587,
+      "loss": 1.4512,
+      "step": 1262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013150970432026656,
+      "loss": 1.4561,
+      "step": 1263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001311992627041779,
+      "loss": 1.5522,
+      "step": 1264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013088900898883564,
+      "loss": 1.5752,
+      "step": 1265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001305789440215693,
+      "loss": 1.7763,
+      "step": 1266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013026906864919323,
+      "loss": 1.6178,
+      "step": 1267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012995938371800363,
+      "loss": 1.5093,
+      "step": 1268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012964989007377677,
+      "loss": 1.4769,
+      "step": 1269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012934058856176663,
+      "loss": 1.4533,
+      "step": 1270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012903148002670205,
+      "loss": 1.5406,
+      "step": 1271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001287225653127854,
+      "loss": 1.4527,
+      "step": 1272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001284138452636892,
+      "loss": 1.1498,
+      "step": 1273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012810532072255452,
+      "loss": 1.3571,
+      "step": 1274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012779699253198863,
+      "loss": 1.5416,
+      "step": 1275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012748886153406223,
+      "loss": 1.4866,
+      "step": 1276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001271809285703078,
+      "loss": 1.5418,
+      "step": 1277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012687319448171668,
+      "loss": 1.6024,
+      "step": 1278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012656566010873726,
+      "loss": 1.3701,
+      "step": 1279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012625832629127256,
+      "loss": 1.5769,
+      "step": 1280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012595119386867753,
+      "loss": 1.7299,
+      "step": 1281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012564426367975745,
+      "loss": 1.5873,
+      "step": 1282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000125337536562765,
+      "loss": 1.5423,
+      "step": 1283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001250310133553985,
+      "loss": 1.5658,
+      "step": 1284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012472469489479928,
+      "loss": 1.5692,
+      "step": 1285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012441858201754934,
+      "loss": 1.5028,
+      "step": 1286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012411267555966946,
+      "loss": 1.4544,
+      "step": 1287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012380697635661645,
+      "loss": 1.4032,
+      "step": 1288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012350148524328125,
+      "loss": 1.6898,
+      "step": 1289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012319620305398655,
+      "loss": 1.4942,
+      "step": 1290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00012289113062248413,
+      "loss": 1.575,
+      "step": 1291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00012258626878195326,
+      "loss": 1.6769,
+      "step": 1292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001222816183649978,
+      "loss": 1.4289,
+      "step": 1293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00012197718020364437,
+      "loss": 1.7212,
+      "step": 1294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00012167295512933991,
+      "loss": 1.3705,
+      "step": 1295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001213689439729492,
+      "loss": 1.4692,
+      "step": 1296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00012106514756475297,
+      "loss": 1.5929,
+      "step": 1297
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00012076156673444546,
+      "loss": 1.7073,
+      "step": 1298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00012045820231113207,
+      "loss": 1.5234,
+      "step": 1299
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00012015505512332712,
+      "loss": 1.5092,
+      "step": 1300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001198521259989518,
+      "loss": 1.4455,
+      "step": 1301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011954941576533167,
+      "loss": 1.5459,
+      "step": 1302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011924692524919447,
+      "loss": 1.3827,
+      "step": 1303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011894465527666793,
+      "loss": 1.5077,
+      "step": 1304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001186426066732773,
+      "loss": 1.3971,
+      "step": 1305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011834078026394339,
+      "loss": 1.6192,
+      "step": 1306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011803917687298028,
+      "loss": 1.3832,
+      "step": 1307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011773779732409264,
+      "loss": 1.7161,
+      "step": 1308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011743664244037425,
+      "loss": 1.6524,
+      "step": 1309
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011713571304430479,
+      "loss": 1.514,
+      "step": 1310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011683500995774856,
+      "loss": 1.5347,
+      "step": 1311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011653453400195169,
+      "loss": 1.3202,
+      "step": 1312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011623428599753966,
+      "loss": 1.5273,
+      "step": 1313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011593426676451593,
+      "loss": 1.4826,
+      "step": 1314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011563447712225875,
+      "loss": 1.4646,
+      "step": 1315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011533491788951947,
+      "loss": 1.6094,
+      "step": 1316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001150355898844204,
+      "loss": 1.5485,
+      "step": 1317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011473649392445183,
+      "loss": 1.4593,
+      "step": 1318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011443763082647081,
+      "loss": 1.6859,
+      "step": 1319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001141390014066982,
+      "loss": 1.6198,
+      "step": 1320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001138406064807166,
+      "loss": 1.4567,
+      "step": 1321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011354244686346852,
+      "loss": 1.4479,
+      "step": 1322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011324452336925327,
+      "loss": 1.5005,
+      "step": 1323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011294683681172581,
+      "loss": 1.438,
+      "step": 1324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011264938800389382,
+      "loss": 1.5529,
+      "step": 1325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011235217775811544,
+      "loss": 1.4482,
+      "step": 1326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011205520688609778,
+      "loss": 1.5138,
+      "step": 1327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011175847619889359,
+      "loss": 1.4835,
+      "step": 1328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011146198650690021,
+      "loss": 1.3981,
+      "step": 1329
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011116573861985642,
+      "loss": 1.4917,
+      "step": 1330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011086973334684078,
+      "loss": 1.3993,
+      "step": 1331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011057397149626924,
+      "loss": 1.6514,
+      "step": 1332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011027845387589279,
+      "loss": 1.4756,
+      "step": 1333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010998318129279575,
+      "loss": 1.5535,
+      "step": 1334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010968815455339285,
+      "loss": 1.4985,
+      "step": 1335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010939337446342763,
+      "loss": 1.4418,
+      "step": 1336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001090988418279699,
+      "loss": 1.4817,
+      "step": 1337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001088045574514136,
+      "loss": 1.5932,
+      "step": 1338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010851052213747493,
+      "loss": 1.6614,
+      "step": 1339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010821673668918957,
+      "loss": 1.6368,
+      "step": 1340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010792320190891093,
+      "loss": 1.4143,
+      "step": 1341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001076299185983078,
+      "loss": 1.5895,
+      "step": 1342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010733688755836215,
+      "loss": 1.6284,
+      "step": 1343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010704410958936719,
+      "loss": 1.5215,
+      "step": 1344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010675158549092455,
+      "loss": 1.4819,
+      "step": 1345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010645931606194297,
+      "loss": 1.4669,
+      "step": 1346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010616730210063536,
+      "loss": 1.4592,
+      "step": 1347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010587554440451702,
+      "loss": 1.4238,
+      "step": 1348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010558404377040352,
+      "loss": 1.5082,
+      "step": 1349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010529280099440799,
+      "loss": 1.4766,
+      "step": 1350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010500181687193977,
+      "loss": 1.5275,
+      "step": 1351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010471109219770155,
+      "loss": 1.5614,
+      "step": 1352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010442062776568735,
+      "loss": 1.517,
+      "step": 1353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010413042436918084,
+      "loss": 1.4797,
+      "step": 1354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001038404828007522,
+      "loss": 1.6415,
+      "step": 1355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010355080385225713,
