diff --git "a/checkpoint-2000/trainer_state.json" "b/checkpoint-2000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2000/trainer_state.json"
@@ -0,0 +1,12019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5797941730685606,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.5832,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.516,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.6493,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.7471,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.7788,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 1.6985,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.6358,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.7444,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 1.8029,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.633,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.6036,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 1.7389,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.5451,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.5136,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.4915,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.5161,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.5376,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-05,
+      "loss": 1.5323,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6e-05,
+      "loss": 1.5795,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-05,
+      "loss": 1.5304,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4e-05,
+      "loss": 1.736,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.4172,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.4842,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6e-05,
+      "loss": 1.5846,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001,
+      "loss": 1.419,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 1.5319,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 1.5199,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 1.5843,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000116,
+      "loss": 1.3958,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012,
+      "loss": 1.3748,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000124,
+      "loss": 1.5378,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.5137,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000132,
+      "loss": 1.5762,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 1.4164,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014,
+      "loss": 1.5456,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000144,
+      "loss": 1.4421,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000148,
+      "loss": 1.341,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000152,
+      "loss": 1.4324,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 1.5214,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016,
+      "loss": 1.2637,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000164,
+      "loss": 1.5638,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000168,
+      "loss": 1.3309,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000172,
+      "loss": 1.3436,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.5341,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018,
+      "loss": 1.3339,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 1.3667,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000188,
+      "loss": 1.3897,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000192,
+      "loss": 1.3349,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000196,
+      "loss": 1.4463,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.3309,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 1.4589,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00020800000000000001,
+      "loss": 1.421,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00021200000000000003,
+      "loss": 1.399,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00021600000000000002,
+      "loss": 1.3131,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00022000000000000003,
+      "loss": 1.2907,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00022400000000000002,
+      "loss": 1.443,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 1.5236,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000232,
+      "loss": 1.3018,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000236,
+      "loss": 1.4164,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00024,
+      "loss": 1.3918,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000244,
+      "loss": 1.5658,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000248,
+      "loss": 1.3968,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000252,
+      "loss": 1.4089,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00025600000000000004,
+      "loss": 1.3224,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 1.5052,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000264,
+      "loss": 1.4618,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000268,
+      "loss": 1.405,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 1.3175,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000276,
+      "loss": 1.2401,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00028,
+      "loss": 1.3094,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000284,
+      "loss": 1.2717,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000288,
+      "loss": 1.2567,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000292,
+      "loss": 1.4094,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000296,
+      "loss": 1.3668,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 1.2809,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000304,
+      "loss": 1.6195,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000308,
+      "loss": 1.5164,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00031200000000000005,
+      "loss": 1.3275,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00031600000000000004,
+      "loss": 1.3314,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00032,
+      "loss": 1.4825,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000324,
+      "loss": 1.4019,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000328,
+      "loss": 1.3452,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000332,
+      "loss": 1.279,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000336,
+      "loss": 1.4439,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00034,
+      "loss": 1.3839,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000344,
+      "loss": 1.2913,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000348,
+      "loss": 1.3672,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00035200000000000005,
+      "loss": 1.4193,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00035600000000000003,
+      "loss": 1.4455,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00036,
+      "loss": 1.2722,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000364,
+      "loss": 1.5172,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00036800000000000005,
+      "loss": 1.4723,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00037200000000000004,
+      "loss": 1.2708,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000376,
+      "loss": 1.4163,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00038,
+      "loss": 1.4054,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000384,
+      "loss": 1.3794,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000388,
+      "loss": 1.3473,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000392,
+      "loss": 1.5271,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 1.4365,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 1.5053,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999999120026855,
+      "loss": 1.3237,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039999964801081954,
+      "loss": 1.5015,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999992080246343,
+      "loss": 1.2449,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999985920445171,
+      "loss": 1.4435,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999978000710098,
+      "loss": 1.4326,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999968321048095,
+      "loss": 1.4118,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999956881467678,
+      "loss": 1.4152,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039999436819789153,
+      "loss": 1.3995,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999928722593421,
+      "loss": 1.3273,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399991200332436,
+      "loss": 1.2412,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999893524186444,
+      "loss": 1.404,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999873285195934,
+      "loss": 1.4241,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399985128637064,
+      "loss": 1.3263,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399982752772992,
+      "loss": 1.4066,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039998020092946824,
+      "loss": 1.4737,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999774731087381,
+      "loss": 1.3736,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999745693132021,
+      "loss": 1.3962,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039997148954541543,
+      "loss": 1.4497,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039996823380808824,
+      "loss": 1.3684,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999648021040855,
+      "loss": 1.396,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399961194436427,
+      "loss": 1.4098,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039995741080828734,
+      "loss": 1.4442,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999534512229961,
+      "loss": 1.3268,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039994931568403764,
+      "loss": 1.4533,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399945004195051,
+      "loss": 1.2902,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999405167598303,
+      "loss": 1.3231,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999358533823242,
+      "loss": 1.4404,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999310140666365,
+      "loss": 1.5074,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039992599881702554,
+      "loss": 1.5619,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039992080763790475,
+      "loss": 1.4426,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039991544053384213,
+      "loss": 1.2351,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039990989750956064,
+      "loss": 1.3252,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039990417856993794,
+      "loss": 1.4116,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039989828372000657,
+      "loss": 1.3452,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998922129649538,
+      "loss": 1.3856,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998859663101218,
+      "loss": 1.3277,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998795437610074,
+      "loss": 1.4707,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039987294532326233,
+      "loss": 1.4154,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399866171002693,
+      "loss": 1.4648,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039985922080526065,
+      "loss": 1.3241,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998520947370812,
+      "loss": 1.3787,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998447928044255,
+      "loss": 1.1477,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399837315013719,
+      "loss": 1.4423,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039982966137154193,
+      "loss": 1.4003,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998218318846293,
+      "loss": 1.3892,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039981382655987095,
+      "loss": 1.4127,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039980564540431125,
+      "loss": 1.4172,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039979728842514935,
+      "loss": 1.3198,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039978875562973927,
+      "loss": 1.3332,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039978004702558966,
+      "loss": 1.4563,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997711626203637,
+      "loss": 1.4628,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039976210242187964,
+      "loss": 1.358,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039975286643811,
+      "loss": 1.2227,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997434546771824,
+      "loss": 1.4093,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997338671473787,
+      "loss": 1.3938,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003997241038571359,
+      "loss": 1.2654,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039971416481504536,
+      "loss": 1.3301,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003997040500298531,
+      "loss": 1.4287,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039969375951045985,
+      "loss": 1.5021,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039968329326592117,
+      "loss": 1.3404,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003996726513054469,
+      "loss": 1.3392,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003996618336384017,
+      "loss": 1.4043,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039965084027430487,
+      "loss": 1.2441,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039963967122283033,
+      "loss": 1.442,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039962832649380635,
+      "loss": 1.3297,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039961680609721626,
+      "loss": 1.3974,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039960511004319746,
+      "loss": 1.4026,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995932383420423,
+      "loss": 1.4038,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995811910041975,
+      "loss": 1.2948,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995689680402644,
+      "loss": 1.504,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039955656946099894,
+      "loss": 1.2047,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995439952773114,
+      "loss": 1.328,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995312455002669,
+      "loss": 1.3518,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995183201410848,
+      "loss": 1.4515,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000399505219211139,
+      "loss": 1.3857,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003994919427219581,
+      "loss": 1.4068,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000399478490685225,
+      "loss": 1.2225,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003994648631127772,
+      "loss": 1.3794,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003994510600166065,
+      "loss": 1.4642,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039943708140885925,
+      "loss": 1.2692,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039942292730183625,
+      "loss": 1.3233,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003994085977079928,
+      "loss": 1.2888,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039939409263993854,
+      "loss": 1.4167,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003993794121104375,
+      "loss": 1.3193,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003993645561324082,
+      "loss": 1.4016,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003993495247189235,
+      "loss": 1.3544,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003993343178832106,
+      "loss": 1.4277,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039931893563865115,
+      "loss": 1.4128,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039930337799878103,
+      "loss": 1.3224,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003992876449772907,
+      "loss": 1.4007,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003992717365880246,
+      "loss": 1.2749,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039925565284498185,
+      "loss": 1.3118,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039923939376231563,
+      "loss": 1.2867,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039922295935433346,
+      "loss": 1.3441,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003992063496354973,
+      "loss": 1.3888,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003991895646204232,
+      "loss": 1.405,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039917260432388146,
+      "loss": 1.4234,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003991554687607968,
+      "loss": 1.4004,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000399138157946248,
+      "loss": 1.4156,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000399120671895468,
+      "loss": 1.4495,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003991030106238442,
+      "loss": 1.4346,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000399085174146918,
+      "loss": 1.2202,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039906716248038507,
+      "loss": 1.473,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003990489756400951,
+      "loss": 1.3089,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003990306136420521,
+      "loss": 1.3049,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003990120765024141,
+      "loss": 1.3764,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003989933642374933,
+      "loss": 1.272,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003989744768637559,
+      "loss": 1.3248,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003989554143978224,
+      "loss": 1.3713,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003989361768564673,
+      "loss": 1.5846,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000398916764256619,
+      "loss": 1.3624,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039889717661536003,
+      "loss": 1.3555,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039887741394992713,
+      "loss": 1.2427,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003988574762777108,
+      "loss": 1.4018,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003988373636162558,
+      "loss": 1.3019,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003988170759832605,
+      "loss": 1.3403,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039879661339657776,
+      "loss": 1.3435,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003987759758742138,
+      "loss": 1.4053,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039875516343432937,
+      "loss": 1.4096,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039873417609523863,
+      "loss": 1.4866,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039871301387541,
+      "loss": 1.4622,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039869167679346573,
+      "loss": 1.3574,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003986701648681817,
+      "loss": 1.4224,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000398648478118488,
+      "loss": 1.3927,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039862661656346816,
+      "loss": 1.3708,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039860458022236,
+      "loss": 1.2435,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003985823691145548,
+      "loss": 1.2099,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003985599832595977,
+      "loss": 1.3541,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039853742267718776,
+      "loss": 1.3682,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039851468738717765,
+      "loss": 1.3608,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039849177740957373,
+      "loss": 1.4069,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039846869276453624,
+      "loss": 1.7075,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000398445433472379,
+      "loss": 1.257,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003984219995535697,
+      "loss": 1.3831,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039839839102872936,
+      "loss": 1.3522,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039837460791863297,
+      "loss": 1.4216,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000398350650244209,
+      "loss": 1.3535,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003983265180265396,
+      "loss": 1.3365,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003983022112868604,
+      "loss": 1.4763,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003982777300465608,
+      "loss": 1.3085,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003982530743271834,
+      "loss": 1.3435,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003982282441504248,
+      "loss": 1.1607,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003982032395381348,
+      "loss": 1.3848,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039817806051231685,
+      "loss": 1.5452,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003981527070951277,
+      "loss": 1.4223,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039812717930887777,
+      "loss": 1.5239,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003981014771760307,
+      "loss": 1.4098,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003980756007192039,
+      "loss": 1.2477,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039804954996116774,
+      "loss": 1.2293,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003980233249248463,
+      "loss": 1.2934,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039799692563331686,
+      "loss": 1.2993,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003979703521098102,
+      "loss": 1.4085,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039794360437771015,
+      "loss": 1.2628,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039791668246055406,
+      "loss": 1.4072,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003978895863820325,
+      "loss": 1.3316,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039786231616598935,
+      "loss": 1.3214,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039783487183642154,
+      "loss": 1.3406,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039780725341747946,
+      "loss": 1.5058,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003977794609334665,
+      "loss": 1.2867,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039775149440883934,
+      "loss": 1.3296,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003977233538682077,
+      "loss": 1.3374,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003976950393363346,
+      "loss": 1.3311,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039766655083813603,
+      "loss": 1.3599,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003976378883986811,
+      "loss": 1.2804,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000397609052043192,
+      "loss": 1.2712,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039758004179704393,
+      "loss": 1.3557,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003975508576857651,
+      "loss": 1.2742,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003975214997350368,
+      "loss": 1.5171,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039749196797069316,
+      "loss": 1.4556,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003974622624187215,
+      "loss": 1.2967,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039743238310526173,
+      "loss": 1.3242,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003974023300566069,
+      "loss": 1.2761,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003973721032992029,
+      "loss": 1.3672,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003973417028596485,
+      "loss": 1.3849,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039731112876469527,
+      "loss": 1.3892,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003972803810412475,
+      "loss": 1.3909,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003972494597163625,
+      "loss": 1.407,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039721836481725,
+      "loss": 1.3331,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039718709637127285,
+      "loss": 1.4107,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039715565440594637,
+      "loss": 1.3444,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039712403894893875,
+      "loss": 1.3164,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003970922500280706,
+      "loss": 1.4395,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039706028767131545,
+      "loss": 1.4244,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039702815190679916,
+      "loss": 1.2813,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039699584276280044,
+      "loss": 1.427,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039696336026775043,
+      "loss": 1.1327,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000396930704450233,
+      "loss": 1.3982,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039689787533898414,
+      "loss": 1.3648,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003968648729628927,
+      "loss": 1.3662,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039683169735099997,
+      "loss": 1.3564,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039679834853249954,
+      "loss": 1.3171,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003967648265367374,
+      "loss": 1.4295,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039673113139321217,
+      "loss": 1.2965,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039669726313157447,
+      "loss": 1.6415,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039666322178162763,
