diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1000/trainer_state.json"
@@ -0,0 +1,6019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.49975012493753124,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 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
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2001,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.9365667152278323e+17,
+  "trial_name": null,
+  "trial_params": null
+}