+      "loss": 1.5257,
+      "step": 1356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010326138831483367,
+      "loss": 1.5953,
+      "step": 1357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010297223697890062,
+      "loss": 1.5818,
+      "step": 1358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010268335063415536,
+      "loss": 1.4314,
+      "step": 1359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010239473006957112,
+      "loss": 1.3441,
+      "step": 1360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010210637607339584,
+      "loss": 1.2736,
+      "step": 1361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010181828943314899,
+      "loss": 1.3509,
+      "step": 1362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010153047093562009,
+      "loss": 1.5463,
+      "step": 1363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010124292136686624,
+      "loss": 1.5247,
+      "step": 1364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010095564151221002,
+      "loss": 1.4775,
+      "step": 1365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001006686321562377,
+      "loss": 1.5266,
+      "step": 1366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010038189408279643,
+      "loss": 1.3547,
+      "step": 1367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010009542807499266,
+      "loss": 1.6053,
+      "step": 1368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.980923491518976e-05,
+      "loss": 1.4517,
+      "step": 1369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.952331538500583e-05,
+      "loss": 1.6798,
+      "step": 1370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.923767026531195e-05,
+      "loss": 1.4975,
+      "step": 1371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.895230033622945e-05,
+      "loss": 1.492,
+      "step": 1372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.866720637712823e-05,
+      "loss": 1.5128,
+      "step": 1373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.83823891666245e-05,
+      "loss": 1.3858,
+      "step": 1374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.809784948257852e-05,
+      "loss": 1.612,
+      "step": 1375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.781358810209287e-05,
+      "loss": 1.3599,
+      "step": 1376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.752960580150981e-05,
+      "loss": 1.6477,
+      "step": 1377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.724590335640949e-05,
+      "loss": 1.5782,
+      "step": 1378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.696248154160776e-05,
+      "loss": 1.6439,
+      "step": 1379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.667934113115395e-05,
+      "loss": 1.7892,
+      "step": 1380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.63964828983291e-05,
+      "loss": 1.6616,
+      "step": 1381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.611390761564333e-05,
+      "loss": 1.7806,
+      "step": 1382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.583161605483415e-05,
+      "loss": 1.4184,
+      "step": 1383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.554960898686411e-05,
+      "loss": 1.4113,
+      "step": 1384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.526788718191881e-05,
+      "loss": 1.4594,
+      "step": 1385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.498645140940488e-05,
+      "loss": 1.4846,
+      "step": 1386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.470530243794765e-05,
+      "loss": 1.3407,
+      "step": 1387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.442444103538919e-05,
+      "loss": 1.4245,
+      "step": 1388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.414386796878621e-05,
+      "loss": 1.6035,
+      "step": 1389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.38635840044079e-05,
+      "loss": 1.5466,
+      "step": 1390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.358358990773402e-05,
+      "loss": 1.4481,
+      "step": 1391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.330388644345259e-05,
+      "loss": 1.5167,
+      "step": 1392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.302447437545784e-05,
+      "loss": 1.4248,
+      "step": 1393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.274535446684827e-05,
+      "loss": 1.7327,
+      "step": 1394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.246652747992429e-05,
+      "loss": 1.3955,
+      "step": 1395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.218799417618662e-05,
+      "loss": 1.7074,
+      "step": 1396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.190975531633368e-05,
+      "loss": 1.5225,
+      "step": 1397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.163181166025976e-05,
+      "loss": 1.587,
+      "step": 1398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.135416396705294e-05,
+      "loss": 1.3871,
+      "step": 1399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.107681299499304e-05,
+      "loss": 1.5438,
+      "step": 1400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.079975950154942e-05,
+      "loss": 1.6617,
+      "step": 1401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.052300424337917e-05,
+      "loss": 1.4866,
+      "step": 1402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.02465479763247e-05,
+      "loss": 1.4448,
+      "step": 1403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.997039145541193e-05,
+      "loss": 1.5473,
+      "step": 1404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.969453543484811e-05,
+      "loss": 1.6725,
+      "step": 1405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.941898066801979e-05,
+      "loss": 1.5296,
+      "step": 1406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.91437279074909e-05,
+      "loss": 1.5941,
+      "step": 1407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.886877790500042e-05,
+      "loss": 1.5012,
+      "step": 1408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.859413141146049e-05,
+      "loss": 1.5981,
+      "step": 1409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.831978917695438e-05,
+      "loss": 1.3714,
+      "step": 1410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.804575195073435e-05,
+      "loss": 1.4957,
+      "step": 1411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.777202048121977e-05,
+      "loss": 1.5818,
+      "step": 1412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.749859551599488e-05,
+      "loss": 1.5382,
+      "step": 1413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.72254778018068e-05,
+      "loss": 1.3759,
+      "step": 1414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.69526680845636e-05,
+      "loss": 1.4997,
+      "step": 1415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.6680167109332e-05,
+      "loss": 1.5585,
+      "step": 1416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.640797562033586e-05,
+      "loss": 1.5791,
+      "step": 1417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.613609436095353e-05,
+      "loss": 1.5011,
+      "step": 1418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.586452407371619e-05,
+      "loss": 1.4529,
+      "step": 1419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.559326550030567e-05,
+      "loss": 1.385,
+      "step": 1420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.532231938155253e-05,
+      "loss": 1.6166,
+      "step": 1421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.505168645743393e-05,
+      "loss": 1.433,
+      "step": 1422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.478136746707188e-05,
+      "loss": 1.526,
+      "step": 1423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.451136314873076e-05,
+      "loss": 1.5146,
+      "step": 1424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.424167423981565e-05,
+      "loss": 1.4147,
+      "step": 1425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.397230147687023e-05,
+      "loss": 1.6492,
+      "step": 1426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.370324559557463e-05,
+      "loss": 1.6118,
+      "step": 1427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.343450733074388e-05,
+      "loss": 1.6142,
+      "step": 1428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.316608741632519e-05,
+      "loss": 1.5937,
+      "step": 1429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.289798658539658e-05,
+      "loss": 1.3966,
+      "step": 1430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.263020557016447e-05,
+      "loss": 1.5064,
+      "step": 1431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.236274510196196e-05,
+      "loss": 1.4262,
+      "step": 1432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.209560591124654e-05,
+      "loss": 1.2924,
+      "step": 1433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.182878872759854e-05,
+      "loss": 1.6428,
+      "step": 1434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.15622942797186e-05,
+      "loss": 1.4808,
+      "step": 1435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.129612329542609e-05,
+      "loss": 1.5953,
+      "step": 1436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.103027650165685e-05,
+      "loss": 1.682,
+      "step": 1437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.076475462446138e-05,
+      "loss": 1.5173,
+      "step": 1438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.049955838900294e-05,
+      "loss": 1.6062,
+      "step": 1439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.023468851955527e-05,
+      "loss": 1.5373,
+      "step": 1440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.997014573950078e-05,
+      "loss": 1.5742,
+      "step": 1441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.970593077132869e-05,
+      "loss": 1.6135,
+      "step": 1442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.944204433663269e-05,
+      "loss": 1.4727,
+      "step": 1443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.917848715610962e-05,
+      "loss": 1.4902,
+      "step": 1444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.891525994955675e-05,
+      "loss": 1.7681,
+      "step": 1445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.865236343587028e-05,
+      "loss": 1.3804,
+      "step": 1446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.838979833304328e-05,
+      "loss": 1.6395,
+      "step": 1447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.812756535816358e-05,
+      "loss": 1.4848,