+      "loss": 1.2551,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039662900737332695,
+      "loss": 1.3234,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003965946199367804,
+      "loss": 1.3994,
+      "step": 297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003965600595022478,
+      "loss": 1.206,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003965253261001416,
+      "loss": 1.2827,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003964904197610261,
+      "loss": 1.3162,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000396455340515618,
+      "loss": 1.5978,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039642008839478615,
+      "loss": 1.3232,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003963846634295513,
+      "loss": 1.2778,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039634906565108667,
+      "loss": 1.3913,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039631329509071724,
+      "loss": 1.2894,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039627735177992024,
+      "loss": 1.3261,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003962412357503246,
+      "loss": 1.3329,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039620494703371175,
+      "loss": 1.3168,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039616848566201453,
+      "loss": 1.3801,
+      "step": 309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039613185166731814,
+      "loss": 1.3349,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039609504508185937,
+      "loss": 1.2949,
+      "step": 311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039605806593802714,
+      "loss": 1.1931,
+      "step": 312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039602091426836205,
+      "loss": 1.5426,
+      "step": 313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003959835901055566,
+      "loss": 1.4269,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039594609348245506,
+      "loss": 1.3779,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039590842443205335,
+      "loss": 1.3227,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003958705829874994,
+      "loss": 1.331,
+      "step": 317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003958325691820925,
+      "loss": 1.2632,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039579438304928385,
+      "loss": 1.3223,
+      "step": 319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003957560246226763,
+      "loss": 1.396,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039571749393602405,
+      "loss": 1.3239,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003956787910232332,
+      "loss": 1.4516,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039563991591836127,
+      "loss": 1.3951,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003956008686556172,
+      "loss": 1.3145,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003955616492693617,
+      "loss": 1.298,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003955222577941067,
+      "loss": 1.3131,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039548269426451557,
+      "loss": 1.2039,
+      "step": 327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003954429587154032,
+      "loss": 1.3435,
+      "step": 328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003954030511817359,
+      "loss": 1.4773,
+      "step": 329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003953629716986311,
+      "loss": 1.4621,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003953227203013577,
+      "loss": 1.4583,
+      "step": 331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003952822970253359,
+      "loss": 1.4588,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003952417019061371,
+      "loss": 1.3788,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003952009349794838,
+      "loss": 1.3653,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003951599962812499,
+      "loss": 1.2425,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039511888584746035,
+      "loss": 1.2731,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039507760371429126,
+      "loss": 1.3504,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003950361499180697,
+      "loss": 1.3654,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039499452449527395,
+      "loss": 1.3129,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039495272748253326,
+      "loss": 1.4244,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039491075891662794,
+      "loss": 1.41,
+      "step": 341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039486861883448907,
+      "loss": 1.2094,
+      "step": 342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039482630727319883,
+      "loss": 1.5471,
+      "step": 343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003947838242699904,
+      "loss": 1.2915,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003947411698622475,
+      "loss": 1.3057,
+      "step": 345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000394698344087505,
+      "loss": 1.374,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039465534698344826,
+      "loss": 1.3786,
+      "step": 347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003946121785879138,
+      "loss": 1.4415,
+      "step": 348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039456883893888845,
+      "loss": 1.2374,
+      "step": 349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039452532807451006,
+      "loss": 1.3662,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003944816460330669,
+      "loss": 1.2629,
+      "step": 351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039443779285299824,
+      "loss": 1.2885,
+      "step": 352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003943937685728934,
+      "loss": 1.3265,
+      "step": 353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039434957323149277,
+      "loss": 1.2861,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039430520686768695,
+      "loss": 1.27,
+      "step": 355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003942606695205172,
+      "loss": 1.1879,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003942159612291752,
+      "loss": 1.4018,
+      "step": 357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000394171082033003,
+      "loss": 1.3663,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003941260319714932,
+      "loss": 1.2719,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039408081108428843,
+      "loss": 1.3557,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003940354194111821,
+      "loss": 1.251,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003939898569921175,
+      "loss": 1.4242,
+      "step": 362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003939441238671884,
+      "loss": 1.464,
+      "step": 363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039389822007663866,
+      "loss": 1.279,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003938521456608625,
+      "loss": 1.4705,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000393805900660404,
+      "loss": 1.2939,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003937594851159577,
+      "loss": 1.3762,
+      "step": 367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039371289906836793,
+      "loss": 1.504,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039366614255862916,
+      "loss": 1.1273,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039361921562788594,
+      "loss": 1.2929,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003935721183174326,
+      "loss": 1.461,
+      "step": 371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039352485066871355,
+      "loss": 1.2933,
+      "step": 372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003934774127233231,
+      "loss": 1.2856,
+      "step": 373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003934298045230053,
+      "loss": 1.3504,
+      "step": 374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003933820261096541,
+      "loss": 1.3782,
+      "step": 375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039333407752531333,
+      "loss": 1.2738,
+      "step": 376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039328595881217625,
+      "loss": 1.4028,
+      "step": 377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003932376700125862,
+      "loss": 1.3697,
+      "step": 378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000393189211169036,
+      "loss": 1.471,
+      "step": 379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003931405823241681,
+      "loss": 1.4331,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039309178352077455,
+      "loss": 1.3678,
+      "step": 381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003930428148017971,
+      "loss": 1.2892,
+      "step": 382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039299367621032673,
+      "loss": 1.381,
+      "step": 383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003929443677896042,
+      "loss": 1.255,
+      "step": 384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039289488958301963,
+      "loss": 1.5373,
+      "step": 385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003928452416341124,
+      "loss": 1.4048,
+      "step": 386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003927954239865715,
+      "loss": 1.3485,
+      "step": 387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039274543668423495,
+      "loss": 1.3841,
+      "step": 388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039269527977109044,
+      "loss": 1.2156,
+      "step": 389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003926449532912746,
+      "loss": 1.4647,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039259445728907327,
+      "loss": 1.3239,
+      "step": 391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003925437918089218,
+      "loss": 1.2882,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039249295689540426,
+      "loss": 1.3821,
+      "step": 393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039244195259325416,
+      "loss": 1.1908,
+      "step": 394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039239077894735383,
+      "loss": 1.1618,
+      "step": 395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003923394360027347,
+      "loss": 1.3317,
+      "step": 396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003922879238045773,
+      "loss": 1.5053,
+      "step": 397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039223624239821073,
+      "loss": 1.3213,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003921843918291135,
+      "loss": 1.1913,
+      "step": 399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003921323721429126,
+      "loss": 1.3521,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039208018338538386,
+      "loss": 1.4003,
+      "step": 401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003920278256024522,
+      "loss": 1.2853,
+      "step": 402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039197529884019096,
+      "loss": 1.6179,
+      "step": 403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039192260314482223,
+      "loss": 1.3391,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039186973856271683,
+      "loss": 1.3379,
+      "step": 405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003918167051403943,
+      "loss": 1.2994,
+      "step": 406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039176350292452244,
+      "loss": 1.3591,
+      "step": 407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039171013196191793,
+      "loss": 1.259,
+      "step": 408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039165659229954566,
+      "loss": 1.402,
+      "step": 409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039160288398451906,
+      "loss": 1.2868,
+      "step": 410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003915490070641002,
+      "loss": 1.4592,
+      "step": 411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039149496158569924,
+      "loss": 1.3848,
+      "step": 412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003914407475968746,
+      "loss": 1.292,
+      "step": 413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003913863651453333,
+      "loss": 1.2964,
+      "step": 414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039133181427893044,
+      "loss": 1.2426,
+      "step": 415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003912770950456693,
+      "loss": 1.3032,
+      "step": 416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039122220749370125,
+      "loss": 1.427,
+      "step": 417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039116715167132593,
+      "loss": 1.1978,
+      "step": 418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039111192762699093,
+      "loss": 1.3663,
+      "step": 419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039105653540929197,
+      "loss": 1.2532,
+      "step": 420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039100097506697276,
+      "loss": 1.2392,
+      "step": 421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039094524664892483,
+      "loss": 1.2665,
+      "step": 422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039088935020418776,
+      "loss": 1.2964,
+      "step": 423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003908332857819488,
+      "loss": 1.3515,
+      "step": 424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003907770534315433,
+      "loss": 1.4667,
+      "step": 425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039072065320245415,
+      "loss": 1.2508,
+      "step": 426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000390664085144312,
+      "loss": 1.1843,
+      "step": 427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003906073493068952,
+      "loss": 1.2708,
+      "step": 428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003905504457401299,
+      "loss": 1.2819,
+      "step": 429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039049337449408955,
+      "loss": 1.3026,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003904361356189954,
+      "loss": 1.3789,
+      "step": 431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039037872916521614,
+      "loss": 1.324,
+      "step": 432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003903211551832678,
+      "loss": 1.4777,
+      "step": 433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003902634137238141,
+      "loss": 1.3832,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039020550483766583,
+      "loss": 1.2534,
+      "step": 435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003901474285757813,
+      "loss": 1.3532,
+      "step": 436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003900891849892661,
+      "loss": 1.1942,
+      "step": 437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000390030774129373,
+      "loss": 1.3259,
+      "step": 438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038997219604750195,
+      "loss": 1.3383,
+      "step": 439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038991345079520014,
+      "loss": 1.3872,
+      "step": 440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038985453842416183,
+      "loss": 1.2964,
+      "step": 441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003897954589862282,
+      "loss": 1.4885,
+      "step": 442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038973621253338776,
+      "loss": 1.1836,
+      "step": 443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003896767991177757,
+      "loss": 1.3709,
+      "step": 444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003896172187916741,
+      "loss": 1.3361,
+      "step": 445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003895574716075123,
+      "loss": 1.4199,
+      "step": 446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038949755761786604,
+      "loss": 1.3185,
+      "step": 447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003894374768754581,
+      "loss": 1.4701,
+      "step": 448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003893772294331578,
+      "loss": 1.2512,
+      "step": 449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038931681534398147,
+      "loss": 1.4217,
+      "step": 450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038925623466109176,
+      "loss": 1.393,
+      "step": 451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038919548743779803,
+      "loss": 1.4419,
+      "step": 452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038913457372755626,
+      "loss": 1.2372,
+      "step": 453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003890734935839688,
+      "loss": 1.4241,
+      "step": 454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003890122470607847,
+      "loss": 1.3903,
+      "step": 455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038895083421189904,
+      "loss": 1.1638,
+      "step": 456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038888925509135365,
+      "loss": 1.3148,
+      "step": 457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038882750975333645,
+      "loss": 1.3197,
+      "step": 458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038876559825218164,
+      "loss": 1.3027,
+      "step": 459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003887035206423697,
+      "loss": 1.2378,
+      "step": 460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003886412769785273,
+      "loss": 1.2874,
+      "step": 461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003885788673154272,
+      "loss": 1.3247,
+      "step": 462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003885162917079881,
+      "loss": 1.4948,
+      "step": 463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038845355021127495,
+      "loss": 1.3641,
+      "step": 464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003883906428804986,
+      "loss": 1.2692,
+      "step": 465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038832756977101574,
+      "loss": 1.2695,
+      "step": 466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038826433093832913,
+      "loss": 1.3523,
+      "step": 467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003882009264380871,
+      "loss": 1.4101,
+      "step": 468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003881373563260841,
+      "loss": 1.1302,
+      "step": 469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003880736206582599,
+      "loss": 1.4,
+      "step": 470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003880097194907003,
+      "loss": 1.2677,
+      "step": 471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038794565287963653,
+      "loss": 1.2799,
+      "step": 472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038788142088144566,
+      "loss": 1.4191,
+      "step": 473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038781702355264995,
+      "loss": 1.1316,
+      "step": 474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038775246094991734,
+      "loss": 1.3546,
+      "step": 475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003876877331300613,
+      "loss": 1.2596,
+      "step": 476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003876228401500404,
+      "loss": 1.5291,
+      "step": 477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003875577820669589,
+      "loss": 1.4081,
+      "step": 478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003874925589380661,
+      "loss": 1.2045,
+      "step": 479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003874271708207565,
+      "loss": 1.3115,
+      "step": 480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038736161777257004,
+      "loss": 1.3006,
+      "step": 481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038729589985119156,
+      "loss": 1.2511,
+      "step": 482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003872300171144511,
+      "loss": 1.3153,
+      "step": 483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038716396962032364,
+      "loss": 1.5334,
+      "step": 484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038709775742692927,
+      "loss": 1.3983,
+      "step": 485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003870313805925329,
+      "loss": 1.2143,
+      "step": 486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003869648391755444,
+      "loss": 1.3641,
+      "step": 487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038689813323451836,
+      "loss": 1.4607,
+      "step": 488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003868312628281543,
+      "loss": 1.2097,
+      "step": 489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003867642280152963,
+      "loss": 1.2142,
+      "step": 490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003866970288549333,
+      "loss": 1.1171,
+      "step": 491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003866296654061986,
+      "loss": 1.3582,
+      "step": 492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038656213772837045,
+      "loss": 1.2705,
+      "step": 493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038649444588087115,
+      "loss": 1.0947,
+      "step": 494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038642658992326794,
+      "loss": 1.3349,
+      "step": 495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038635856991527204,
+      "loss": 1.3444,
+      "step": 496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003862903859167393,
+      "loss": 1.2091,
+      "step": 497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038622203798766984,
+      "loss": 1.3876,
+      "step": 498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000386153526188208,
+      "loss": 1.3659,
+      "step": 499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038608485057864227,
+      "loss": 1.3617,
+      "step": 500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003860160112194054,
+      "loss": 1.3491,
+      "step": 501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038594700817107415,
+      "loss": 1.2504,
+      "step": 502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003858778414943694,
+      "loss": 1.297,
+      "step": 503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038580851125015583,
+      "loss": 1.4889,
+      "step": 504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003857390174994424,
+      "loss": 1.3215,
+      "step": 505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038566936030338153,
+      "loss": 1.336,
+      "step": 506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038559953972326985,
+      "loss": 1.4424,
+      "step": 507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003855295558205475,
+      "loss": 1.3839,
+      "step": 508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038545940865679845,
+      "loss": 1.4157,
+      "step": 509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038538909829375034,
+      "loss": 1.3802,
+      "step": 510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003853186247932744,
+      "loss": 1.4247,
+      "step": 511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003852479882173854,
+      "loss": 1.2501,
+      "step": 512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038517718862824166,
+      "loss": 1.3293,
+      "step": 513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038510622608814486,
+      "loss": 1.3841,
+      "step": 514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038503510065954026,
+      "loss": 1.3453,
+      "step": 515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038496381240501614,
+      "loss": 1.2539,
+      "step": 516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003848923613873044,
+      "loss": 1.2811,
+      "step": 517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038482074766927997,
+      "loss": 1.3435,
+      "step": 518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003847489713139609,
+      "loss": 1.2876,
+      "step": 519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038467703238450864,
+      "loss": 1.185,
+      "step": 520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038460493094422734,
+      "loss": 1.196,
+      "step": 521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003845326670565645,
+      "loss": 1.3964,
+      "step": 522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003844602407851103,
+      "loss": 1.4147,
+      "step": 523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003843876521935979,
+      "loss": 1.352,
+      "step": 524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038431490134590337,
+      "loss": 1.2607,
+      "step": 525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003842419883060454,
+      "loss": 1.2796,
+      "step": 526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003841689131381856,
+      "loss": 1.3396,
+      "step": 527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003840956759066282,
+      "loss": 1.4369,
+      "step": 528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003840222766758198,
+      "loss": 1.5503,
+      "step": 529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038394871551035003,
+      "loss": 1.3106,
+      "step": 530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003838749924749505,
+      "loss": 1.3839,
+      "step": 531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038380110763449563,
+      "loss": 1.3257,
+      "step": 532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038372706105400207,
+      "loss": 1.4503,
+      "step": 533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038365285279862876,
+      "loss": 1.2723,
+      "step": 534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000383578482933677,