+      "step": 1448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.786566522741226e-05,
+      "loss": 1.5668,
+      "step": 1449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.760409865606102e-05,
+      "loss": 1.5328,
+      "step": 1450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.734286635847073e-05,
+      "loss": 1.4281,
+      "step": 1451
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.708196904808935e-05,
+      "loss": 1.5234,
+      "step": 1452
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.682140743744986e-05,
+      "loss": 1.4104,
+      "step": 1453
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.65611822381686e-05,
+      "loss": 1.3889,
+      "step": 1454
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.630129416094296e-05,
+      "loss": 1.5667,
+      "step": 1455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.604174391554973e-05,
+      "loss": 1.4768,
+      "step": 1456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.578253221084297e-05,
+      "loss": 1.6435,
+      "step": 1457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.552365975475217e-05,
+      "loss": 1.5696,
+      "step": 1458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.526512725428043e-05,
+      "loss": 1.7266,
+      "step": 1459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.500693541550224e-05,
+      "loss": 1.4111,
+      "step": 1460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.474908494356179e-05,
+      "loss": 1.5795,
+      "step": 1461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.449157654267093e-05,
+      "loss": 1.5585,
+      "step": 1462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.423441091610727e-05,
+      "loss": 1.3697,
+      "step": 1463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.39775887662125e-05,
+      "loss": 1.5309,
+      "step": 1464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.37211107943898e-05,
+      "loss": 1.5493,
+      "step": 1465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.346497770110282e-05,
+      "loss": 1.4651,
+      "step": 1466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.320919018587304e-05,
+      "loss": 1.4814,
+      "step": 1467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.295374894727818e-05,
+      "loss": 1.3825,
+      "step": 1468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.269865468295047e-05,
+      "loss": 1.4483,
+      "step": 1469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.244390808957409e-05,
+      "loss": 1.4974,
+      "step": 1470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.218950986288415e-05,
+      "loss": 1.418,
+      "step": 1471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.193546069766411e-05,
+      "loss": 1.3873,
+      "step": 1472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.168176128774399e-05,
+      "loss": 1.5701,
+      "step": 1473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.142841232599906e-05,
+      "loss": 1.5859,
+      "step": 1474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.117541450434686e-05,
+      "loss": 1.7759,
+      "step": 1475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.092276851374647e-05,
+      "loss": 1.6146,
+      "step": 1476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.067047504419579e-05,
+      "loss": 1.5228,
+      "step": 1477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.041853478473e-05,
+      "loss": 1.6275,
+      "step": 1478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.016694842341984e-05,
+      "loss": 1.6343,
+      "step": 1479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.991571664736916e-05,
+      "loss": 1.3821,
+      "step": 1480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.966484014271373e-05,
+      "loss": 1.494,
+      "step": 1481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.941431959461889e-05,
+      "loss": 1.4935,
+      "step": 1482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.916415568727777e-05,
+      "loss": 1.4673,
+      "step": 1483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.891434910390979e-05,
+      "loss": 1.5045,
+      "step": 1484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.8664900526758e-05,
+      "loss": 1.5553,
+      "step": 1485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.841581063708819e-05,
+      "loss": 1.3835,
+      "step": 1486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.816708011518623e-05,
+      "loss": 1.4642,
+      "step": 1487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.791870964035658e-05,
+      "loss": 1.7471,
+      "step": 1488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.76706998909206e-05,
+      "loss": 1.5005,
+      "step": 1489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.742305154421408e-05,
+      "loss": 1.7341,
+      "step": 1490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.717576527658618e-05,
+      "loss": 1.6091,
+      "step": 1491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.692884176339694e-05,
+      "loss": 1.3284,
+      "step": 1492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.668228167901573e-05,
+      "loss": 1.4199,
+      "step": 1493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.643608569681958e-05,
+      "loss": 1.6035,
+      "step": 1494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.61902544891907e-05,
+      "loss": 1.4549,
+      "step": 1495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.594478872751555e-05,
+      "loss": 1.4991,
+      "step": 1496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.569968908218207e-05,
+      "loss": 1.6751,
+      "step": 1497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.545495622257869e-05,
+      "loss": 1.4197,
+      "step": 1498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.521059081709191e-05,
+      "loss": 1.4976,
+      "step": 1499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.496659353310466e-05,
+      "loss": 1.3893,
+      "step": 1500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.472296503699473e-05,
+      "loss": 1.5723,
+      "step": 1501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.447970599413231e-05,
+      "loss": 1.4579,
+      "step": 1502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.423681706887903e-05,
+      "loss": 1.5235,
+      "step": 1503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.399429892458541e-05,
+      "loss": 1.3772,
+      "step": 1504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.375215222358935e-05,
+      "loss": 1.4735,
+      "step": 1505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.351037762721457e-05,
+      "loss": 1.5303,
+      "step": 1506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.326897579576807e-05,
+      "loss": 1.3491,
+      "step": 1507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.302794738853927e-05,
+      "loss": 1.5621,
+      "step": 1508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.27872930637975e-05,
+      "loss": 1.5875,
+      "step": 1509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.254701347879035e-05,
+      "loss": 1.4622,
+      "step": 1510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.230710928974232e-05,
+      "loss": 1.6829,
+      "step": 1511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.206758115185216e-05,
+      "loss": 1.5249,
+      "step": 1512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.18284297192921e-05,
+      "loss": 1.5523,
+      "step": 1513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.15896556452052e-05,
+      "loss": 1.4929,
+      "step": 1514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.135125958170405e-05,
+      "loss": 1.2468,
+      "step": 1515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.111324217986898e-05,
+      "loss": 1.4509,
+      "step": 1516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.087560408974584e-05,
+      "loss": 1.4516,
+      "step": 1517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.06383459603449e-05,
+      "loss": 1.6536,
+      "step": 1518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.040146843963852e-05,
+      "loss": 1.5844,
+      "step": 1519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.0164972174559586e-05,
+      "loss": 1.3181,
+      "step": 1520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.992885781099997e-05,
+      "loss": 1.4771,
+      "step": 1521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.96931259938081e-05,
+      "loss": 1.4503,
+      "step": 1522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9457777366788084e-05,
+      "loss": 1.5759,
+      "step": 1523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.922281257269726e-05,
+      "loss": 1.3737,
+      "step": 1524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8988232253244704e-05,
+      "loss": 1.5489,
+      "step": 1525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.875403704908964e-05,
+      "loss": 1.7822,
+      "step": 1526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.852022759983913e-05,
+      "loss": 1.3443,
+      "step": 1527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.828680454404714e-05,
+      "loss": 1.3229,
+      "step": 1528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.805376851921212e-05,
+      "loss": 1.4909,
+      "step": 1529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7821120161775435e-05,
+      "loss": 1.5012,
+      "step": 1530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.758886010712003e-05,
+      "loss": 1.4465,
+      "step": 1531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.73569889895679e-05,
+      "loss": 1.3189,
+      "step": 1532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.712550744237923e-05,
+      "loss": 1.5343,
+      "step": 1533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.6894416097750034e-05,
+      "loss": 1.4563,
+      "step": 1534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.666371558681063e-05,
+      "loss": 1.4985,
+      "step": 1535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.643340653962421e-05,
+      "loss": 1.3521,
+      "step": 1536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.6203489585184375e-05,
+      "loss": 1.3999,
+      "step": 1537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.5973965351414346e-05,
+      "loss": 1.3275,
+      "step": 1538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.5744834465164564e-05,
+      "loss": 1.511,
+      "step": 1539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.551609755221112e-05,
+      "loss": 1.3924,
+      "step": 1540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.528775523725449e-05,
+      "loss": 1.5096,
+      "step": 1541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.5059808143916935e-05,
+      "loss": 1.5209,
+      "step": 1542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.483225689474183e-05,
+      "loss": 1.4124,
+      "step": 1543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.460510211119119e-05,
+      "loss": 1.5415,
+      "step": 1544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4378344413644265e-05,
+      "loss": 1.4963,
+      "step": 1545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.415198442139604e-05,
+      "loss": 1.3565,
+      "step": 1546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.392602275265497e-05,
+      "loss": 1.6187,
+      "step": 1547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.370046002454201e-05,
+      "loss": 1.387,
+      "step": 1548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.347529685308836e-05,
+      "loss": 1.5391,
+      "step": 1549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.325053385323402e-05,
+      "loss": 1.5568,
+      "step": 1550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.302617163882626e-05,
+      "loss": 1.4682,
+      "step": 1551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.280221082261745e-05,
+      "loss": 1.4907,
+      "step": 1552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.257865201626404e-05,
+      "loss": 1.531,
+      "step": 1553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.235549583032435e-05,
+      "loss": 1.5039,
+      "step": 1554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.213274287425711e-05,
+      "loss": 1.5405,
+      "step": 1555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.1910393756420017e-05,
+      "loss": 1.5242,
+      "step": 1556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.168844908406749e-05,
+      "loss": 1.4804,
+      "step": 1557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.146690946334969e-05,
+      "loss": 1.4277,
+      "step": 1558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.12457754993104e-05,
+      "loss": 1.3221,
+      "step": 1559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.102504779588555e-05,
+      "loss": 1.6861,
+      "step": 1560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0804726955901505e-05,
+      "loss": 1.5792,
+      "step": 1561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.058481358107343e-05,
+      "loss": 1.4429,
+      "step": 1562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.036530827200392e-05,
+      "loss": 1.6149,
+      "step": 1563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.014621162818061e-05,
+      "loss": 1.4497,
+      "step": 1564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9927524247975555e-05,
+      "loss": 1.4262,
+      "step": 1565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9709246728642747e-05,
+      "loss": 1.4911,
+      "step": 1566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.94913796663169e-05,
+      "loss": 1.4118,
+      "step": 1567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.927392365601189e-05,
+      "loss": 1.5412,
+      "step": 1568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.905687929161862e-05,
+      "loss": 1.5675,
+      "step": 1569
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.884024716590412e-05,
+      "loss": 1.5406,
+      "step": 1570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.862402787050935e-05,
+      "loss": 1.5776,
+      "step": 1571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.840822199594781e-05,
+      "loss": 1.5056,
+      "step": 1572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.819283013160412e-05,
+      "loss": 1.6191,
+      "step": 1573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.79778528657318e-05,
+      "loss": 1.3349,
+      "step": 1574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7763290785452476e-05,
+      "loss": 1.4504,
+      "step": 1575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.754914447675363e-05,
+      "loss": 1.5173,
+      "step": 1576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7335414524487245e-05,
+      "loss": 1.604,
+      "step": 1577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7122101512368424e-05,
+      "loss": 1.4894,
+      "step": 1578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.690920602297315e-05,
+      "loss": 1.5064,
+      "step": 1579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.669672863773758e-05,
+      "loss": 1.4289,
+      "step": 1580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.648466993695566e-05,
+      "loss": 1.3812,
+      "step": 1581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.627303049977796e-05,
+      "loss": 1.6637,
+      "step": 1582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.60618109042102e-05,
+      "loss": 1.4851,
+      "step": 1583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.58510117271111e-05,
+      "loss": 1.467,
+      "step": 1584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5640633544191545e-05,
+      "loss": 1.5849,
+      "step": 1585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5430676930012414e-05,
+      "loss": 1.4256,
+      "step": 1586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.522114245798332e-05,
+      "loss": 1.5615,
+      "step": 1587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5012030700361085e-05,
+      "loss": 1.5331,
+      "step": 1588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.480334222824778e-05,
+      "loss": 1.3991,
+      "step": 1589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4595077611589765e-05,
+      "loss": 1.5719,
+      "step": 1590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.438723741917559e-05,
+      "loss": 1.3425,
+      "step": 1591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.417982221863473e-05,
+      "loss": 1.6173,
+      "step": 1592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.3972832576436096e-05,
+      "loss": 1.5314,
+      "step": 1593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.376626905788608e-05,
+      "loss": 1.4374,
+      "step": 1594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.3560132227127606e-05,
+      "loss": 1.5889,
+      "step": 1595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.335442264713807e-05,
+      "loss": 1.3955,
+      "step": 1596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.314914087972808e-05,
+      "loss": 1.5982,
+      "step": 1597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.294428748553987e-05,
+      "loss": 1.4567,
+      "step": 1598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2739863024045625e-05,
+      "loss": 1.7278,
+      "step": 1599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2535868053546326e-05,
+      "loss": 1.4764,
+      "step": 1600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2332303131169756e-05,
+      "loss": 1.4648,
+      "step": 1601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.212916881286926e-05,
+      "loss": 1.6258,
+      "step": 1602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.19264656534222e-05,
+      "loss": 1.6907,
+      "step": 1603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1724194206428326e-05,
+      "loss": 1.5267,
+      "step": 1604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.15223550243085e-05,
+      "loss": 1.2853,
+      "step": 1605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.132094865830289e-05,
+      "loss": 1.3756,
+      "step": 1606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1119975658469635e-05,
+      "loss": 1.5751,
+      "step": 1607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.091943657368333e-05,
+      "loss": 1.6317,
+      "step": 1608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0719331951633486e-05,
+      "loss": 1.5201,
+      "step": 1609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.051966233882314e-05,
+      "loss": 1.539,
+      "step": 1610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0320428280567213e-05,
+      "loss": 1.4511,
+      "step": 1611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.012163032099106e-05,
+      "loss": 1.5473,
+      "step": 1612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.992326900302907e-05,
+      "loss": 1.4899,
+      "step": 1613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9725344868423054e-05,
+      "loss": 1.6457,
+      "step": 1614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9527858457721e-05,
+      "loss": 1.5946,
+      "step": 1615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.933081031027528e-05,
+      "loss": 1.645,
+      "step": 1616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9134200964241344e-05,
+      "loss": 1.2908,
+      "step": 1617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.893803095657631e-05,
+      "loss": 1.5992,
+      "step": 1618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.874230082303729e-05,
+      "loss": 1.3276,
+      "step": 1619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8547011098180306e-05,
+      "loss": 1.441,
+      "step": 1620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.835216231535834e-05,
+      "loss": 1.4115,
+      "step": 1621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.815775500672028e-05,
+      "loss": 1.4233,
+      "step": 1622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.796378970320917e-05,
+      "loss": 1.5179,
+      "step": 1623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7770266934561e-05,
+      "loss": 1.5091,
+      "step": 1624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.757718722930321e-05,
+      "loss": 1.5677,
+      "step": 1625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7384551114753076e-05,
+      "loss": 1.5057,
+      "step": 1626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.719235911701642e-05,
+      "loss": 1.4919,
+      "step": 1627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7000611760986193e-05,
+      "loss": 1.3167,
+      "step": 1628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6809309570340946e-05,
+      "loss": 1.5101,
+      "step": 1629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.661845306754341e-05,
+      "loss": 1.4361,
+      "step": 1630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6428042773839244e-05,
+      "loss": 1.5722,
+      "step": 1631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6238079209255346e-05,
+      "loss": 1.3537,
+      "step": 1632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.604856289259857e-05,