+      "loss": 1.4161,
+      "step": 535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003835039515245904,
+      "loss": 1.1735,
+      "step": 536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003834292586369545,
+      "loss": 1.2848,
+      "step": 537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038335440433649693,
+      "loss": 1.3721,
+      "step": 538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038327938868908763,
+      "loss": 1.3287,
+      "step": 539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038320421176073823,
+      "loss": 1.5261,
+      "step": 540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003831288736176025,
+      "loss": 1.3888,
+      "step": 541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000383053374325976,
+      "loss": 1.3134,
+      "step": 542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038297771395229594,
+      "loss": 1.3813,
+      "step": 543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038290189256314156,
+      "loss": 1.2592,
+      "step": 544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038282591022523356,
+      "loss": 1.2625,
+      "step": 545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003827497670054344,
+      "loss": 1.2192,
+      "step": 546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003826734629707481,
+      "loss": 1.2154,
+      "step": 547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003825969981883201,
+      "loss": 1.3196,
+      "step": 548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038252037272543734,
+      "loss": 1.3649,
+      "step": 549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003824435866495282,
+      "loss": 1.2522,
+      "step": 550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038236664002816244,
+      "loss": 1.2861,
+      "step": 551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038228953292905094,
+      "loss": 1.3513,
+      "step": 552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003822122654200458,
+      "loss": 1.2638,
+      "step": 553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003821348375691405,
+      "loss": 1.2824,
+      "step": 554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003820572494444694,
+      "loss": 1.4368,
+      "step": 555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038197950111430794,
+      "loss": 1.3823,
+      "step": 556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038190159264707267,
+      "loss": 1.4946,
+      "step": 557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038182352411132085,
+      "loss": 1.3771,
+      "step": 558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003817452955757507,
+      "loss": 1.367,
+      "step": 559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038166690710920124,
+      "loss": 1.2109,
+      "step": 560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038158835878065225,
+      "loss": 1.1682,
+      "step": 561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003815096506592241,
+      "loss": 1.2641,
+      "step": 562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003814307828141779,
+      "loss": 1.4155,
+      "step": 563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038135175531491515,
+      "loss": 1.4218,
+      "step": 564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038127256823097794,
+      "loss": 1.198,
+      "step": 565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038119322163204886,
+      "loss": 1.3554,
+      "step": 566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038111371558795064,
+      "loss": 1.1691,
+      "step": 567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003810340501686466,
+      "loss": 1.326,
+      "step": 568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003809542254442401,
+      "loss": 1.4071,
+      "step": 569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003808742414849748,
+      "loss": 1.3133,
+      "step": 570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003807940983612344,
+      "loss": 1.3361,
+      "step": 571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003807137961435427,
+      "loss": 1.2362,
+      "step": 572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038063333490256354,
+      "loss": 1.431,
+      "step": 573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038055271470910053,
+      "loss": 1.2436,
+      "step": 574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038047193563409744,
+      "loss": 1.2789,
+      "step": 575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038039099774863754,
+      "loss": 1.2921,
+      "step": 576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038030990112394413,
+      "loss": 1.2558,
+      "step": 577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003802286458313799,
+      "loss": 1.3656,
+      "step": 578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038014723194244746,
+      "loss": 1.2975,
+      "step": 579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003800656595287888,
+      "loss": 1.3769,
+      "step": 580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003799839286621855,
+      "loss": 1.4537,
+      "step": 581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003799020394145584,
+      "loss": 1.2094,
+      "step": 582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003798199918579679,
+      "loss": 1.3608,
+      "step": 583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003797377860646138,
+      "loss": 1.2369,
+      "step": 584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037965542210683474,
+      "loss": 1.3911,
+      "step": 585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000379572900057109,
+      "loss": 1.292,
+      "step": 586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003794902199880536,
+      "loss": 1.3406,
+      "step": 587
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003794073819724249,
+      "loss": 1.2936,
+      "step": 588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003793243860831181,
+      "loss": 1.529,
+      "step": 589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003792412323931673,
+      "loss": 1.2615,
+      "step": 590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037915792097574557,
+      "loss": 1.2682,
+      "step": 591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003790744519041647,
+      "loss": 1.4804,
+      "step": 592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037899082525187525,
+      "loss": 1.2158,
+      "step": 593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003789070410924664,
+      "loss": 1.3411,
+      "step": 594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037882309949966597,
+      "loss": 1.2307,
+      "step": 595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037873900054734034,
+      "loss": 1.2769,
+      "step": 596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037865474430949427,
+      "loss": 1.4223,
+      "step": 597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037857033086027104,
+      "loss": 1.2942,
+      "step": 598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003784857602739522,
+      "loss": 1.2689,
+      "step": 599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003784010326249576,
+      "loss": 1.3241,
+      "step": 600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037831614798784525,
+      "loss": 1.222,
+      "step": 601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003782311064373114,
+      "loss": 1.3674,
+      "step": 602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003781459080481903,
+      "loss": 1.4221,
+      "step": 603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003780605528954543,
+      "loss": 1.4172,
+      "step": 604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003779750410542136,
+      "loss": 1.3146,
+      "step": 605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003778893725997163,
+      "loss": 1.2114,
+      "step": 606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037780354760734837,
+      "loss": 1.2844,
+      "step": 607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003777175661526336,
+      "loss": 1.3779,
+      "step": 608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003776314283112331,
+      "loss": 1.0988,
+      "step": 609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003775451341589461,
+      "loss": 1.3803,
+      "step": 610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037745868377170907,
+      "loss": 1.1517,
+      "step": 611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000377372077225596,
+      "loss": 1.2965,
+      "step": 612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037728531459681826,
+      "loss": 1.4926,
+      "step": 613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037719839596172476,
+      "loss": 1.4066,
+      "step": 614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037711132139680155,
+      "loss": 1.3498,
+      "step": 615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037702409097867186,
+      "loss": 1.4738,
+      "step": 616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037693670478409607,
+      "loss": 1.3854,
+      "step": 617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037684916288997176,
+      "loss": 1.2015,
+      "step": 618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037676146537333344,
+      "loss": 1.3336,
+      "step": 619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037667361231135256,
+      "loss": 1.2285,
+      "step": 620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037658560378133745,
+      "loss": 1.3738,
+      "step": 621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003764974398607333,
+      "loss": 1.2951,
+      "step": 622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037640912062712186,
+      "loss": 1.3998,
+      "step": 623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037632064615822185,
+      "loss": 1.3421,
+      "step": 624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003762320165318883,
+      "loss": 1.2505,
+      "step": 625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000376143231826113,
+      "loss": 1.3863,
+      "step": 626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037605429211902397,
+      "loss": 1.2849,
+      "step": 627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037596519748888594,
+      "loss": 1.221,
+      "step": 628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003758759480140997,
+      "loss": 1.195,
+      "step": 629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003757865437732023,
+      "loss": 1.4152,
+      "step": 630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037569698484486724,
+      "loss": 1.3788,
+      "step": 631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037560727130790385,
+      "loss": 1.2949,
+      "step": 632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037551740324125773,
+      "loss": 1.3709,
+      "step": 633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037542738072401027,
+      "loss": 1.2895,
+      "step": 634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003753372038353789,
+      "loss": 1.3311,
+      "step": 635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037524687265471697,
+      "loss": 1.2309,
+      "step": 636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003751563872615133,
+      "loss": 1.3925,
+      "step": 637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037506574773539276,
+      "loss": 1.3566,
+      "step": 638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003749749541561156,
+      "loss": 1.3243,
+      "step": 639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037488400660357774,
+      "loss": 1.3834,
+      "step": 640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037479290515781067,
+      "loss": 1.2708,
+      "step": 641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037470164989898116,
+      "loss": 1.3351,
+      "step": 642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037461024090739136,
+      "loss": 1.3267,
+      "step": 643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037451867826347876,
+      "loss": 1.3807,
+      "step": 644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037442696204781605,
+      "loss": 1.1451,
+      "step": 645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000374335092341111,
+      "loss": 1.2059,
+      "step": 646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003742430692242065,
+      "loss": 1.1875,
+      "step": 647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003741508927780804,
+      "loss": 1.4249,
+      "step": 648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003740585630838455,
+      "loss": 1.4409,
+      "step": 649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003739660802227495,
+      "loss": 1.4722,
+      "step": 650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037387344427617476,
+      "loss": 1.2156,
+      "step": 651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037378065532563847,
+      "loss": 1.3374,
+      "step": 652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037368771345279244,
+      "loss": 1.24,
+      "step": 653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037359461873942297,
+      "loss": 1.3434,
+      "step": 654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003735013712674509,
+      "loss": 1.3179,
+      "step": 655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003734079711189316,
+      "loss": 1.3183,
+      "step": 656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003733144183760546,
+      "loss": 1.2632,
+      "step": 657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003732207131211438,
+      "loss": 1.4223,
+      "step": 658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003731268554366573,
+      "loss": 1.3397,
+      "step": 659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037303284540518745,
+      "loss": 1.2703,
+      "step": 660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037293868310946045,
+      "loss": 1.5002,
+      "step": 661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037284436863233657,
+      "loss": 1.3052,
+      "step": 662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003727499020568101,
+      "loss": 1.3705,
+      "step": 663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037265528346600913,
+      "loss": 1.4041,
+      "step": 664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003725605129431953,
+      "loss": 1.1651,
+      "step": 665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003724655905717643,
+      "loss": 1.3479,
+      "step": 666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003723705164352451,
+      "loss": 1.2762,
+      "step": 667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037227529061730055,
+      "loss": 1.2119,
+      "step": 668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037217991320172674,
+      "loss": 1.3735,
+      "step": 669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003720843842724532,
+      "loss": 1.288,
+      "step": 670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003719887039135429,
+      "loss": 1.3138,
+      "step": 671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000371892872209192,
+      "loss": 1.3225,
+      "step": 672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003717968892437297,
+      "loss": 1.2151,
+      "step": 673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037170075510161853,
+      "loss": 1.3943,
+      "step": 674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003716044698674539,
+      "loss": 1.3298,
+      "step": 675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003715080336259643,
+      "loss": 1.2433,
+      "step": 676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000371411446462011,
+      "loss": 1.2967,
+      "step": 677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003713147084605881,
+      "loss": 1.3173,
+      "step": 678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003712178197068224,
+      "loss": 1.3028,
+      "step": 679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037112078028597343,
+      "loss": 1.2044,
+      "step": 680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037102359028343335,
+      "loss": 1.1363,
+      "step": 681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003709262497847267,
+      "loss": 1.4618,
+      "step": 682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037082875887551047,
+      "loss": 1.3237,
+      "step": 683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003707311176415741,
+      "loss": 1.2146,
+      "step": 684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003706333261688392,
+      "loss": 1.2813,
+      "step": 685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003705353845433597,
+      "loss": 1.1922,
+      "step": 686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003704372928513215,
+      "loss": 1.2796,
+      "step": 687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003703390511790428,
+      "loss": 1.277,
+      "step": 688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003702406596129735,
+      "loss": 1.4421,
+      "step": 689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003701421182396956,
+      "loss": 1.4526,
+      "step": 690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003700434271459229,
+      "loss": 1.1334,
+      "step": 691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036994458641850076,
+      "loss": 1.3087,
+      "step": 692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003698455961444065,
+      "loss": 1.2478,
+      "step": 693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003697464564107489,
+      "loss": 1.1383,
+      "step": 694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036964716730476826,
+      "loss": 1.4408,
+      "step": 695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036954772891383626,
+      "loss": 1.3716,
+      "step": 696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003694481413254561,
+      "loss": 1.2962,
+      "step": 697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036934840462726206,
+      "loss": 1.3133,
+      "step": 698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003692485189070199,
+      "loss": 1.1496,
+      "step": 699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036914848425262625,
+      "loss": 1.3681,
+      "step": 700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000369048300752109,
+      "loss": 1.4004,
+      "step": 701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000368947968493627,
+      "loss": 1.2389,
+      "step": 702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036884748756546977,
+      "loss": 1.3265,
+      "step": 703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036874685805605794,
+      "loss": 1.2333,
+      "step": 704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036864608005394273,
+      "loss": 1.2833,
+      "step": 705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003685451536478061,
+      "loss": 1.2394,
+      "step": 706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003684440789264606,
+      "loss": 1.2999,
+      "step": 707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036834285597884923,
+      "loss": 1.289,
+      "step": 708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003682414848940455,
+      "loss": 1.2782,
+      "step": 709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036813996576125316,
+      "loss": 1.1605,
+      "step": 710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036803829866980645,
+      "loss": 1.285,
+      "step": 711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003679364837091696,
+      "loss": 1.3159,
+      "step": 712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036783452096893703,
+      "loss": 1.5144,
+      "step": 713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036773241053883324,
+      "loss": 1.3787,
+      "step": 714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036763015250871266,
+      "loss": 1.2825,
+      "step": 715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003675277469685597,
+      "loss": 1.3308,
+      "step": 716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003674251940084883,
+      "loss": 1.3313,
+      "step": 717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036732249371874243,
+      "loss": 1.2538,
+      "step": 718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003672196461896956,
+      "loss": 1.2696,
+      "step": 719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036711665151185083,
+      "loss": 1.2787,
+      "step": 720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003670135097758407,
+      "loss": 1.3181,
+      "step": 721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003669102210724271,
+      "loss": 1.406,
+      "step": 722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003668067854925014,
+      "loss": 1.3445,
+      "step": 723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003667032031270841,
+      "loss": 1.3145,
+      "step": 724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036659947406732484,
+      "loss": 1.2812,
+      "step": 725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036649559840450254,
+      "loss": 1.3694,
+      "step": 726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003663915762300249,
+      "loss": 1.3665,
+      "step": 727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036628740763542864,
+      "loss": 1.3879,
+      "step": 728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003661830927123793,
+      "loss": 1.261,
+      "step": 729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003660786315526713,
+      "loss": 1.2177,
+      "step": 730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036597402424822764,
+      "loss": 1.1199,
+      "step": 731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036586927089109987,
+      "loss": 1.2116,
+      "step": 732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003657643715734682,
+      "loss": 1.38,
+      "step": 733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036565932638764115,
+      "loss": 1.3854,
+      "step": 734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036555413542605576,
+      "loss": 1.2881,
+      "step": 735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003654487987812771,
+      "loss": 1.3507,
+      "step": 736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003653433165459987,
+      "loss": 1.3011,
+      "step": 737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003652376888130421,
+      "loss": 1.1903,
+      "step": 738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036513191567535687,
+      "loss": 1.5096,
+      "step": 739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003650259972260204,
+      "loss": 1.3719,
+      "step": 740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003649199335582382,
+      "loss": 1.3139,
+      "step": 741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036481372476534346,
+      "loss": 1.3557,
+      "step": 742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000364707370940797,
+      "loss": 1.2022,
+      "step": 743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003646008721781874,
+      "loss": 1.415,
+      "step": 744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036449422857123067,
+      "loss": 1.4802,
+      "step": 745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003643874402137703,
+      "loss": 1.4193,
+      "step": 746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003642805071997772,
+      "loss": 1.2783,
+      "step": 747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036417342962334956,
+      "loss": 1.4414,
+      "step": 748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036406620757871276,
+      "loss": 1.2048,
+      "step": 749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036395884116021933,
+      "loss": 1.4374,
+      "step": 750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003638513304623488,
+      "loss": 1.2123,
+      "step": 751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036374367557970777,
+      "loss": 1.2275,
+      "step": 752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003636358766070296,
+      "loss": 1.2117,
+      "step": 753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003635279336391745,
+      "loss": 1.1179,
+      "step": 754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036341984677112937,
+      "loss": 1.3382,
+      "step": 755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003633116160980077,
+      "loss": 1.3122,
+      "step": 756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003632032417150496,
+      "loss": 1.2746,
+      "step": 757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036309472371762175,
+      "loss": 1.1844,
+      "step": 758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003629860622012169,
+      "loss": 1.4231,
+      "step": 759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003628772572614544,
+      "loss": 1.4839,
+      "step": 760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003627683089940796,
+      "loss": 1.4891,
+      "step": 761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000362659217494964,
+      "loss": 1.3534,
+      "step": 762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036254998286010536,
+      "loss": 1.3463,
+      "step": 763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000362440605185627,
+      "loss": 1.3156,
+      "step": 764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003623310845677785,
+      "loss": 1.3955,