+      "loss": 1.4854,
+      "step": 1633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5859494341454344e-05,
+      "loss": 1.4207,
+      "step": 1634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.56708740721851e-05,
+      "loss": 1.3087,
+      "step": 1635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.548270259992921e-05,
+      "loss": 1.4631,
+      "step": 1636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5294980438599066e-05,
+      "loss": 1.5242,
+      "step": 1637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.510770810088015e-05,
+      "loss": 1.5441,
+      "step": 1638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.492088609822934e-05,
+      "loss": 1.5334,
+      "step": 1639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.473451494087363e-05,
+      "loss": 1.3592,
+      "step": 1640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.454859513780868e-05,
+      "loss": 1.4713,
+      "step": 1641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.436312719679757e-05,
+      "loss": 1.4357,
+      "step": 1642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.417811162436926e-05,
+      "loss": 1.3486,
+      "step": 1643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3993548925817166e-05,
+      "loss": 1.3834,
+      "step": 1644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3809439605197955e-05,
+      "loss": 1.312,
+      "step": 1645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.362578416533002e-05,
+      "loss": 1.4625,
+      "step": 1646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.344258310779229e-05,
+      "loss": 1.4697,
+      "step": 1647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3259836932922605e-05,
+      "loss": 1.4243,
+      "step": 1648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.307754613981655e-05,
+      "loss": 1.6532,
+      "step": 1649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.289571122632595e-05,
+      "loss": 1.5521,
+      "step": 1650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.271433268905764e-05,
+      "loss": 1.5757,
+      "step": 1651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2533411023372085e-05,
+      "loss": 1.4505,
+      "step": 1652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.235294672338196e-05,
+      "loss": 1.5972,
+      "step": 1653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2172940281950794e-05,
+      "loss": 1.4879,
+      "step": 1654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.199339219069173e-05,
+      "loss": 1.4165,
+      "step": 1655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.181430293996601e-05,
+      "loss": 1.3632,
+      "step": 1656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.163567301888195e-05,
+      "loss": 1.5905,
+      "step": 1657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.145750291529319e-05,
+      "loss": 1.4664,
+      "step": 1658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.127979311579765e-05,
+      "loss": 1.6243,
+      "step": 1659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.110254410573614e-05,
+      "loss": 1.3478,
+      "step": 1660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.092575636919097e-05,
+      "loss": 1.7581,
+      "step": 1661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.074943038898464e-05,
+      "loss": 1.4673,
+      "step": 1662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.057356664667872e-05,
+      "loss": 1.5611,
+      "step": 1663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0398165622572205e-05,
+      "loss": 1.3356,
+      "step": 1664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0223227795700416e-05,
+      "loss": 1.4599,
+      "step": 1665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0048753643833637e-05,
+      "loss": 1.5329,
+      "step": 1666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9874743643475776e-05,
+      "loss": 1.4561,
+      "step": 1667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.970119826986326e-05,
+      "loss": 1.497,
+      "step": 1668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.95281179969634e-05,
+      "loss": 1.505,
+      "step": 1669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.935550329747334e-05,
+      "loss": 1.3881,
+      "step": 1670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9183354642818738e-05,
+      "loss": 1.5118,
+      "step": 1671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9011672503152355e-05,
+      "loss": 1.5736,
+      "step": 1672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.884045734735301e-05,
+      "loss": 1.5678,
+      "step": 1673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8669709643024e-05,
+      "loss": 1.4046,
+      "step": 1674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8499429856492055e-05,
+      "loss": 1.5102,
+      "step": 1675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8329618452805928e-05,
+      "loss": 1.4656,
+      "step": 1676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8160275895735177e-05,
+      "loss": 1.6555,
+      "step": 1677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7991402647769026e-05,
+      "loss": 1.5405,
+      "step": 1678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7822999170114816e-05,
+      "loss": 1.5596,
+      "step": 1679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7655065922696954e-05,
+      "loss": 1.5635,
+      "step": 1680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7487603364155655e-05,
+      "loss": 1.3659,
+      "step": 1681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7320611951845543e-05,
+      "loss": 1.3942,
+      "step": 1682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7154092141834664e-05,
+      "loss": 1.5691,
+      "step": 1683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6988044388902923e-05,
+      "loss": 1.5243,
+      "step": 1684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6822469146541028e-05,
+      "loss": 1.3723,
+      "step": 1685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6657366866949308e-05,
+      "loss": 1.454,
+      "step": 1686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.649273800103622e-05,
+      "loss": 1.567,
+      "step": 1687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6328582998417516e-05,
+      "loss": 1.3523,
+      "step": 1688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6164902307414596e-05,
+      "loss": 1.4375,
+      "step": 1689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6001696375053563e-05,
+      "loss": 1.434,
+      "step": 1690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5838965647063873e-05,
+      "loss": 1.4511,
+      "step": 1691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5676710567877148e-05,
+      "loss": 1.4878,
+      "step": 1692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5514931580626144e-05,
+      "loss": 1.5302,
+      "step": 1693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.535362912714301e-05,
+      "loss": 1.633,
+      "step": 1694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5192803647958817e-05,
+      "loss": 1.4009,
+      "step": 1695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5032455582301763e-05,
+      "loss": 1.4321,
+      "step": 1696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4872585368096225e-05,
+      "loss": 1.4418,
+      "step": 1697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4713193441961636e-05,
+      "loss": 1.479,
+      "step": 1698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4554280239210915e-05,
+      "loss": 1.3998,
+      "step": 1699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.439584619384989e-05,
+      "loss": 1.4941,
+      "step": 1700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4237891738575534e-05,
+      "loss": 1.5128,
+      "step": 1701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.408041730477506e-05,
+      "loss": 1.1936,
+      "step": 1702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3923423322524818e-05,
+      "loss": 1.4058,
+      "step": 1703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3766910220588767e-05,
+      "loss": 1.6485,
+      "step": 1704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3610878426417803e-05,
+      "loss": 1.3647,
+      "step": 1705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3455328366148188e-05,
+      "loss": 1.5596,
+      "step": 1706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3300260464600497e-05,
+      "loss": 1.6531,
+      "step": 1707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3145675145278655e-05,
+      "loss": 1.5407,
+      "step": 1708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2991572830368325e-05,
+      "loss": 1.4708,
+      "step": 1709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.283795394073638e-05,
+      "loss": 1.5763,
+      "step": 1710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2684818895929194e-05,
+      "loss": 1.5763,
+      "step": 1711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2532168114171757e-05,
+      "loss": 1.4242,
+      "step": 1712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.238000201236654e-05,
+      "loss": 1.6482,
+      "step": 1713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2228321006092247e-05,
+      "loss": 1.5563,
+      "step": 1714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.207712550960286e-05,
+      "loss": 1.641,
+      "step": 1715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1926415935826295e-05,
+      "loss": 1.5114,
+      "step": 1716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1776192696363372e-05,
+      "loss": 1.5597,
+      "step": 1717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1626456201486712e-05,
+      "loss": 1.581,
+      "step": 1718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1477206860139564e-05,
+      "loss": 1.4744,
+      "step": 1719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1328445079934835e-05,
+      "loss": 1.4128,
+      "step": 1720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1180171267153726e-05,
+      "loss": 1.5995,
+      "step": 1721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1032385826744805e-05,
+      "loss": 1.3006,
+      "step": 1722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0885089162322857e-05,
+      "loss": 1.7836,
+      "step": 1723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0738281676167736e-05,
+      "loss": 1.4046,
+      "step": 1724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.059196376922348e-05,
+      "loss": 1.3634,
+      "step": 1725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0446135841096713e-05,
+      "loss": 1.3388,
+      "step": 1726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0300798290056223e-05,
+      "loss": 1.4948,
+      "step": 1727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0155951513031357e-05,
+      "loss": 1.456,
+      "step": 1728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0011595905611103e-05,
+      "loss": 1.3373,
+      "step": 1729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9867731862043206e-05,
+      "loss": 1.7381,
+      "step": 1730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9724359775232592e-05,
+      "loss": 1.5819,
+      "step": 1731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.958148003674094e-05,
+      "loss": 1.6322,
+      "step": 1732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9439093036785105e-05,
+      "loss": 1.42,
+      "step": 1733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9297199164236178e-05,
+      "loss": 1.5373,
+      "step": 1734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.915579880661873e-05,
+      "loss": 1.7121,
+      "step": 1735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9014892350109113e-05,
+      "loss": 1.5765,
+      "step": 1736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.887448017953519e-05,
+      "loss": 1.7314,
+      "step": 1737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8734562678374635e-05,
+      "loss": 1.3345,
+      "step": 1738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8595140228754194e-05,
+      "loss": 1.3661,
+      "step": 1739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.845621321144866e-05,
+      "loss": 1.548,
+      "step": 1740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8317782005879615e-05,
+      "loss": 1.433,
+      "step": 1741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.817984699011468e-05,
+      "loss": 1.3648,
+      "step": 1742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8042408540866272e-05,
+      "loss": 1.4532,
+      "step": 1743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7905467033490585e-05,
+      "loss": 1.5611,
+      "step": 1744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7769022841986805e-05,
+      "loss": 1.6119,
+      "step": 1745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7633076338995647e-05,
+      "loss": 1.5655,
+      "step": 1746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.749762789579883e-05,
+      "loss": 1.536,
+      "step": 1747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7362677882317713e-05,
+      "loss": 1.423,
+      "step": 1748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7228226667112368e-05,
+      "loss": 1.4794,
+      "step": 1749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7094274617380755e-05,
+      "loss": 1.4831,
+      "step": 1750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6960822098957374e-05,
+      "loss": 1.405,
+      "step": 1751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6827869476312607e-05,
+      "loss": 1.45,
+      "step": 1752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.669541711255156e-05,
+      "loss": 1.6667,
+      "step": 1753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656346536941298e-05,
+      "loss": 1.631,
+      "step": 1754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6432014607268555e-05,
+      "loss": 1.4798,
+      "step": 1755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6301065185121554e-05,
+      "loss": 1.4823,
+      "step": 1756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6170617460606197e-05,
+      "loss": 1.486,
+      "step": 1757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6040671789986472e-05,
+      "loss": 1.3193,
+      "step": 1758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5911228528155164e-05,
+      "loss": 1.4193,
+      "step": 1759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5782288028633085e-05,
+      "loss": 1.5819,
+      "step": 1760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5653850643567725e-05,
+      "loss": 1.6686,
+      "step": 1761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5525916723732737e-05,
+      "loss": 1.4264,
+      "step": 1762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5398486618526674e-05,
+      "loss": 1.4875,
+      "step": 1763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5271560675972128e-05,
+      "loss": 1.5749,
+      "step": 1764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5145139242714857e-05,
+      "loss": 1.4587,
+      "step": 1765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.501922266402258e-05,
+      "loss": 1.4071,
+      "step": 1766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4893811283784443e-05,
+      "loss": 1.5193,
+      "step": 1767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4768905444509684e-05,
+      "loss": 1.5266,
+      "step": 1768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4644505487326898e-05,
+      "loss": 1.658,
+      "step": 1769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4520611751983182e-05,
+      "loss": 1.5443,
+      "step": 1770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4397224576842916e-05,
+      "loss": 1.4865,
+      "step": 1771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4274344298887166e-05,
+      "loss": 1.7879,
+      "step": 1772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.415197125371257e-05,
+      "loss": 1.4343,
+      "step": 1773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.403010577553041e-05,
+      "loss": 1.5379,
+      "step": 1774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3908748197165944e-05,
+      "loss": 1.3095,
+      "step": 1775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3787898850057024e-05,
+      "loss": 1.3313,
+      "step": 1776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3667558064253727e-05,
+      "loss": 1.5196,
+      "step": 1777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.354772616841713e-05,
+      "loss": 1.4472,
+      "step": 1778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3428403489818398e-05,
+      "loss": 1.3119,
+      "step": 1779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3309590354338164e-05,
+      "loss": 1.5333,
+      "step": 1780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3191287086465243e-05,
+      "loss": 1.6633,
+      "step": 1781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3073494009296139e-05,
+      "loss": 1.362,
+      "step": 1782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2956211444533827e-05,
+      "loss": 1.3867,
+      "step": 1783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2839439712487133e-05,
+      "loss": 1.3955,
+      "step": 1784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2723179132069817e-05,
+      "loss": 1.5394,
+      "step": 1785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2607430020799405e-05,
+      "loss": 1.4558,
+      "step": 1786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2492192694796822e-05,
+      "loss": 1.4158,
+      "step": 1787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.237746746878503e-05,
+      "loss": 1.4332,
+      "step": 1788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2263254656088618e-05,
+      "loss": 1.4337,
+      "step": 1789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.214955456863256e-05,
+      "loss": 1.4629,
+      "step": 1790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2036367516941594e-05,
+      "loss": 1.4211,
+      "step": 1791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1923693810139424e-05,
+      "loss": 1.3978,
+      "step": 1792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1811533755947502e-05,
+      "loss": 1.5042,
+      "step": 1793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1699887660684705e-05,
+      "loss": 1.6496,
+      "step": 1794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1588755829266152e-05,
+      "loss": 1.4824,
+      "step": 1795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.147813856520239e-05,
+      "loss": 1.5267,
+      "step": 1796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1368036170598828e-05,
+      "loss": 1.491,
+      "step": 1797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1258448946154444e-05,
+      "loss": 1.4966,
+      "step": 1798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1149377191161491e-05,
+      "loss": 1.555,
+      "step": 1799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1040821203504336e-05,
+      "loss": 1.4457,
+      "step": 1800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.093278127965871e-05,
+      "loss": 1.5241,
+      "step": 1801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0825257714691029e-05,
+      "loss": 1.357,
+      "step": 1802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0718250802257302e-05,
+      "loss": 1.3904,
+      "step": 1803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0611760834602757e-05,
+      "loss": 1.3859,
+      "step": 1804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0505788102560598e-05,
+      "loss": 1.2844,
+      "step": 1805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0400332895551513e-05,
+      "loss": 1.5293,
+      "step": 1806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0295395501582805e-05,
+      "loss": 1.5423,
+      "step": 1807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0190976207247472e-05,
+      "loss": 1.511,
+      "step": 1808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.008707529772368e-05,
+      "loss": 1.3079,
+      "step": 1809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.983693056773757e-06,
+      "loss": 1.5143,
+      "step": 1810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.880829766743471e-06,
+      "loss": 1.6085,
+      "step": 1811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.778485708561436e-06,
+      "loss": 1.4507,
+      "step": 1812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.676661161737977e-06,
+      "loss": 1.4321,
+      "step": 1813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.575356404364799e-06,
+      "loss": 1.5854,
+      "step": 1814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.4745717131139e-06,
+      "loss": 1.721,
+      "step": 1815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.374307363236946e-06,
+      "loss": 1.2434,
+      "step": 1816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.274563628564581e-06,
+      "loss": 1.498,