+      "step": 765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036222142110293505,
+      "loss": 1.3864,
+      "step": 766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036211161488759747,
+      "loss": 1.3406,
+      "step": 767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003620016660183923,
+      "loss": 1.3364,
+      "step": 768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036189157459207166,
+      "loss": 1.1346,
+      "step": 769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036178134070551305,
+      "loss": 1.3928,
+      "step": 770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003616709644557193,
+      "loss": 1.2873,
+      "step": 771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036156044593981847,
+      "loss": 1.289,
+      "step": 772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000361449785255064,
+      "loss": 1.3108,
+      "step": 773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036133898249883425,
+      "loss": 1.4273,
+      "step": 774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003612280377686327,
+      "loss": 1.1892,
+      "step": 775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036111695116208765,
+      "loss": 1.3329,
+      "step": 776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036100572277695256,
+      "loss": 1.4551,
+      "step": 777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036089435271110517,
+      "loss": 1.4236,
+      "step": 778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036078284106254825,
+      "loss": 1.283,
+      "step": 779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003606711879294091,
+      "loss": 1.268,
+      "step": 780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003605593934099393,
+      "loss": 1.1755,
+      "step": 781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003604474576025153,
+      "loss": 1.3168,
+      "step": 782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036033538060563743,
+      "loss": 1.3726,
+      "step": 783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036022316251793044,
+      "loss": 1.1281,
+      "step": 784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036011080343814324,
+      "loss": 1.3585,
+      "step": 785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003599983034651489,
+      "loss": 1.3135,
+      "step": 786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003598856626979442,
+      "loss": 1.37,
+      "step": 787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035977288123565016,
+      "loss": 1.1796,
+      "step": 788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003596599591775113,
+      "loss": 1.1075,
+      "step": 789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035954689662289613,
+      "loss": 1.335,
+      "step": 790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003594336936712966,
+      "loss": 1.2655,
+      "step": 791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003593203504223282,
+      "loss": 1.385,
+      "step": 792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003592068669757301,
+      "loss": 1.101,
+      "step": 793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035909324343136453,
+      "loss": 1.326,
+      "step": 794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003589794798892172,
+      "loss": 1.3805,
+      "step": 795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035886557644939705,
+      "loss": 1.3828,
+      "step": 796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000358751533212136,
+      "loss": 1.1955,
+      "step": 797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035863735027778895,
+      "loss": 1.29,
+      "step": 798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035852302774683393,
+      "loss": 1.316,
+      "step": 799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035840856571987167,
+      "loss": 1.3937,
+      "step": 800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035829396429762567,
+      "loss": 1.3658,
+      "step": 801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003581792235809421,
+      "loss": 1.3005,
+      "step": 802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003580643436707898,
+      "loss": 1.235,
+      "step": 803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003579493246682598,
+      "loss": 1.2511,
+      "step": 804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035783416667456593,
+      "loss": 1.2629,
+      "step": 805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000357718869791044,
+      "loss": 1.2455,
+      "step": 806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035760343411915226,
+      "loss": 1.3293,
+      "step": 807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000357487859760471,
+      "loss": 1.4308,
+      "step": 808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003573721468167024,
+      "loss": 1.2123,
+      "step": 809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035725629538967097,
+      "loss": 1.4557,
+      "step": 810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003571403055813227,
+      "loss": 1.2088,
+      "step": 811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003570241774937256,
+      "loss": 1.3433,
+      "step": 812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003569079112290692,
+      "loss": 1.3293,
+      "step": 813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035679150688966476,
+      "loss": 1.3839,
+      "step": 814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035667496457794483,
+      "loss": 1.4165,
+      "step": 815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035655828439646366,
+      "loss": 1.2507,
+      "step": 816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003564414664478966,
+      "loss": 1.3885,
+      "step": 817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003563245108350404,
+      "loss": 1.1639,
+      "step": 818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003562074176608127,
+      "loss": 1.3476,
+      "step": 819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035609018702825257,
+      "loss": 1.3051,
+      "step": 820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035597281904051957,
+      "loss": 1.2524,
+      "step": 821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003558553138008946,
+      "loss": 1.1534,
+      "step": 822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035573767141277885,
+      "loss": 1.2203,
+      "step": 823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035561989197969475,
+      "loss": 1.2969,
+      "step": 824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035550197560528486,
+      "loss": 1.4392,
+      "step": 825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003553839223933126,
+      "loss": 1.3154,
+      "step": 826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035526573244766134,
+      "loss": 1.2765,
+      "step": 827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035514740587233525,
+      "loss": 1.18,
+      "step": 828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003550289427714586,
+      "loss": 1.2128,
+      "step": 829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003549103432492755,
+      "loss": 1.2001,
+      "step": 830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035479160741015063,
+      "loss": 1.2746,
+      "step": 831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035467273535856814,
+      "loss": 1.2801,
+      "step": 832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003545537271991323,
+      "loss": 1.347,
+      "step": 833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003544345830365671,
+      "loss": 1.3524,
+      "step": 834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003543153029757162,
+      "loss": 1.23,
+      "step": 835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035419588712154286,
+      "loss": 1.3264,
+      "step": 836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003540763355791298,
+      "loss": 1.4038,
+      "step": 837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003539566484536792,
+      "loss": 1.3156,
+      "step": 838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003538368258505125,
+      "loss": 1.1655,
+      "step": 839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003537168678750704,
+      "loss": 1.2485,
+      "step": 840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035359677463291263,
+      "loss": 1.3553,
+      "step": 841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000353476546229718,
+      "loss": 1.3247,
+      "step": 842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003533561827712844,
+      "loss": 1.2529,
+      "step": 843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003532356843635284,
+      "loss": 1.3355,
+      "step": 844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035311505111248527,
+      "loss": 1.2757,
+      "step": 845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035299428312430913,
+      "loss": 1.4812,
+      "step": 846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003528733805052725,
+      "loss": 1.335,
+      "step": 847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003527523433617665,
+      "loss": 1.4521,
+      "step": 848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035263117180030055,
+      "loss": 1.2322,
+      "step": 849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003525098659275024,
+      "loss": 1.3555,
+      "step": 850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035238842585011784,
+      "loss": 1.3167,
+      "step": 851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000352266851675011,
+      "loss": 1.2408,
+      "step": 852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003521451435091638,
+      "loss": 1.2397,
+      "step": 853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003520233014596762,
+      "loss": 1.4012,
+      "step": 854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000351901325633766,
+      "loss": 1.2655,
+      "step": 855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003517792161387685,
+      "loss": 1.2994,
+      "step": 856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003516569730821369,
+      "loss": 1.3843,
+      "step": 857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003515345965714418,
+      "loss": 1.2291,
+      "step": 858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003514120867143712,
+      "loss": 1.2817,
+      "step": 859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003512894436187304,
+      "loss": 1.228,
+      "step": 860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035116666739244223,
+      "loss": 1.312,
+      "step": 861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003510437581435463,
+      "loss": 1.2025,
+      "step": 862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035092071598019956,
+      "loss": 1.2523,
+      "step": 863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003507975410106758,
+      "loss": 1.3244,
+      "step": 864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003506742333433656,
+      "loss": 1.3762,
+      "step": 865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003505507930867764,
+      "loss": 1.2588,
+      "step": 866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003504272203495324,
+      "loss": 1.2548,
+      "step": 867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003503035152403743,
+      "loss": 1.1942,
+      "step": 868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035017967786815917,
+      "loss": 1.2061,
+      "step": 869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003500557083418606,
+      "loss": 1.3105,
+      "step": 870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003499316067705685,
+      "loss": 1.2977,
+      "step": 871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034980737326348885,
+      "loss": 1.2818,
+      "step": 872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034968300792994386,
+      "loss": 1.3441,
+      "step": 873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003495585108793717,
+      "loss": 1.3399,
+      "step": 874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003494338822213264,
+      "loss": 1.2834,
+      "step": 875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003493091220654777,
+      "loss": 1.1703,
+      "step": 876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003491842305216114,
+      "loss": 1.3597,
+      "step": 877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034905920769962863,
+      "loss": 1.2565,
+      "step": 878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034893405370954605,
+      "loss": 1.2479,
+      "step": 879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003488087686614959,
+      "loss": 1.3164,
+      "step": 880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003486833526657256,
+      "loss": 1.2071,
+      "step": 881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034855780583259785,
+      "loss": 1.2601,
+      "step": 882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034843212827259055,
+      "loss": 1.3907,
+      "step": 883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003483063200962966,
+      "loss": 1.3849,
+      "step": 884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034818038141442366,
+      "loss": 1.3005,
+      "step": 885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034805431233779456,
+      "loss": 1.3962,
+      "step": 886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034792811297734656,
+      "loss": 1.3415,
+      "step": 887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034780178344413185,
+      "loss": 1.2538,
+      "step": 888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003476753238493169,
+      "loss": 1.2713,
+      "step": 889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003475487343041829,
+      "loss": 1.2862,
+      "step": 890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034742201492012504,
+      "loss": 1.1766,
+      "step": 891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003472951658086532,
+      "loss": 1.3389,
+      "step": 892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003471681870813911,
+      "loss": 1.2903,
+      "step": 893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034704107885007656,
+      "loss": 1.3663,
+      "step": 894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003469138412265614,
+      "loss": 1.2847,
+      "step": 895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034678647432281147,
+      "loss": 1.2622,
+      "step": 896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000346658978250906,
+      "loss": 1.3262,
+      "step": 897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003465313531230383,
+      "loss": 1.558,
+      "step": 898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000346403599051515,
+      "loss": 1.3492,
+      "step": 899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003462757161487562,
+      "loss": 1.574,
+      "step": 900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034614770452729544,
+      "loss": 1.2393,
+      "step": 901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003460195642997795,
+      "loss": 1.3901,
+      "step": 902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034589129557896845,
+      "loss": 1.1878,
+      "step": 903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034576289847773514,
+      "loss": 1.2544,
+      "step": 904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003456343731090657,
+      "loss": 1.301,
+      "step": 905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000345505719586059,
+      "loss": 1.125,
+      "step": 906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003453769380219266,
+      "loss": 1.4236,
+      "step": 907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003452480285299929,
+      "loss": 1.3812,
+      "step": 908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034511899122369476,
+      "loss": 1.2932,
+      "step": 909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034498982621658156,
+      "loss": 1.3466,
+      "step": 910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034486053362231496,
+      "loss": 1.331,
+      "step": 911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034473111355466906,
+      "loss": 1.2288,
+      "step": 912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034460156612753004,
+      "loss": 1.3641,
+      "step": 913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003444718914548961,
+      "loss": 1.1936,
+      "step": 914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003443420896508775,
+      "loss": 1.1695,
+      "step": 915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003442121608296964,
+      "loss": 1.3313,
+      "step": 916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003440821051056866,
+      "loss": 1.2438,
+      "step": 917
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034395192259329365,
+      "loss": 1.0805,
+      "step": 918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034382161340707466,
+      "loss": 1.2497,
+      "step": 919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003436911776616982,
+      "loss": 1.2729,
+      "step": 920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034356061547194427,
+      "loss": 1.599,
+      "step": 921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003434299269527041,
+      "loss": 1.3383,
+      "step": 922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003432991122189801,
+      "loss": 1.301,
+      "step": 923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003431681713858856,
+      "loss": 1.289,
+      "step": 924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034303710456864506,
+      "loss": 1.2264,
+      "step": 925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034290591188259386,
+      "loss": 1.2434,
+      "step": 926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000342774593443178,
+      "loss": 1.5302,
+      "step": 927
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034264314936595407,
+      "loss": 1.2321,
+      "step": 928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034251157976658943,
+      "loss": 1.2389,
+      "step": 929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034237988476086177,
+      "loss": 1.2864,
+      "step": 930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034224806446465927,
+      "loss": 1.3782,
+      "step": 931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034211611899398,
+      "loss": 1.3262,
+      "step": 932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003419840484649326,
+      "loss": 1.4719,
+      "step": 933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003418518529937356,
+      "loss": 1.1969,
+      "step": 934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034171953269671745,
+      "loss": 1.3223,
+      "step": 935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003415870876903164,
+      "loss": 1.1886,
+      "step": 936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003414545180910805,
+      "loss": 1.3605,
+      "step": 937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003413218240156675,
+      "loss": 1.4476,
+      "step": 938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034118900558084464,
+      "loss": 1.3892,
+      "step": 939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003410560629034884,
+      "loss": 1.399,
+      "step": 940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034092299610058506,
+      "loss": 1.2616,
+      "step": 941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034078980528922954,
+      "loss": 1.2666,
+      "step": 942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003406564905866264,
+      "loss": 1.2589,
+      "step": 943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034052305211008883,
+      "loss": 1.3343,
+      "step": 944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034038948997703916,
+      "loss": 1.2685,
+      "step": 945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034025580430500864,
+      "loss": 1.3386,
+      "step": 946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034012199521163677,
+      "loss": 1.2239,
+      "step": 947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003399880628146722,
+      "loss": 1.188,
+      "step": 948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003398540072319717,
+      "loss": 1.2265,
+      "step": 949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003397198285815007,
+      "loss": 1.4477,
+      "step": 950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033958552698133275,
+      "loss": 1.3151,
+      "step": 951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033945110254964965,
+      "loss": 1.3016,
+      "step": 952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033931655540474134,
+      "loss": 1.3619,
+      "step": 953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033918188566500556,
+      "loss": 1.4516,
+      "step": 954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003390470934489482,
+      "loss": 1.2323,
+      "step": 955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033891217887518276,
+      "loss": 1.3176,
+      "step": 956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003387771420624304,
+      "loss": 1.3262,
+      "step": 957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003386419831295199,
+      "loss": 1.1469,
+      "step": 958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003385067021953875,
+      "loss": 1.2472,
+      "step": 959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033837129937907683,
+      "loss": 1.4968,
+      "step": 960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003382357747997387,
+      "loss": 1.1333,
+      "step": 961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033810012857663104,
+      "loss": 1.2604,
+      "step": 962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000337964360829119,
+      "loss": 1.3023,
+      "step": 963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003378284716766745,
+      "loss": 1.5207,
+      "step": 964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033769246123887625,
+      "loss": 1.2229,
+      "step": 965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033755632963540993,
+      "loss": 1.3655,
+      "step": 966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033742007698606765,
+      "loss": 1.1568,
+      "step": 967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000337283703410748,
+      "loss": 1.1576,
+      "step": 968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003371472090294562,
+      "loss": 1.1474,
+      "step": 969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003370105939623035,
+      "loss": 1.3913,
+      "step": 970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003368738583295077,
+      "loss": 1.3809,
+      "step": 971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033673700225139225,
+      "loss": 1.3749,
+      "step": 972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003366000258483869,
+      "loss": 1.3265,
+      "step": 973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033646292924102727,
+      "loss": 1.3178,
+      "step": 974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003363257125499546,
+      "loss": 1.3419,
+      "step": 975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003361883758959159,
+      "loss": 1.3618,
+      "step": 976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003360509193997639,
+      "loss": 1.412,
+      "step": 977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003359133431824565,
+      "loss": 1.0222,
+      "step": 978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000335775647365057,
+      "loss": 1.4098,
+      "step": 979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033563783206873413,
+      "loss": 1.254,
+      "step": 980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003354998974147616,
+      "loss": 1.2008,
+      "step": 981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003353618435245183,
+      "loss": 1.2168,
+      "step": 982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003352236705194878,
+      "loss": 1.3808,
+      "step": 983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033508537852125877,
+      "loss": 1.2343,
+      "step": 984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033494696765152426,
+      "loss": 1.2673,
+      "step": 985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033480843803208236,
+      "loss": 1.3375,
+      "step": 986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003346697897848352,
+      "loss": 1.2231,
+      "step": 987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003345310230317897,
+      "loss": 1.2481,
+      "step": 988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003343921378950568,
+      "loss": 1.2603,
+      "step": 989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003342531344968516,
+      "loss": 1.3673,
+      "step": 990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003341140129594935,
+      "loss": 1.1984,
+      "step": 991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003339747734054057,
+      "loss": 1.2851,
+      "step": 992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033383541595711515,
+      "loss": 1.2433,
+      "step": 993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003336959407372527,
+      "loss": 1.3085,
+      "step": 994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003335563478685529,
+      "loss": 1.316,