+      "step": 1817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.17534078150546e-06,
+      "loss": 1.68,
+      "step": 1818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.076639093045814e-06,
+      "loss": 1.4945,
+      "step": 1819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.978458832748437e-06,
+      "loss": 1.4391,
+      "step": 1820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.880800268752043e-06,
+      "loss": 1.4764,
+      "step": 1821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.783663667770703e-06,
+      "loss": 1.6106,
+      "step": 1822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.687049295092698e-06,
+      "loss": 1.5949,
+      "step": 1823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.590957414580336e-06,
+      "loss": 1.5353,
+      "step": 1824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.495388288668782e-06,
+      "loss": 1.5349,
+      "step": 1825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.400342178365562e-06,
+      "loss": 1.3926,
+      "step": 1826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.305819343249788e-06,
+      "loss": 1.7055,
+      "step": 1827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.211820041471475e-06,
+      "loss": 1.5366,
+      "step": 1828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.118344529750843e-06,
+      "loss": 1.4641,
+      "step": 1829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.025393063377574e-06,
+      "loss": 1.4215,
+      "step": 1830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.932965896210109e-06,
+      "loss": 1.4099,
+      "step": 1831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.841063280675042e-06,
+      "loss": 1.5401,
+      "step": 1832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.74968546776631e-06,
+      "loss": 1.4222,
+      "step": 1833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.658832707044638e-06,
+      "loss": 1.5288,
+      "step": 1834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.568505246636748e-06,
+      "loss": 1.634,
+      "step": 1835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.4787033332346825e-06,
+      "loss": 1.2635,
+      "step": 1836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.389427212095234e-06,
+      "loss": 1.526,
+      "step": 1837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.3006771270391685e-06,
+      "loss": 1.5969,
+      "step": 1838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.212453320450685e-06,
+      "loss": 1.4708,
+      "step": 1839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.124756033276536e-06,
+      "loss": 1.4838,
+      "step": 1840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.037585505025601e-06,
+      "loss": 1.5253,
+      "step": 1841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.950941973768132e-06,
+      "loss": 1.4627,
+      "step": 1842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.8648256761350224e-06,
+      "loss": 1.5297,
+      "step": 1843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.779236847317383e-06,
+      "loss": 1.5432,
+      "step": 1844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.694175721065632e-06,
+      "loss": 1.5786,
+      "step": 1845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.6096425296890755e-06,
+      "loss": 1.3737,
+      "step": 1846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.525637504055127e-06,
+      "loss": 1.5205,
+      "step": 1847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.442160873588732e-06,
+      "loss": 1.5186,
+      "step": 1848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.359212866271836e-06,
+      "loss": 1.4775,
+      "step": 1849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.2767937086425366e-06,
+      "loss": 1.4804,
+      "step": 1850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.19490362579469e-06,
+      "loss": 1.5721,
+      "step": 1851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.1135428413771515e-06,
+      "loss": 1.5261,
+      "step": 1852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.032711577593264e-06,
+      "loss": 1.4337,
+      "step": 1853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.952410055200153e-06,
+      "loss": 1.5172,
+      "step": 1854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.872638493508231e-06,
+      "loss": 1.4595,
+      "step": 1855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.793397110380516e-06,
+      "loss": 1.4478,
+      "step": 1856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.714686122232027e-06,
+      "loss": 1.4076,
+      "step": 1857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.636505744029252e-06,
+      "loss": 1.6518,
+      "step": 1858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.558856189289596e-06,
+      "loss": 1.3948,
+      "step": 1859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.481737670080622e-06,
+      "loss": 1.5249,
+      "step": 1860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.4051503970197205e-06,
+      "loss": 1.2953,
+      "step": 1861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.329094579273308e-06,
+      "loss": 1.3525,
+      "step": 1862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2535704245563866e-06,
+      "loss": 1.5081,
+      "step": 1863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.178578139131918e-06,
+      "loss": 1.6195,
+      "step": 1864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.104117927810314e-06,
+      "loss": 1.5943,
+      "step": 1865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.030189993948841e-06,
+      "loss": 1.2921,
+      "step": 1866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.95679453945106e-06,
+      "loss": 1.5332,
+      "step": 1867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.883931764766292e-06,
+      "loss": 1.4457,
+      "step": 1868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8116018688890265e-06,
+      "loss": 1.4234,
+      "step": 1869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.73980504935847e-06,
+      "loss": 1.5769,
+      "step": 1870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6685415022579286e-06,
+      "loss": 1.5136,
+      "step": 1871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.597811422214293e-06,
+      "loss": 1.6175,
+      "step": 1872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.527615002397489e-06,
+      "loss": 1.3397,
+      "step": 1873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4579524345200295e-06,
+      "loss": 1.6076,
+      "step": 1874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.388823908836348e-06,
+      "loss": 1.5147,
+      "step": 1875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.320229614142468e-06,
+      "loss": 1.571,
+      "step": 1876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.252169737775291e-06,
+      "loss": 1.8309,
+      "step": 1877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.184644465612198e-06,
+      "loss": 1.5114,
+      "step": 1878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1176539820705575e-06,
+      "loss": 1.5716,
+      "step": 1879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.051198470107109e-06,
+      "loss": 1.496,
+      "step": 1880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.98527811121765e-06,
+      "loss": 1.4383,
+      "step": 1881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.919893085436322e-06,
+      "loss": 1.4402,
+      "step": 1882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.855043571335282e-06,
+      "loss": 1.4978,
+      "step": 1883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7907297460241465e-06,
+      "loss": 1.3977,
+      "step": 1884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7269517851494797e-06,
+      "loss": 1.5101,
+      "step": 1885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6637098628944598e-06,
+      "loss": 1.776,
+      "step": 1886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6010041519782157e-06,
+      "loss": 1.4289,
+      "step": 1887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5388348236554016e-06,
+      "loss": 1.5133,
+      "step": 1888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4772020477158663e-06,
+      "loss": 1.3617,
+      "step": 1889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.416105992484009e-06,
+      "loss": 1.3404,
+      "step": 1890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3555468248184006e-06,
+      "loss": 1.4797,
+      "step": 1891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2955247101114082e-06,
+      "loss": 1.539,
+      "step": 1892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.236039812288594e-06,
+      "loss": 1.4868,
+      "step": 1893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1770922938082926e-06,
+      "loss": 1.4764,
+      "step": 1894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1186823156613254e-06,
+      "loss": 1.3227,
+      "step": 1895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0608100373703095e-06,
+      "loss": 1.6015,
+      "step": 1896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0034756169894816e-06,
+      "loss": 1.5561,
+      "step": 1897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.946679211104053e-06,
+      "loss": 1.5038,
+      "step": 1898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8904209748299436e-06,
+      "loss": 1.3569,
+      "step": 1899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8347010618132275e-06,
+      "loss": 1.6143,
+      "step": 1900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.779519624229776e-06,
+      "loss": 1.3514,
+      "step": 1901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7248768127848824e-06,
+      "loss": 1.549,
+      "step": 1902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.670772776712771e-06,
+      "loss": 1.4779,
+      "step": 1903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6172076637762e-06,
+      "loss": 1.599,
+      "step": 1904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5641816202661485e-06,
+      "loss": 1.401,
+      "step": 1905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5116947910012844e-06,
+      "loss": 1.7322,
+      "step": 1906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4597473193276986e-06,
+      "loss": 1.4035,
+      "step": 1907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4083393471183936e-06,
+      "loss": 1.6523,
+      "step": 1908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3574710147729717e-06,
+      "loss": 1.4911,
+      "step": 1909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.30714246121726e-06,