+      "step": 995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033341663747385366,
+      "loss": 1.3461,
+      "step": 996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003332768096760964,
+      "loss": 1.1993,
+      "step": 997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003331368645983257,
+      "loss": 1.2851,
+      "step": 998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003329968023636896,
+      "loss": 1.4083,
+      "step": 999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033285662309543915,
+      "loss": 1.2496,
+      "step": 1000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003327163269169282,
+      "loss": 1.2417,
+      "step": 1001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003325759139516137,
+      "loss": 1.1427,
+      "step": 1002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003324353843230553,
+      "loss": 1.3431,
+      "step": 1003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033229473815491535,
+      "loss": 1.432,
+      "step": 1004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003321539755709585,
+      "loss": 1.3202,
+      "step": 1005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033201309669505223,
+      "loss": 1.3503,
+      "step": 1006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003318721016511661,
+      "loss": 1.2937,
+      "step": 1007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003317309905633719,
+      "loss": 1.3295,
+      "step": 1008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033158976355584374,
+      "loss": 1.2516,
+      "step": 1009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003314484207528575,
+      "loss": 1.2637,
+      "step": 1010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003313069622787911,
+      "loss": 1.2511,
+      "step": 1011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033116538825812413,
+      "loss": 1.2837,
+      "step": 1012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000331023698815438,
+      "loss": 1.2124,
+      "step": 1013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003308818940754156,
+      "loss": 1.3627,
+      "step": 1014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033073997416284125,
+      "loss": 1.3624,
+      "step": 1015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033059793920260064,
+      "loss": 1.2367,
+      "step": 1016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003304557893196808,
+      "loss": 1.3101,
+      "step": 1017
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003303135246391698,
+      "loss": 1.3609,
+      "step": 1018
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003301711452862567,
+      "loss": 1.1903,
+      "step": 1019
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033002865138623147,
+      "loss": 1.2604,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000329886043064485,
+      "loss": 1.2647,
+      "step": 1021
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003297433204465088,
+      "loss": 1.218,
+      "step": 1022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032960048365789486,
+      "loss": 1.2481,
+      "step": 1023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003294575328243357,
+      "loss": 1.3108,
+      "step": 1024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032931446807162436,
+      "loss": 1.219,
+      "step": 1025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003291712895256538,
+      "loss": 1.4056,
+      "step": 1026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003290279973124174,
+      "loss": 1.3875,
+      "step": 1027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032888459155800846,
+      "loss": 1.237,
+      "step": 1028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003287410723886201,
+      "loss": 1.381,
+      "step": 1029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003285974399305455,
+      "loss": 1.3096,
+      "step": 1030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003284536943101772,
+      "loss": 1.3856,
+      "step": 1031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003283098356540077,
+      "loss": 1.3452,
+      "step": 1032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003281658640886285,
+      "loss": 1.2547,
+      "step": 1033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032802177974073076,
+      "loss": 1.1737,
+      "step": 1034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032787758273710495,
+      "loss": 1.3169,
+      "step": 1035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003277332732046405,
+      "loss": 1.1254,
+      "step": 1036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032758885127032606,
+      "loss": 1.2771,
+      "step": 1037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032744431706124876,
+      "loss": 1.5731,
+      "step": 1038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003272996707045951,
+      "loss": 1.3528,
+      "step": 1039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003271549123276498,
+      "loss": 1.3671,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003270100420577965,
+      "loss": 1.375,
+      "step": 1041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032686506002251704,
+      "loss": 1.3423,
+      "step": 1042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032671996634939173,
+      "loss": 1.3171,
+      "step": 1043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003265747611660991,
+      "loss": 1.1313,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003264294446004159,
+      "loss": 1.6061,
+      "step": 1045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003262840167802167,
+      "loss": 1.283,
+      "step": 1046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032613847783347414,
+      "loss": 1.1906,
+      "step": 1047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032599282788825854,
+      "loss": 1.3627,
+      "step": 1048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000325847067072738,
+      "loss": 1.3425,
+      "step": 1049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000325701195515178,
+      "loss": 1.404,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032555521334394174,
+      "loss": 1.132,
+      "step": 1051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003254091206874895,
+      "loss": 1.3062,
+      "step": 1052
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000325262917674379,
+      "loss": 1.1618,
+      "step": 1053
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032511660443326485,
+      "loss": 1.2118,
+      "step": 1054
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032497018109289887,
+      "loss": 1.343,
+      "step": 1055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032482364778212966,
+      "loss": 1.4845,
+      "step": 1056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003246770046299025,
+      "loss": 1.1748,
+      "step": 1057
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003245302517652596,
+      "loss": 1.3668,
+      "step": 1058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003243833893173393,
+      "loss": 1.0851,
+      "step": 1059
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032423641741537683,
+      "loss": 1.3395,
+      "step": 1060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003240893361887034,
+      "loss": 1.3099,
+      "step": 1061
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032394214576674657,
+      "loss": 1.4818,
+      "step": 1062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032379484627902995,
+      "loss": 1.3774,
+      "step": 1063
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032364743785517317,
+      "loss": 1.348,
+      "step": 1064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032349992062489163,
+      "loss": 1.293,
+      "step": 1065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032335229471799654,
+      "loss": 1.1384,
+      "step": 1066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003232045602643948,
+      "loss": 1.3489,
+      "step": 1067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032305671739408864,
+      "loss": 1.0874,
+      "step": 1068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000322908766237176,
+      "loss": 1.1842,
+      "step": 1069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032276070692384965,
+      "loss": 1.2471,
+      "step": 1070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003226125395843981,
+      "loss": 1.4924,
+      "step": 1071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003224642643492045,
+      "loss": 1.2523,
+      "step": 1072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000322315881348747,
+      "loss": 1.2867,
+      "step": 1073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003221673907135988,
+      "loss": 1.463,
+      "step": 1074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003220187925744275,
+      "loss": 1.2859,
+      "step": 1075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032187008706199567,
+      "loss": 1.072,
+      "step": 1076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032172127430716007,
+      "loss": 1.3431,
+      "step": 1077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003215723544408719,
+      "loss": 1.3594,
+      "step": 1078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032142332759417665,
+      "loss": 1.3436,
+      "step": 1079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032127419389821403,
+      "loss": 1.3454,
+      "step": 1080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003211249534842176,
+      "loss": 1.3333,
+      "step": 1081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003209756064835149,
+      "loss": 1.3765,
+      "step": 1082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003208261530275274,
+      "loss": 1.4229,
+      "step": 1083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032067659324776996,
+      "loss": 1.045,
+      "step": 1084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032052692727585133,
+      "loss": 1.3477,
+      "step": 1085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003203771552434734,
+      "loss": 1.2725,
+      "step": 1086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032022727728243163,
+      "loss": 1.3644,
+      "step": 1087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003200772935246146,
+      "loss": 1.3548,
+      "step": 1088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000319927204102004,
+      "loss": 1.3695,
+      "step": 1089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003197770091466744,
+      "loss": 1.2361,
+      "step": 1090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003196267087907934,
+      "loss": 1.2436,
+      "step": 1091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031947630316662127,
+      "loss": 1.4142,
+      "step": 1092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031932579240651093,
+      "loss": 1.1265,
+      "step": 1093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031917517664290776,
+      "loss": 1.2766,
+      "step": 1094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003190244560083496,
+      "loss": 1.3594,
+      "step": 1095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031887363063546665,
+      "loss": 1.2358,
+      "step": 1096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003187227006569812,
+      "loss": 1.2309,
+      "step": 1097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003185716662057073,
+      "loss": 1.3444,
+      "step": 1098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031842052741455146,
+      "loss": 1.3418,
+      "step": 1099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031826928441651176,
+      "loss": 1.3559,
+      "step": 1100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031811793734467787,
+      "loss": 1.276,
+      "step": 1101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031796648633223125,
+      "loss": 1.2734,
+      "step": 1102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031781493151244463,
+      "loss": 1.2315,
+      "step": 1103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003176632730186822,
+      "loss": 1.3014,
+      "step": 1104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031751151098439946,
+      "loss": 1.3195,
+      "step": 1105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003173596455431428,
+      "loss": 1.119,
+      "step": 1106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003172076768285497,
+      "loss": 1.3873,
+      "step": 1107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031705560497434883,
+      "loss": 1.4302,
+      "step": 1108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031690343011435895,
+      "loss": 1.2712,
+      "step": 1109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031675115238249013,
+      "loss": 1.2812,
+      "step": 1110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031659877191274257,
+      "loss": 1.2545,
+      "step": 1111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031644628883920696,
+      "loss": 1.2387,
+      "step": 1112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003162937032960644,
+      "loss": 1.2183,
+      "step": 1113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003161410154175861,
+      "loss": 1.2704,
+      "step": 1114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031598822533813315,
+      "loss": 1.308,
+      "step": 1115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003158353331921568,
+      "loss": 1.3652,
+      "step": 1116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031568233911419795,
+      "loss": 1.2184,
+      "step": 1117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003155292432388874,
+      "loss": 1.2892,
+      "step": 1118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031537604570094544,
+      "loss": 1.3794,
+      "step": 1119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031522274663518167,
+      "loss": 1.3267,
+      "step": 1120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003150693461764952,
+      "loss": 1.3899,
+      "step": 1121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003149158444598743,
+      "loss": 1.1889,
+      "step": 1122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031476224162039633,
+      "loss": 1.2387,
+      "step": 1123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003146085377932277,
+      "loss": 1.3521,
+      "step": 1124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003144547331136237,
+      "loss": 1.4573,
+      "step": 1125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003143008277169282,
+      "loss": 1.2822,
+      "step": 1126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031414682173857395,
+      "loss": 1.4768,
+      "step": 1127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003139927153140819,
+      "loss": 1.4126,
+      "step": 1128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031383850857906177,
+      "loss": 1.2651,
+      "step": 1129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003136842016692112,
+      "loss": 1.2556,
+      "step": 1130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003135297947203162,
+      "loss": 1.4173,
+      "step": 1131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031337528786825074,
+      "loss": 1.2882,
+      "step": 1132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003132206812489766,
+      "loss": 1.2622,
+      "step": 1133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031306597499854356,
+      "loss": 1.3657,
+      "step": 1134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031291116925308895,
+      "loss": 1.2806,
+      "step": 1135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031275626414883765,
+      "loss": 1.1607,
+      "step": 1136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031260125982210195,
+      "loss": 1.4189,
+      "step": 1137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031244615640928155,
+      "loss": 1.3719,
+      "step": 1138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031229095404686333,
+      "loss": 1.3134,
+      "step": 1139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003121356528714211,
+      "loss": 1.2751,
+      "step": 1140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003119802530196158,
+      "loss": 1.4261,
+      "step": 1141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000311824754628195,
+      "loss": 1.0773,
+      "step": 1142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003116691578339933,
+      "loss": 1.2708,
+      "step": 1143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031151346277393156,
+      "loss": 1.1252,
+      "step": 1144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031135766958501727,
+      "loss": 1.2471,
+      "step": 1145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031120177840434434,
+      "loss": 1.1925,
+      "step": 1146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003110457893690928,
+      "loss": 1.3149,
+      "step": 1147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031088970261652866,
+      "loss": 1.2653,
+      "step": 1148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031073351828400425,
+      "loss": 1.2462,
+      "step": 1149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003105772365089575,
+      "loss": 1.3278,
+      "step": 1150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003104208574289122,
+      "loss": 1.4563,
+      "step": 1151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003102643811814778,
+      "loss": 1.3731,
+      "step": 1152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031010780790434906,
+      "loss": 1.1842,
+      "step": 1153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030995113773530633,
+      "loss": 1.2848,
+      "step": 1154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003097943708122152,
+      "loss": 1.2751,
+      "step": 1155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003096375072730262,
+      "loss": 1.3703,
+      "step": 1156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030948054725577525,
+      "loss": 1.4415,
+      "step": 1157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003093234908985828,
+      "loss": 1.2485,
+      "step": 1158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003091663383396543,
+      "loss": 1.3278,
+      "step": 1159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003090090897172797,
+      "loss": 1.4458,
+      "step": 1160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003088517451698336,
+      "loss": 1.41,
+      "step": 1161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030869430483577504,
+      "loss": 1.3387,
+      "step": 1162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030853676885364724,
+      "loss": 1.2245,
+      "step": 1163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030837913736207756,
+      "loss": 1.2714,
+      "step": 1164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003082214104997776,
+      "loss": 1.1188,
+      "step": 1165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030806358840554263,
+      "loss": 1.3079,
+      "step": 1166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000307905671218252,
+      "loss": 1.3317,
+      "step": 1167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003077476590768686,
+      "loss": 1.1684,
+      "step": 1168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030758955212043865,
+      "loss": 1.4122,
+      "step": 1169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030743135048809226,
+      "loss": 1.4393,
+      "step": 1170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003072730543190424,
+      "loss": 1.4578,
+      "step": 1171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030711466375258576,
+      "loss": 1.2782,
+      "step": 1172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003069561789281015,
+      "loss": 1.2647,
+      "step": 1173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030679759998505213,
+      "loss": 1.3026,
+      "step": 1174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003066389270629829,
+      "loss": 1.1644,
+      "step": 1175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003064801603015216,
+      "loss": 1.2328,
+      "step": 1176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003063212998403789,
+      "loss": 1.4035,
+      "step": 1177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030616234581934746,
+      "loss": 1.3095,
+      "step": 1178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030600329837830276,
+      "loss": 1.169,
+      "step": 1179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003058441576572023,
+      "loss": 1.318,
+      "step": 1180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030568492379608544,
+      "loss": 1.3981,
+      "step": 1181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003055255969350739,
+      "loss": 1.274,
+      "step": 1182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000305366177214371,
+      "loss": 1.1882,
+      "step": 1183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030520666477426177,
+      "loss": 1.2955,
+      "step": 1184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003050470597551128,
+      "loss": 1.272,
+      "step": 1185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003048873622973725,
+      "loss": 1.2953,
+      "step": 1186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030472757254157,
+      "loss": 1.2946,
+      "step": 1187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003045676906283162,
+      "loss": 1.3895,
+      "step": 1188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003044077166983028,
+      "loss": 1.341,
+      "step": 1189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003042476508923025,
+      "loss": 1.1957,
+      "step": 1190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003040874933511692,
+      "loss": 1.3568,
+      "step": 1191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003039272442158369,
+      "loss": 1.1424,
+      "step": 1192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003037669036273207,
+      "loss": 1.3181,
+      "step": 1193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003036064717267161,
+      "loss": 1.2626,
+      "step": 1194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003034459486551986,
+      "loss": 1.1842,
+      "step": 1195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003032853345540244,
+      "loss": 1.286,
+      "step": 1196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003031246295645296,
+      "loss": 1.277,
+      "step": 1197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003029638338281302,
+      "loss": 1.2859,
+      "step": 1198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003028029474863221,
+      "loss": 1.2462,
+      "step": 1199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030264197068068105,
+      "loss": 1.3812,
+      "step": 1200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030248090355286224,
+      "loss": 1.1455,
+      "step": 1201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003023197462446005,
+      "loss": 1.2305,
+      "step": 1202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030215849889770973,
+      "loss": 1.3159,
+      "step": 1203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030199716165408365,
+      "loss": 1.2649,
+      "step": 1204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003018357346556943,
+      "loss": 1.2146,
+      "step": 1205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030167421804459327,
+      "loss": 1.2537,
+      "step": 1206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030151261196291094,
+      "loss": 1.2755,
+      "step": 1207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030135091655285616,
+      "loss": 1.2214,
+      "step": 1208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003011891319567167,
+      "loss": 1.2191,
+      "step": 1209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030102725831685845,
+      "loss": 1.3751,
+      "step": 1210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003008652957757261,
+      "loss": 1.2667,
+      "step": 1211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003007032444758421,
+      "loss": 1.2965,
+      "step": 1212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030054110455980746,
+      "loss": 1.3041,
+      "step": 1213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003003788761703008,
+      "loss": 1.3534,
+      "step": 1214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003002165594500789,
+      "loss": 1.2117,
+      "step": 1215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030005415454197596,
+      "loss": 1.1901,
+      "step": 1216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000299891661588904,
+      "loss": 1.1561,
+      "step": 1217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002997290807338524,
+      "loss": 1.2261,
+      "step": 1218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002995664121198881,
+      "loss": 1.3413,
+      "step": 1219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029940365589015497,
+      "loss": 1.3146,
+      "step": 1220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002992408121878742,
+      "loss": 1.1425,
+      "step": 1221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029907788115634377,
+      "loss": 1.2512,
+      "step": 1222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029891486293893874,
+      "loss": 1.3225,
+      "step": 1223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002987517576791107,
+      "loss": 1.1951,