+      "loss": 1.4443,
+      "step": 1910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.257353823902819e-06,
+      "loss": 1.5876,
+      "step": 1911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2081052388067902e-06,
+      "loss": 1.4541,
+      "step": 1912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.159396840431249e-06,
+      "loss": 1.4649,
+      "step": 1913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1112287618030523e-06,
+      "loss": 1.4844,
+      "step": 1914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0636011344733696e-06,
+      "loss": 1.5646,
+      "step": 1915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0165140885173516e-06,
+      "loss": 1.2648,
+      "step": 1916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.969967752533819e-06,
+      "loss": 1.5677,
+      "step": 1917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9239622536447953e-06,
+      "loss": 1.4764,
+      "step": 1918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8784977174952646e-06,
+      "loss": 1.51,
+      "step": 1919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8335742682527912e-06,
+      "loss": 1.4352,
+      "step": 1920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7891920286072117e-06,
+      "loss": 1.5023,
+      "step": 1921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7453511197702554e-06,
+      "loss": 1.4443,
+      "step": 1922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.702051661475168e-06,
+      "loss": 1.3801,
+      "step": 1923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6592937719765778e-06,
+      "loss": 1.5101,
+      "step": 1924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6170775680499183e-06,
+      "loss": 1.323,
+      "step": 1925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.575403164991296e-06,
+      "loss": 1.1511,
+      "step": 1926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5342706766171112e-06,
+      "loss": 1.4834,
+      "step": 1927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4936802152637264e-06,
+      "loss": 1.3955,
+      "step": 1928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4536318917871773e-06,
+      "loss": 1.4266,
+      "step": 1929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4141258155629278e-06,
+      "loss": 1.4888,
+      "step": 1930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.375162094485427e-06,
+      "loss": 1.4385,
+      "step": 1931
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.33674083496802e-06,
+      "loss": 1.6035,
+      "step": 1932
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2988621419424141e-06,
+      "loss": 1.3936,
+      "step": 1933
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2615261188586136e-06,
+      "loss": 1.53,
+      "step": 1934
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2247328676845415e-06,
+      "loss": 1.5621,
+      "step": 1935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1884824889057066e-06,
+      "loss": 1.3635,
+      "step": 1936
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1527750815250481e-06,
+      "loss": 1.2758,
+      "step": 1937
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.117610743062536e-06,
+      "loss": 1.518,
+      "step": 1938
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0829895695550374e-06,
+      "loss": 1.4032,
+      "step": 1939
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.048911655555962e-06,
+      "loss": 1.7788,
+      "step": 1940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0153770941350172e-06,
+      "loss": 1.4456,
+      "step": 1941
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.823859768780086e-07,
+      "loss": 1.4624,
+      "step": 1942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.499383938864847e-07,
+      "loss": 1.3749,
+      "step": 1943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.180344337776259e-07,
+      "loss": 1.4968,
+      "step": 1944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.866741836838888e-07,
+      "loss": 1.6657,
+      "step": 1945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.558577292528069e-07,
+      "loss": 1.6438,
+      "step": 1946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.255851546467908e-07,
+      "loss": 1.3037,
+      "step": 1947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.958565425428388e-07,
+      "loss": 1.6018,
+      "step": 1948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.666719741323824e-07,
+      "loss": 1.392,
+      "step": 1949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.380315291209528e-07,
+      "loss": 1.5374,
+      "step": 1950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.099352857281138e-07,
+      "loss": 1.3997,
+      "step": 1951
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.823833206871077e-07,
+      "loss": 1.4063,
+      "step": 1952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.55375709244721e-07,
+      "loss": 1.5195,
+      "step": 1953
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.289125251610851e-07,
+      "loss": 1.5045,
+      "step": 1954
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.029938407094316e-07,
+      "loss": 1.532,
+      "step": 1955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.776197266759376e-07,
+      "loss": 1.6174,
+      "step": 1956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.527902523594586e-07,
+      "loss": 1.4325,
+      "step": 1957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.285054855714622e-07,
+      "loss": 1.4119,
+      "step": 1958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.047654926357392e-07,
+      "loss": 1.4863,
+      "step": 1959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.815703383882264e-07,
+      "loss": 1.4364,
+      "step": 1960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.589200861769394e-07,
+      "loss": 1.5309,
+      "step": 1961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.368147978616399e-07,
+      "loss": 1.5563,
+      "step": 1962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.152545338137692e-07,
+      "loss": 1.5772,
+      "step": 1963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9423935291629244e-07,
+      "loss": 1.2974,
+      "step": 1964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.737693125634545e-07,
+      "loss": 1.5887,
+      "step": 1965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.538444686607356e-07,
+      "loss": 1.3868,
+      "step": 1966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3446487562456276e-07,
+      "loss": 1.4866,
+      "step": 1967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1563058638230947e-07,
+      "loss": 1.4536,
+      "step": 1968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.973416523720296e-07,
+      "loss": 1.398,
+      "step": 1969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7959812354234615e-07,
+      "loss": 1.4143,
+      "step": 1970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6240004835242916e-07,
+      "loss": 1.4668,
+      "step": 1971
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.457474737716625e-07,
+      "loss": 1.5373,
+      "step": 1972
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2964044527966633e-07,
+      "loss": 1.3918,
+      "step": 1973
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1407900686616355e-07,
+      "loss": 1.5909,
+      "step": 1974
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9906320103071363e-07,
+      "loss": 1.5743,
+      "step": 1975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8459306878284566e-07,
+      "loss": 1.5593,
+      "step": 1976
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7066864964170314e-07,
+      "loss": 1.1698,
+      "step": 1977
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5728998163608843e-07,
+      "loss": 1.374,
+      "step": 1978
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4445710130428502e-07,
+      "loss": 1.4814,
+      "step": 1979
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.32170043693991e-07,
+      "loss": 1.4802,
+      "step": 1980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2042884236223018e-07,
+      "loss": 1.6539,
+      "step": 1981
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0923352937526332e-07,
+      "loss": 1.3322,
+      "step": 1982
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.858413530841048e-08,
+      "loss": 1.545,
+      "step": 1983
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.848068924611763e-08,
+      "loss": 1.5501,
+      "step": 1984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.892321878173459e-08,
+      "loss": 1.4385,
+      "step": 1985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.991175001755945e-08,
+      "loss": 1.3686,
+      "step": 1986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.144630756470537e-08,
+      "loss": 1.3645,
+      "step": 1987
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3526914542989524e-08,
+      "loss": 1.2885,
+      "step": 1988
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.615359258097751e-08,
+      "loss": 1.6891,
+      "step": 1989
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.932636181585014e-08,
+      "loss": 1.5689,
+      "step": 1990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.304524089335903e-08,
+      "loss": 1.3357,
+      "step": 1991
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7310246967782172e-08,
+      "loss": 1.5573,
+      "step": 1992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2121395701901747e-08,
+      "loss": 1.6638,
+      "step": 1993
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7478701266893105e-08,
+      "loss": 1.4344,
+      "step": 1994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3382176342346953e-08,
+      "loss": 1.3828,
+      "step": 1995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.831832116269368e-09,
+      "loss": 1.5397,
+      "step": 1996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.827678284926365e-09,
+      "loss": 1.5922,
+      "step": 1997
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.369723052910502e-09,
+      "loss": 1.5328,
+      "step": 1998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4579731331408896e-09,
+      "loss": 1.5408,
+      "step": 1999
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.092433746752164e-09,
+      "loss": 1.4483,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2001,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 3.8712881787070054e+17,
+  "trial_name": null,
+  "trial_params": null
+}