+      "step": 1224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029858856552038776,
+      "loss": 1.2961,
+      "step": 1225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002984252866063749,
+      "loss": 1.4236,
+      "step": 1226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002982619210807531,
+      "loss": 1.2997,
+      "step": 1227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002980984690872795,
+      "loss": 1.3658,
+      "step": 1228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002979349307697876,
+      "loss": 1.232,
+      "step": 1229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029777130627218665,
+      "loss": 1.3018,
+      "step": 1230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029760759573846195,
+      "loss": 1.1708,
+      "step": 1231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029744379931267415,
+      "loss": 1.2785,
+      "step": 1232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002972799171389599,
+      "loss": 1.3665,
+      "step": 1233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029711594936153104,
+      "loss": 1.2077,
+      "step": 1234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029695189612467473,
+      "loss": 1.2757,
+      "step": 1235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002967877575727536,
+      "loss": 1.581,
+      "step": 1236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000296623533850205,
+      "loss": 1.2972,
+      "step": 1237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029645922510154147,
+      "loss": 1.097,
+      "step": 1238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002962948314713503,
+      "loss": 1.1225,
+      "step": 1239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029613035310429344,
+      "loss": 1.3598,
+      "step": 1240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029596579014510747,
+      "loss": 1.4496,
+      "step": 1241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002958011427386033,
+      "loss": 1.1856,
+      "step": 1242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002956364110296663,
+      "loss": 1.1327,
+      "step": 1243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000295471595163256,
+      "loss": 1.2706,
+      "step": 1244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002953066952844058,
+      "loss": 1.0502,
+      "step": 1245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002951417115382232,
+      "loss": 1.3791,
+      "step": 1246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029497664406988956,
+      "loss": 1.2774,
+      "step": 1247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029481149302465975,
+      "loss": 1.3849,
+      "step": 1248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029464625854786224,
+      "loss": 1.2147,
+      "step": 1249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029448094078489894,
+      "loss": 1.2812,
+      "step": 1250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002943155398812451,
+      "loss": 1.2631,
+      "step": 1251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029415005598244895,
+      "loss": 1.4973,
+      "step": 1252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000293984489234132,
+      "loss": 1.3568,
+      "step": 1253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002938188397819885,
+      "loss": 1.4428,
+      "step": 1254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002936531077717855,
+      "loss": 1.4517,
+      "step": 1255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002934872933493628,
+      "loss": 1.1518,
+      "step": 1256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002933213966606325,
+      "loss": 1.2996,
+      "step": 1257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029315541785157937,
+      "loss": 1.1926,
+      "step": 1258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029298935706826017,
+      "loss": 1.3596,
+      "step": 1259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000292823214456804,
+      "loss": 1.3128,
+      "step": 1260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002926569901634119,
+      "loss": 1.1898,
+      "step": 1261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029249068433435683,
+      "loss": 1.2303,
+      "step": 1262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029232429711598335,
+      "loss": 1.1898,
+      "step": 1263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002921578286547078,
+      "loss": 1.2474,
+      "step": 1264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029199127909701803,
+      "loss": 1.2588,
+      "step": 1265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000291824648589473,
+      "loss": 1.3361,
+      "step": 1266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029165793727870323,
+      "loss": 1.2607,
+      "step": 1267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002914911453114101,
+      "loss": 1.3589,
+      "step": 1268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029132427283436613,
+      "loss": 1.2802,
+      "step": 1269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002911573199944147,
+      "loss": 1.298,
+      "step": 1270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029099028693846954,
+      "loss": 1.3343,
+      "step": 1271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029082317381351556,
+      "loss": 1.2028,
+      "step": 1272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002906559807666077,
+      "loss": 1.299,
+      "step": 1273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029048870794487133,
+      "loss": 1.2366,
+      "step": 1274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029032135549550203,
+      "loss": 1.2781,
+      "step": 1275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029015392356576556,
+      "loss": 1.2501,
+      "step": 1276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002899864123029974,
+      "loss": 1.2104,
+      "step": 1277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028981882185460304,
+      "loss": 1.402,
+      "step": 1278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002896511523680575,
+      "loss": 1.2098,
+      "step": 1279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002894834039909055,
+      "loss": 1.2936,
+      "step": 1280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002893155768707612,
+      "loss": 1.2416,
+      "step": 1281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002891476711553077,
+      "loss": 1.1811,
+      "step": 1282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002889796869922977,
+      "loss": 1.217,
+      "step": 1283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002888116245295527,
+      "loss": 1.2277,
+      "step": 1284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002886434839149632,
+      "loss": 1.3703,
+      "step": 1285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028847526529648834,
+      "loss": 1.4413,
+      "step": 1286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002883069688221561,
+      "loss": 1.3558,
+      "step": 1287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002881385946400628,
+      "loss": 1.2862,
+      "step": 1288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002879701428983731,
+      "loss": 1.1239,
+      "step": 1289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028780161374532026,
+      "loss": 1.2821,
+      "step": 1290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002876330073292052,
+      "loss": 1.4575,
+      "step": 1291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028746432379839703,
+      "loss": 1.256,
+      "step": 1292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028729556330133284,
+      "loss": 1.1911,
+      "step": 1293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002871267259865173,
+      "loss": 1.0427,
+      "step": 1294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002869578120025227,
+      "loss": 1.3246,
+      "step": 1295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002867888214979888,
+      "loss": 1.3458,
+      "step": 1296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002866197546216226,
+      "loss": 1.3394,
+      "step": 1297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028645061152219865,
+      "loss": 1.1475,
+      "step": 1298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028628139234855815,
+      "loss": 1.3651,
+      "step": 1299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028611209724960956,
+      "loss": 1.4035,
+      "step": 1300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002859427263743279,
+      "loss": 1.2352,
+      "step": 1301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000285773279871755,
+      "loss": 1.3415,
+      "step": 1302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028560375789099936,
+      "loss": 1.3337,
+      "step": 1303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002854341605812357,
+      "loss": 1.2691,
+      "step": 1304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000285264488091705,
+      "loss": 1.2242,
+      "step": 1305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002850947405717146,
+      "loss": 1.267,
+      "step": 1306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028492491817063765,
+      "loss": 1.2977,
+      "step": 1307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002847550210379135,
+      "loss": 1.065,
+      "step": 1308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002845850493230469,
+      "loss": 1.1993,
+      "step": 1309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002844150031756085,
+      "loss": 1.198,
+      "step": 1310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002842448827452342,
+      "loss": 1.1242,
+      "step": 1311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002840746881816256,
+      "loss": 1.1063,
+      "step": 1312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002839044196345492,
+      "loss": 1.2793,
+      "step": 1313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002837340772538368,
+      "loss": 1.3933,
+      "step": 1314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028356366118938506,
+      "loss": 1.2991,
+      "step": 1315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002833931715911556,
+      "loss": 1.3674,
+      "step": 1316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028322260860917474,
+      "loss": 1.181,
+      "step": 1317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002830519723935332,
+      "loss": 1.2209,
+      "step": 1318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002828812630943864,
+      "loss": 1.3288,
+      "step": 1319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028271048086195386,
+      "loss": 1.294,
+      "step": 1320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002825396258465193,
+      "loss": 1.3559,
+      "step": 1321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002823686981984307,
+      "loss": 1.1715,
+      "step": 1322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002821976980680996,
+      "loss": 1.2196,
+      "step": 1323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002820266256060018,
+      "loss": 1.271,
+      "step": 1324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028185548096267617,
+      "loss": 1.5067,
+      "step": 1325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028168426428872554,
+      "loss": 1.3284,
+      "step": 1326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002815129757348161,
+      "loss": 1.2798,
+      "step": 1327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028134161545167693,
+      "loss": 1.3363,
+      "step": 1328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002811701835901007,
+      "loss": 1.2926,
+      "step": 1329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002809986803009427,
+      "loss": 1.1234,
+      "step": 1330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028082710573512133,
+      "loss": 1.2989,
+      "step": 1331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002806554600436175,
+      "loss": 1.3239,
+      "step": 1332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002804837433774748,
+      "loss": 1.46,
+      "step": 1333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028031195588779946,
+      "loss": 1.3381,
+      "step": 1334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002801400977257596,
+      "loss": 1.3373,
+      "step": 1335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027996816904258604,
+      "loss": 1.2269,
+      "step": 1336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002797961699895712,
+      "loss": 1.3315,
+      "step": 1337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002796241007180697,
+      "loss": 1.2946,
+      "step": 1338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000279451961379498,
+      "loss": 1.2438,
+      "step": 1339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027927975212533386,
+      "loss": 1.3579,
+      "step": 1340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000279107473107117,
+      "loss": 1.1844,
+      "step": 1341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027893512447644826,
+      "loss": 1.2203,
+      "step": 1342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002787627063849898,
+      "loss": 1.1703,
+      "step": 1343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002785902189844649,
+      "loss": 1.3602,
+      "step": 1344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002784176624266578,
+      "loss": 1.2242,
+      "step": 1345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027824503686341373,
+      "loss": 1.2376,
+      "step": 1346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027807234244663846,
+      "loss": 1.2791,
+      "step": 1347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002778995793282985,
+      "loss": 1.4444,
+      "step": 1348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002777267476604208,
+      "loss": 1.1773,
+      "step": 1349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002775538475950925,
+      "loss": 1.3084,
+      "step": 1350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027738087928446103,
+      "loss": 1.3622,
+      "step": 1351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027720784288073387,
+      "loss": 1.3555,
+      "step": 1352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027703473853617834,
+      "loss": 1.2227,
+      "step": 1353
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027686156640312177,
+      "loss": 1.304,
+      "step": 1354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002766883266339509,
+      "loss": 1.3173,
+      "step": 1355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002765150193811121,
+      "loss": 1.2876,
+      "step": 1356
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000276341644797111,
+      "loss": 1.2845,
+      "step": 1357
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002761682030345126,
+      "loss": 1.3135,
+      "step": 1358
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027599469424594117,
+      "loss": 1.3552,
+      "step": 1359
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002758211185840796,
+      "loss": 1.1442,
+      "step": 1360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027564747620166985,
+      "loss": 1.2293,
+      "step": 1361
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002754737672515126,
+      "loss": 1.3055,
+      "step": 1362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027529999188646707,
+      "loss": 1.1836,
+      "step": 1363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027512615025945076,
+      "loss": 1.2333,
+      "step": 1364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002749522425234398,
+      "loss": 1.3836,
+      "step": 1365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027477826883146836,
+      "loss": 1.3763,
+      "step": 1366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002746042293366284,
+      "loss": 1.148,
+      "step": 1367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027443012419207024,
+      "loss": 1.2524,
+      "step": 1368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002742559535510016,
+      "loss": 1.1413,
+      "step": 1369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000274081717566688,
+      "loss": 1.2911,
+      "step": 1370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002739074163924524,
+      "loss": 1.3519,
+      "step": 1371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002737330501816752,
+      "loss": 1.3324,
+      "step": 1372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027355861908779397,
+      "loss": 1.2163,
+      "step": 1373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002733841232643033,
+      "loss": 1.2817,
+      "step": 1374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002732095628647549,
+      "loss": 1.1298,
+      "step": 1375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027303493804275736,
+      "loss": 1.3882,
+      "step": 1376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002728602489519756,
+      "loss": 1.2812,
+      "step": 1377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027268549574613153,
+      "loss": 1.2963,
+      "step": 1378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027251067857900305,
+      "loss": 1.4427,
+      "step": 1379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027233579760442485,
+      "loss": 1.2234,
+      "step": 1380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002721608529762873,
+      "loss": 1.4004,
+      "step": 1381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027198584484853703,
+      "loss": 1.2462,
+      "step": 1382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027181077337517644,
+      "loss": 1.3864,
+      "step": 1383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027163563871026377,
+      "loss": 1.2195,
+      "step": 1384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027146044100791287,
+      "loss": 1.4295,
+      "step": 1385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000271285180422293,
+      "loss": 1.4245,
+      "step": 1386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027110985710762864,
+      "loss": 1.3367,
+      "step": 1387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027093447121819984,
+      "loss": 1.1771,
+      "step": 1388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002707590229083412,
+      "loss": 1.3098,
+      "step": 1389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027058351233244266,
+      "loss": 1.2718,
+      "step": 1390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027040793964494883,
+      "loss": 1.3501,
+      "step": 1391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000270232305000359,
+      "loss": 1.2444,
+      "step": 1392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027005660855322676,
+      "loss": 1.4745,
+      "step": 1393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026988085045816045,
+      "loss": 1.2654,
+      "step": 1394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002697050308698224,
+      "loss": 1.2273,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002695291499429291,
+      "loss": 1.1599,
+      "step": 1396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002693532078322511,
+      "loss": 1.228,
+      "step": 1397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026917720469261263,
+      "loss": 1.2249,
+      "step": 1398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002690011406788918,
+      "loss": 1.1718,
+      "step": 1399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002688250159460203,
+      "loss": 1.2489,
+      "step": 1400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026864883064898305,
+      "loss": 1.1626,
+      "step": 1401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026847258494281837,
+      "loss": 1.1026,
+      "step": 1402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002682962789826179,
+      "loss": 1.3496,
+      "step": 1403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002681199129235259,
+      "loss": 1.2368,
+      "step": 1404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002679434869207399,
+      "loss": 1.2877,
+      "step": 1405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002677670011295101,
+      "loss": 1.345,
+      "step": 1406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002675904557051392,
+      "loss": 1.3206,
+      "step": 1407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002674138508029824,
+      "loss": 1.3423,
+      "step": 1408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002672371865784474,
+      "loss": 1.321,
+      "step": 1409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002670604631869938,
+      "loss": 1.3775,
+      "step": 1410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026688368078413353,
+      "loss": 1.3151,
+      "step": 1411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026670683952543037,
+      "loss": 1.4244,
+      "step": 1412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002665299395664998,
+      "loss": 1.1129,
+      "step": 1413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002663529810630091,
+      "loss": 1.2802,
+      "step": 1414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026617596417067706,
+      "loss": 1.2053,
+      "step": 1415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026599888904527364,
+      "loss": 1.3049,
+      "step": 1416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026582175584262033,
+      "loss": 1.3669,
+      "step": 1417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002656445647185895,
+      "loss": 1.2236,
+      "step": 1418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002654673158291046,
+      "loss": 1.28,
+      "step": 1419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026529000933013996,
+      "loss": 1.3546,
+      "step": 1420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026511264537772047,
+      "loss": 1.229,
+      "step": 1421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002649352241279216,
+      "loss": 1.2918,
+      "step": 1422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002647577457368694,
+      "loss": 1.1362,
+      "step": 1423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026458021036074005,
+      "loss": 1.2805,
+      "step": 1424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002644026181557599,
+      "loss": 1.3399,
+      "step": 1425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026422496927820525,
+      "loss": 1.3102,
+      "step": 1426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002640472638844025,
+      "loss": 1.4234,
+      "step": 1427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002638695021307275,
+      "loss": 1.1533,
+      "step": 1428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002636916841736058,
+      "loss": 1.3985,
+      "step": 1429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026351381016951253,
+      "loss": 1.2707,
+      "step": 1430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026333588027497196,
+      "loss": 1.2683,
+      "step": 1431
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002631578946465577,
+      "loss": 1.2254,
+      "step": 1432
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026297985344089217,
+      "loss": 1.3023,
+      "step": 1433
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002628017568146471,
+      "loss": 1.4252,
+      "step": 1434
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002626236049245425,
+      "loss": 1.392,
+      "step": 1435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002624453979273473,
+      "loss": 1.2346,
+      "step": 1436
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026226713597987894,
+      "loss": 1.4382,
+      "step": 1437
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002620888192390031,
+      "loss": 1.2343,
+      "step": 1438
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002619104478616338,
+      "loss": 1.469,
+      "step": 1439
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002617320220047329,
+      "loss": 1.4174,
+      "step": 1440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026155354182531056,
+      "loss": 1.1898,
+      "step": 1441
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002613750074804244,
+      "loss": 1.3007,
+      "step": 1442
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026119641912717995,
+      "loss": 1.2316,
+      "step": 1443
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026101777692273003,
+      "loss": 1.1826,
+      "step": 1444
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026083908102427514,
+      "loss": 1.2212,
+      "step": 1445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026066033158906275,
+      "loss": 1.0836,
+      "step": 1446
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002604815287743876,
+      "loss": 1.1287,
+      "step": 1447
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026030267273759134,
+      "loss": 1.2268,
+      "step": 1448
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002601237636360626,
+      "loss": 1.2117,
+      "step": 1449
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025994480162723645,
+      "loss": 1.2665,
+      "step": 1450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025976578686859466,
+      "loss": 1.405,
+      "step": 1451
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002595867195176655,
+      "loss": 1.2319,
+      "step": 1452
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002594075997320234,
+      "loss": 1.2847,
+      "step": 1453
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002592284276692889,
+      "loss": 1.4436,
+      "step": 1454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025904920348712864,
+      "loss": 1.2914,
+      "step": 1455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002588699273432551,
+      "loss": 1.0885,
+      "step": 1456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025869059939542647,
+      "loss": 1.3028,
+      "step": 1457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025851121980144647,
+      "loss": 1.2751,
+      "step": 1458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025833178871916446,
+      "loss": 1.1309,
+      "step": 1459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002581523063064748,
+      "loss": 1.4444,
+      "step": 1460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025797277272131736,
+      "loss": 1.1246,
+      "step": 1461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025779318812167674,
+      "loss": 1.4859,
+      "step": 1462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025761355266558264,
+      "loss": 1.2375,
+      "step": 1463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025743386651110945,
+      "loss": 1.3053,
+      "step": 1464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002572541298163761,
+      "loss": 1.3145,
+      "step": 1465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002570743427395461,
+      "loss": 1.4664,
+      "step": 1466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002568945054388272,
+      "loss": 1.2516,
+      "step": 1467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002567146180724714,
+      "loss": 1.2222,
+      "step": 1468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002565346807987748,
+      "loss": 1.1783,
+      "step": 1469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025635469377607736,
+      "loss": 1.1859,
+      "step": 1470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002561746571627629,
+      "loss": 1.3463,
+      "step": 1471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002559945711172587,
+      "loss": 1.2379,
+      "step": 1472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002558144357980355,
+      "loss": 1.3192,
+      "step": 1473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025563425136360785,
+      "loss": 1.2627,
+      "step": 1474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025545401797253303,
+      "loss": 1.5463,
+      "step": 1475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002552737357834116,
+      "loss": 1.0699,
+      "step": 1476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002550934049548871,
+      "loss": 1.4229,
+      "step": 1477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025491302564564577,
+      "loss": 1.2421,
+      "step": 1478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002547325980144166,
+      "loss": 1.3587,
+      "step": 1479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025455212221997097,
+      "loss": 1.3002,
+      "step": 1480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025437159842112286,
+      "loss": 1.3108,
+      "step": 1481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025419102677672826,
+      "loss": 1.3039,
+      "step": 1482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025401040744568547,
+      "loss": 1.2134,
+      "step": 1483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025382974058693456,
+      "loss": 1.2693,
+      "step": 1484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002536490263594575,
+      "loss": 1.136,
+      "step": 1485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002534682649222781,
+      "loss": 1.3232,
+      "step": 1486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002532874564344613,
+      "loss": 1.3243,
+      "step": 1487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000253106601055114,
+      "loss": 1.3572,
+      "step": 1488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002529256989433839,
+      "loss": 1.3136,
+      "step": 1489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002527447502584601,
+      "loss": 1.2716,
+      "step": 1490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002525637551595726,
+      "loss": 1.3324,
+      "step": 1491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025238271380599207,
+      "loss": 1.358,
+      "step": 1492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025220162635703023,
+      "loss": 1.2872,
+      "step": 1493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000252020492972039,
+      "loss": 1.2299,
+      "step": 1494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000251839313810411,
+      "loss": 1.3384,
+      "step": 1495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000251658089031579,
+      "loss": 1.4771,
+      "step": 1496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000251476818795016,
+      "loss": 1.3173,
+      "step": 1497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002512955032602348,
+      "loss": 1.2148,
+      "step": 1498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025111414258678833,
+      "loss": 1.2605,
+      "step": 1499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002509327369342691,
+      "loss": 1.1375,
+      "step": 1500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025075128646230907,
+      "loss": 1.1814,
+      "step": 1501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002505697913305799,
+      "loss": 1.2528,
+      "step": 1502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025038825169879255,
+      "loss": 1.2179,
+      "step": 1503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002502066677266968,
+      "loss": 1.2665,
+      "step": 1504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025002503957408177,
+      "loss": 1.2886,
+      "step": 1505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002498433674007753,
+      "loss": 1.186,
+      "step": 1506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024966165136664414,
+      "loss": 1.2297,
+      "step": 1507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024947989163159347,
+      "loss": 1.4288,
+      "step": 1508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002492980883555669,
+      "loss": 1.3706,
+      "step": 1509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002491162416985465,
+      "loss": 1.396,
+      "step": 1510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002489343518205525,
+      "loss": 1.255,
+      "step": 1511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002487524188816429,
+      "loss": 1.498,
+      "step": 1512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024857044304191407,
+      "loss": 1.4793,
+      "step": 1513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024838842446149963,
+      "loss": 1.3647,
+      "step": 1514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002482063633005712,
+      "loss": 1.0928,
+      "step": 1515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002480242597193376,
+      "loss": 1.2955,
+      "step": 1516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024784211387804525,
+      "loss": 1.225,
+      "step": 1517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002476599259369775,
+      "loss": 1.2221,
+      "step": 1518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024747769605645473,
+      "loss": 1.2738,
+      "step": 1519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024729542439683454,
+      "loss": 1.3595,
+      "step": 1520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024711311111851096,
+      "loss": 1.3011,
+      "step": 1521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024693075638191486,
+      "loss": 1.156,
+      "step": 1522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002467483603475135,
+      "loss": 1.3377,
+      "step": 1523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002465659231758104,
+      "loss": 1.1649,
+      "step": 1524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002463834450273455,
+      "loss": 1.2601,
+      "step": 1525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024620092606269465,
+      "loss": 1.2853,
+      "step": 1526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002460183664424695,
+      "loss": 1.1838,
+      "step": 1527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024583576632731776,
+      "loss": 1.039,
+      "step": 1528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002456531258779226,
+      "loss": 1.2585,
+      "step": 1529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024547044525500265,
+      "loss": 1.373,
+      "step": 1530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000245287724619312,
+      "loss": 1.3463,
+      "step": 1531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002451049641316399,
+      "loss": 1.4139,
+      "step": 1532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024492216395281076,
+      "loss": 1.3381,
+      "step": 1533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002447393242436836,
+      "loss": 1.3236,
+      "step": 1534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024455644516515273,
+      "loss": 1.2707,
+      "step": 1535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024437352687814657,
+      "loss": 1.3322,
+      "step": 1536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024419056954362844,
+      "loss": 1.2577,
+      "step": 1537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024400757332259586,
+      "loss": 1.2023,
+      "step": 1538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002438245383760806,
+      "loss": 1.2588,
+      "step": 1539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024364146486514853,
+      "loss": 1.2778,
+      "step": 1540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024345835295089927,
+      "loss": 1.38,
+      "step": 1541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002432752027944666,
+      "loss": 1.2384,
+      "step": 1542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002430920145570176,
+      "loss": 1.1527,
+      "step": 1543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024290878839975305,
+      "loss": 1.2395,
+      "step": 1544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024272552448390697,
+      "loss": 1.4901,
+      "step": 1545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002425422229707468,
+      "loss": 1.1814,
+      "step": 1546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024235888402157293,
+      "loss": 1.2667,
+      "step": 1547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024217550779771858,
+      "loss": 1.3515,
+      "step": 1548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024199209446055004,
+      "loss": 1.3413,
+      "step": 1549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024180864417146613,
+      "loss": 1.4546,
+      "step": 1550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024162515709189812,
+      "loss": 1.2312,
+      "step": 1551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024144163338330965,
+      "loss": 1.0816,
+      "step": 1552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002412580732071968,
+      "loss": 1.3829,
+      "step": 1553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024107447672508747,
+      "loss": 1.1685,
+      "step": 1554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024089084409854175,
+      "loss": 1.2811,
+      "step": 1555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024070717548915126,
+      "loss": 1.2654,
+      "step": 1556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024052347105853962,
+      "loss": 1.3404,
+      "step": 1557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024033973096836166,
+      "loss": 1.3266,
+      "step": 1558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024015595538030385,
+      "loss": 1.2525,
+      "step": 1559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023997214445608362,
+      "loss": 1.2767,
+      "step": 1560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023978829835744975,
+      "loss": 1.2979,
+      "step": 1561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023960441724618195,
+      "loss": 1.2431,
+      "step": 1562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002394205012840904,
+      "loss": 1.2869,
+      "step": 1563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002392365506330165,
+      "loss": 1.199,
+      "step": 1564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023905256545483168,
+      "loss": 1.3731,
+      "step": 1565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023886854591143798,
+      "loss": 1.2633,
+      "step": 1566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023868449216476778,
+      "loss": 1.2219,
+      "step": 1567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023850040437678333,
+      "loss": 1.2433,
+      "step": 1568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000238316282709477,
+      "loss": 1.2935,
+      "step": 1569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002381321273248708,
+      "loss": 1.2649,
+      "step": 1570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023794793838501663,
+      "loss": 1.3732,
+      "step": 1571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023776371605199582,
+      "loss": 1.2717,
+      "step": 1572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023757946048791903,
+      "loss": 1.3932,
+      "step": 1573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023739517185492618,
+      "loss": 1.3593,
+      "step": 1574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023721085031518636,
+      "loss": 1.2325,
+      "step": 1575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023702649603089758,
+      "loss": 1.3133,
+      "step": 1576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023684210916428665,
+      "loss": 1.3409,
+      "step": 1577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023665768987760904,
+      "loss": 1.4334,
+      "step": 1578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002364732383331488,
+      "loss": 1.1466,
+      "step": 1579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023628875469321834,
+      "loss": 1.2477,
+      "step": 1580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023610423912015827,
+      "loss": 1.3581,
+      "step": 1581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023591969177633737,
+      "loss": 1.2247,
+      "step": 1582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023573511282415227,
+      "loss": 1.1193,
+      "step": 1583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023555050242602765,
+      "loss": 1.4805,
+      "step": 1584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023536586074441557,
+      "loss": 1.2427,
+      "step": 1585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023518118794179586,
+      "loss": 1.2371,
+      "step": 1586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002349964841806755,
+      "loss": 1.3099,
+      "step": 1587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023481174962358884,
+      "loss": 1.1764,
+      "step": 1588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002346269844330975,
+      "loss": 1.3289,
+      "step": 1589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023444218877178972,
+      "loss": 1.1783,
+      "step": 1590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023425736280228073,
+      "loss": 1.2568,
+      "step": 1591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023407250668721252,
+      "loss": 1.1521,
+      "step": 1592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002338876205892534,
+      "loss": 1.2124,
+      "step": 1593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023370270467109835,
+      "loss": 1.1642,
+      "step": 1594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002335177590954681,
+      "loss": 1.2952,
+      "step": 1595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023333278402511013,
+      "loss": 1.3528,
+      "step": 1596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002331477796227973,
+      "loss": 1.2231,
+      "step": 1597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023296274605132862,
+      "loss": 1.2521,
+      "step": 1598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002327776834735286,
+      "loss": 1.2415,
+      "step": 1599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023259259205224737,
+      "loss": 1.4175,
+      "step": 1600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023240747195036048,
+      "loss": 1.4344,
+      "step": 1601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002322223233307685,
+      "loss": 1.0662,
+      "step": 1602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023203714635639742,
+      "loss": 1.2483,
+      "step": 1603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023185194119019787,
+      "loss": 1.2279,
+      "step": 1604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023166670799514542,
+      "loss": 1.3382,
+      "step": 1605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002314814469342404,
+      "loss": 1.2369,
+      "step": 1606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023129615817050748,
+      "loss": 1.3321,
+      "step": 1607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023111084186699582,
+      "loss": 1.19,
+      "step": 1608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002309254981867789,
+      "loss": 1.3365,
+      "step": 1609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000230740127292954,
+      "loss": 1.3058,
+      "step": 1610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023055472934864265,
+      "loss": 1.2124,
+      "step": 1611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023036930451699004,
+      "loss": 1.1486,
+      "step": 1612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023018385296116498,
+      "loss": 1.2061,
+      "step": 1613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022999837484435993,
+      "loss": 1.3723,
+      "step": 1614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022981287032979063,
+      "loss": 1.4198,
+      "step": 1615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022962733958069613,
+      "loss": 1.2578,
+      "step": 1616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022944178276033833,
+      "loss": 1.2816,
+      "step": 1617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022925620003200247,
+      "loss": 1.4694,
+      "step": 1618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002290705915589962,
+      "loss": 1.3997,
+      "step": 1619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022888495750465006,
+      "loss": 1.3305,
+      "step": 1620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022869929803231705,
+      "loss": 1.2784,
+      "step": 1621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002285136133053725,
+      "loss": 1.2551,
+      "step": 1622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000228327903487214,
+      "loss": 1.2356,
+      "step": 1623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022814216874126114,
+      "loss": 1.0749,
+      "step": 1624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022795640923095554,
+      "loss": 1.2624,
+      "step": 1625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022777062511976067,
+      "loss": 1.3135,
+      "step": 1626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022758481657116143,
+      "loss": 1.3176,
+      "step": 1627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022739898374866436,
+      "loss": 1.1743,
+      "step": 1628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022721312681579745,
+      "loss": 1.237,
+      "step": 1629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002270272459361097,
+      "loss": 1.3762,
+      "step": 1630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002268413412731714,
+      "loss": 1.3103,
+      "step": 1631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022665541299057363,
+      "loss": 1.2611,
+      "step": 1632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022646946125192826,
+      "loss": 1.4042,
+      "step": 1633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022628348622086782,
+      "loss": 1.2531,
+      "step": 1634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022609748806104533,
+      "loss": 1.1799,
+      "step": 1635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022591146693613418,
+      "loss": 1.0764,
+      "step": 1636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000225725423009828,
+      "loss": 1.2878,
+      "step": 1637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022553935644584048,
+      "loss": 1.3434,
+      "step": 1638
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002253532674079051,
+      "loss": 1.3413,
+      "step": 1639
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002251671560597753,
+      "loss": 1.2488,
+      "step": 1640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022498102256522407,
+      "loss": 1.2143,
+      "step": 1641
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022479486708804376,
+      "loss": 1.2988,
+      "step": 1642
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022460868979204635,
+      "loss": 1.4368,
+      "step": 1643
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022442249084106278,
+      "loss": 1.1675,
+      "step": 1644
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022423627039894317,
+      "loss": 1.1322,
+      "step": 1645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002240500286295564,
+      "loss": 1.2587,
+      "step": 1646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022386376569679035,
+      "loss": 1.2603,
+      "step": 1647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002236774817645514,
+      "loss": 1.3065,
+      "step": 1648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022349117699676426,
+      "loss": 1.2715,
+      "step": 1649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022330485155737232,
+      "loss": 1.3207,
+      "step": 1650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022311850561033682,
+      "loss": 1.1798,
+      "step": 1651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022293213931963725,
+      "loss": 1.2045,
+      "step": 1652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002227457528492709,
+      "loss": 1.1409,
+      "step": 1653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022255934636325288,
+      "loss": 1.2178,
+      "step": 1654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022237292002561597,
+      "loss": 1.2155,
+      "step": 1655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022218647400041016,
+      "loss": 1.2181,
+      "step": 1656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022200000845170315,
+      "loss": 1.1984,
+      "step": 1657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022181352354357946,
+      "loss": 1.2853,
+      "step": 1658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022162701944014086,
+      "loss": 1.2788,
+      "step": 1659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000221440496305506,
+      "loss": 1.2327,
+      "step": 1660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022125395430381013,
+      "loss": 1.2452,
+      "step": 1661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002210673935992053,
+      "loss": 1.0371,
+      "step": 1662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022088081435585978,
+      "loss": 1.1435,
+      "step": 1663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022069421673795843,
+      "loss": 1.2373,
+      "step": 1664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022050760090970215,
+      "loss": 1.257,
+      "step": 1665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002203209670353078,
+      "loss": 1.2781,
+      "step": 1666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022013431527900812,
+      "loss": 1.1342,
+      "step": 1667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002199476458050517,
+      "loss": 1.2176,
+      "step": 1668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002197609587777027,
+      "loss": 1.2025,
+      "step": 1669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021957425436124072,
+      "loss": 1.2804,
+      "step": 1670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021938753271996046,
+      "loss": 1.0952,
+      "step": 1671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021920079401817216,
+      "loss": 1.2449,
+      "step": 1672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002190140384202007,
+      "loss": 1.2866,
+      "step": 1673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002188272660903861,
+      "loss": 1.2772,
+      "step": 1674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002186404771930829,
+      "loss": 1.3276,
+      "step": 1675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021845367189266042,
+      "loss": 1.2041,
+      "step": 1676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021826685035350227,
+      "loss": 1.2675,
+      "step": 1677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021808001274000626,
+      "loss": 1.244,
+      "step": 1678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002178931592165847,
+      "loss": 1.3775,
+      "step": 1679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002177062899476635,
+      "loss": 1.2603,
+      "step": 1680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002175194050976826,
+      "loss": 1.2707,
+      "step": 1681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021733250483109578,
+      "loss": 1.3891,
+      "step": 1682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021714558931237017,
+      "loss": 1.1821,
+      "step": 1683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021695865870598638,
+      "loss": 1.3482,
+      "step": 1684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021677171317643836,
+      "loss": 1.208,
+      "step": 1685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021658475288823317,
+      "loss": 1.2911,
+      "step": 1686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021639777800589084,
+      "loss": 1.4132,
+      "step": 1687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002162107886939442,
+      "loss": 1.2284,
+      "step": 1688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021602378511693888,
+      "loss": 1.2242,
+      "step": 1689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021583676743943295,
+      "loss": 0.9711,
+      "step": 1690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021564973582599697,
+      "loss": 1.2089,
+      "step": 1691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021546269044121376,
+      "loss": 1.3476,
+      "step": 1692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002152756314496782,
+      "loss": 1.3041,
+      "step": 1693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021508855901599723,
+      "loss": 1.3109,
+      "step": 1694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021490147330478952,
+      "loss": 1.1583,
+      "step": 1695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021471437448068547,
+      "loss": 1.2687,
+      "step": 1696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021452726270832702,
+      "loss": 1.2888,
+      "step": 1697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021434013815236754,
+      "loss": 1.3057,
+      "step": 1698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021415300097747157,
+      "loss": 1.3882,
+      "step": 1699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021396585134831483,
+      "loss": 1.3033,
+      "step": 1700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021377868942958398,
+      "loss": 1.1855,
+      "step": 1701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021359151538597647,
+      "loss": 1.3596,
+      "step": 1702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021340432938220038,
+      "loss": 1.3835,
+      "step": 1703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021321713158297447,
+      "loss": 1.2502,
+      "step": 1704
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021302992215302768,
+      "loss": 1.3437,
+      "step": 1705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002128427012570993,
+      "loss": 1.3033,
+      "step": 1706
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021265546905993872,
+      "loss": 1.0176,
+      "step": 1707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021246822572630521,
+      "loss": 1.2505,
+      "step": 1708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021228097142096802,
+      "loss": 1.2274,
+      "step": 1709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021209370630870562,
+      "loss": 1.1612,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021190643055430655,
+      "loss": 1.2879,
+      "step": 1711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002117191443225683,
+      "loss": 1.2691,
+      "step": 1712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002115318477782978,
+      "loss": 1.2698,
+      "step": 1713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002113445410863109,
+      "loss": 1.1628,
+      "step": 1714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002111572244114325,
+      "loss": 1.2225,
+      "step": 1715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021096989791849632,
+      "loss": 1.4484,
+      "step": 1716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002107825617723445,
+      "loss": 1.2193,
+      "step": 1717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002105952161378279,
+      "loss": 1.2302,
+      "step": 1718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002104078611798057,
+      "loss": 1.1346,
+      "step": 1719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002102204970631451,
+      "loss": 1.3832,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021003312395272155,
+      "loss": 1.0992,
+      "step": 1721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020984574201341838,
+      "loss": 1.1517,
+      "step": 1722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020965835141012668,
+      "loss": 1.1362,
+      "step": 1723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020947095230774512,
+      "loss": 1.3007,
+      "step": 1724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002092835448711798,
+      "loss": 1.3205,
+      "step": 1725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002090961292653444,
+      "loss": 1.2617,
+      "step": 1726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020890870565515944,
+      "loss": 1.1763,
+      "step": 1727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020872127420555278,
+      "loss": 1.326,
+      "step": 1728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020853383508145904,
+      "loss": 1.3401,
+      "step": 1729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020834638844781958,
+      "loss": 1.35,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002081589344695825,
+      "loss": 1.3658,
+      "step": 1731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002079714733117021,
+      "loss": 1.1912,
+      "step": 1732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020778400513913936,
+      "loss": 1.2274,
+      "step": 1733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020759653011686108,
+      "loss": 1.3335,
+      "step": 1734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020740904840984024,
+      "loss": 1.448,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020722156018305587,
+      "loss": 1.2446,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020703406560149246,
+      "loss": 1.2779,
+      "step": 1737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002068465648301402,
+      "loss": 1.2144,
+      "step": 1738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020665905803399475,
+      "loss": 1.2173,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020647154537805707,
+      "loss": 1.2191,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020628402702733333,
+      "loss": 1.3689,
+      "step": 1741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020609650314683443,
+      "loss": 1.2727,
+      "step": 1742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020590897390157658,
+      "loss": 1.204,
+      "step": 1743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020572143945658033,
+      "loss": 1.0915,
+      "step": 1744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000205533899976871,
+      "loss": 1.1278,
+      "step": 1745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020534635562747836,
+      "loss": 1.296,
+      "step": 1746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020515880657343626,
+      "loss": 1.2309,
+      "step": 1747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020497125297978306,
+      "loss": 1.2488,
+      "step": 1748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002047836950115606,
+      "loss": 1.3245,
+      "step": 1749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020459613283381508,
+      "loss": 1.2342,
+      "step": 1750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020440856661159607,
+      "loss": 1.3295,
+      "step": 1751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020422099650995682,
+      "loss": 1.2263,
+      "step": 1752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020403342269395398,
+      "loss": 1.2933,
+      "step": 1753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020384584532864755,
+      "loss": 1.2176,
+      "step": 1754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002036582645791005,
+      "loss": 1.1661,
+      "step": 1755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020347068061037882,
+      "loss": 1.1245,
+      "step": 1756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020328309358755138,
+      "loss": 1.0989,
+      "step": 1757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020309550367568987,
+      "loss": 1.3625,
+      "step": 1758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020290791103986817,
+      "loss": 1.2103,
+      "step": 1759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020272031584516283,
+      "loss": 1.2691,
+      "step": 1760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020253271825665262,
+      "loss": 1.1926,
+      "step": 1761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020234511843941842,
+      "loss": 1.2589,
+      "step": 1762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020215751655854294,
+      "loss": 1.2766,
+      "step": 1763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020196991277911076,
+      "loss": 1.3038,
+      "step": 1764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020178230726620834,
+      "loss": 1.4479,
+      "step": 1765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020159470018492334,
+      "loss": 1.2686,
+      "step": 1766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000201407091700345,
+      "loss": 1.2697,
+      "step": 1767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020121948197756378,
+      "loss": 1.2278,
+      "step": 1768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020103187118167113,
+      "loss": 1.3453,
+      "step": 1769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020084425947775964,
+      "loss": 1.3718,
+      "step": 1770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020065664703092234,
+      "loss": 1.2697,
+      "step": 1771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020046903400625337,
+      "loss": 1.3464,
+      "step": 1772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020028142056884707,
+      "loss": 1.1875,
+      "step": 1773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020009380688379823,
+      "loss": 1.1808,
+      "step": 1774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019990619311620187,
+      "loss": 1.4971,
+      "step": 1775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019971857943115298,
+      "loss": 1.3897,
+      "step": 1776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019953096599374665,
+      "loss": 1.1833,
+      "step": 1777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001993433529690777,
+      "loss": 1.2285,
+      "step": 1778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001991557405222404,
+      "loss": 1.1576,
+      "step": 1779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019896812881832886,
+      "loss": 1.3432,
+      "step": 1780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019878051802243627,
+      "loss": 1.2711,
+      "step": 1781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000198592908299655,
+      "loss": 1.2324,
+      "step": 1782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019840529981507673,
+      "loss": 1.1896,
+      "step": 1783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001982176927337917,
+      "loss": 1.3337,
+      "step": 1784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019803008722088923,
+      "loss": 1.1819,
+      "step": 1785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019784248344145713,
+      "loss": 1.1241,
+      "step": 1786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019765488156058163,
+      "loss": 1.1671,
+      "step": 1787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019746728174334735,
+      "loss": 1.4065,
+      "step": 1788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019727968415483719,
+      "loss": 1.1221,
+      "step": 1789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019709208896013188,
+      "loss": 1.4375,
+      "step": 1790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019690449632431015,
+      "loss": 1.1551,
+      "step": 1791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019671690641244867,
+      "loss": 1.2459,
+      "step": 1792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019652931938962122,
+      "loss": 1.1629,
+      "step": 1793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001963417354208996,
+      "loss": 1.3189,
+      "step": 1794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019615415467135253,
+      "loss": 1.2855,
+      "step": 1795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019596657730604599,
+      "loss": 1.1692,
+      "step": 1796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019577900349004328,
+      "loss": 1.2151,
+      "step": 1797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000195591433388404,
+      "loss": 1.1155,
+      "step": 1798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019540386716618497,
+      "loss": 1.1559,
+      "step": 1799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019521630498843946,
+      "loss": 1.265,
+      "step": 1800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000195028747020217,
+      "loss": 1.3046,
+      "step": 1801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001948411934265637,
+      "loss": 1.1584,
+      "step": 1802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001946536443725217,
+      "loss": 1.2032,
+      "step": 1803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019446610002312904,
+      "loss": 1.2944,
+      "step": 1804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019427856054341977,
+      "loss": 1.204,
+      "step": 1805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001940910260984235,
+      "loss": 1.1544,
+      "step": 1806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001939034968531656,
+      "loss": 1.2579,
+      "step": 1807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001937159729726668,
+      "loss": 1.0909,
+      "step": 1808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019352845462194298,
+      "loss": 1.3664,
+      "step": 1809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019334094196600524,
+      "loss": 1.1503,
+      "step": 1810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019315343516985984,
+      "loss": 1.34,
+      "step": 1811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001929659343985076,
+      "loss": 1.2586,
+      "step": 1812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019277843981694413,
+      "loss": 1.3166,
+      "step": 1813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001925909515901598,
+      "loss": 1.2709,
+      "step": 1814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019240346988313897,
+      "loss": 1.3008,
+      "step": 1815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019221599486086071,
+      "loss": 1.1732,
+      "step": 1816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019202852668829797,
+      "loss": 1.1985,
+      "step": 1817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019184106553041754,
+      "loss": 1.2671,
+      "step": 1818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019165361155218044,
+      "loss": 1.4103,
+      "step": 1819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000191466164918541,
+      "loss": 1.1864,
+      "step": 1820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019127872579444727,
+      "loss": 1.215,
+      "step": 1821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019109129434484063,
+      "loss": 1.0866,
+      "step": 1822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019090387073465565,
+      "loss": 1.1126,
+      "step": 1823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019071645512882022,
+      "loss": 1.2809,
+      "step": 1824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019052904769225498,
+      "loss": 1.1704,
+      "step": 1825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001903416485898734,
+      "loss": 1.2752,
+      "step": 1826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019015425798658158,
+      "loss": 1.2683,
+      "step": 1827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018996687604727847,
+      "loss": 1.1622,
+      "step": 1828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018977950293685492,
+      "loss": 1.3076,
+      "step": 1829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001895921388201944,
+      "loss": 1.3259,
+      "step": 1830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018940478386217214,
+      "loss": 1.2067,
+      "step": 1831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018921743822765549,
+      "loss": 1.231,
+      "step": 1832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018903010208150376,
+      "loss": 1.1243,
+      "step": 1833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001888427755885675,
+      "loss": 1.2191,
+      "step": 1834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001886554589136891,
+      "loss": 1.2055,
+      "step": 1835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018846815222170227,
+      "loss": 1.2521,
+      "step": 1836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018828085567743172,
+      "loss": 1.3047,
+      "step": 1837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018809356944569347,
+      "loss": 1.3104,
+      "step": 1838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018790629369129443,
+      "loss": 1.3183,
+      "step": 1839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018771902857903205,
+      "loss": 1.319,
+      "step": 1840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001875317742736948,
+      "loss": 1.2781,
+      "step": 1841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001873445309400613,
+      "loss": 1.2302,
+      "step": 1842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018715729874290072,
+      "loss": 1.2587,
+      "step": 1843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001869700778469724,
+      "loss": 1.2196,
+      "step": 1844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018678286841702557,
+      "loss": 1.1679,
+      "step": 1845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001865956706177996,
+      "loss": 1.2004,
+      "step": 1846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018640848461402357,
+      "loss": 1.1903,
+      "step": 1847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018622131057041604,
+      "loss": 1.2346,
+      "step": 1848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018603414865168513,
+      "loss": 1.1841,
+      "step": 1849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018584699902252848,
+      "loss": 1.285,
+      "step": 1850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001856598618476325,
+      "loss": 1.129,
+      "step": 1851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018547273729167295,
+      "loss": 1.3469,
+      "step": 1852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018528562551931455,
+      "loss": 1.1996,
+      "step": 1853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018509852669521053,
+      "loss": 1.2382,
+      "step": 1854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018491144098400285,
+      "loss": 1.1282,
+      "step": 1855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018472436855032184,
+      "loss": 1.2367,
+      "step": 1856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018453730955878626,
+      "loss": 1.313,
+      "step": 1857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018435026417400308,
+      "loss": 1.1984,
+      "step": 1858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001841632325605671,
+      "loss": 1.2781,
+      "step": 1859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001839762148830612,
+      "loss": 1.3216,
+      "step": 1860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001837892113060559,
+      "loss": 1.3867,
+      "step": 1861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018360222199410918,
+      "loss": 1.271,
+      "step": 1862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018341524711176684,
+      "loss": 1.2565,
+      "step": 1863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001832282868235617,
+      "loss": 1.1966,
+      "step": 1864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018304134129401366,
+      "loss": 1.1986,
+      "step": 1865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018285441068762993,
+      "loss": 1.395,
+      "step": 1866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018266749516890424,
+      "loss": 1.2065,
+      "step": 1867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001824805949023174,
+      "loss": 1.1612,
+      "step": 1868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001822937100523366,
+      "loss": 1.2618,
+      "step": 1869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018210684078341535,
+      "loss": 1.1181,
+      "step": 1870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018191998725999373,
+      "loss": 1.2668,
+      "step": 1871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018173314964649783,
+      "loss": 1.2307,
+      "step": 1872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018154632810733962,
+      "loss": 1.3796,
+      "step": 1873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018135952280691708,
+      "loss": 1.2441,
+      "step": 1874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018117273390961397,
+      "loss": 1.3401,
+      "step": 1875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018098596157979934,
+      "loss": 1.1781,
+      "step": 1876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018079920598182794,
+      "loss": 1.2586,
+      "step": 1877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018061246728003958,
+      "loss": 1.2424,
+      "step": 1878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018042574563875935,
+      "loss": 1.2856,
+      "step": 1879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018023904122229734,
+      "loss": 1.2349,
+      "step": 1880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018005235419494832,
+      "loss": 1.2043,
+      "step": 1881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017986568472099196,
+      "loss": 1.3362,
+      "step": 1882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017967903296469234,
+      "loss": 1.2164,
+      "step": 1883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001794923990902979,
+      "loss": 1.1486,
+      "step": 1884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017930578326204156,
+      "loss": 1.3329,
+      "step": 1885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017911918564414027,
+      "loss": 1.0274,
+      "step": 1886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017893260640079473,
+      "loss": 1.3219,
+      "step": 1887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001787460456961899,
+      "loss": 1.2519,
+      "step": 1888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017855950369449408,
+      "loss": 1.2589,
+      "step": 1889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017837298055985916,
+      "loss": 1.3642,
+      "step": 1890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017818647645642061,
+      "loss": 0.9672,
+      "step": 1891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001779999915482969,
+      "loss": 1.2383,
+      "step": 1892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017781352599958983,
+      "loss": 1.3419,
+      "step": 1893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017762707997438413,
+      "loss": 1.2397,
+      "step": 1894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017744065363674717,
+      "loss": 1.221,
+      "step": 1895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001772542471507291,
+      "loss": 1.3259,
+      "step": 1896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001770678606803628,
+      "loss": 1.3263,
+      "step": 1897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001768814943896632,
+      "loss": 1.2401,
+      "step": 1898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001766951484426277,
+      "loss": 1.176,
+      "step": 1899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017650882300323576,
+      "loss": 1.2862,
+      "step": 1900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017632251823544866,
+      "loss": 1.2146,
+      "step": 1901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001761362343032097,
+      "loss": 1.2925,
+      "step": 1902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017594997137044364,
+      "loss": 1.281,
+      "step": 1903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017576372960105687,
+      "loss": 1.0849,
+      "step": 1904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017557750915893732,
+      "loss": 1.3125,
+      "step": 1905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017539131020795367,
+      "loss": 1.1598,
+      "step": 1906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017520513291195626,
+      "loss": 1.204,
+      "step": 1907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017501897743477603,
+      "loss": 1.2307,
+      "step": 1908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001748328439402247,
+      "loss": 1.2054,
+      "step": 1909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001746467325920949,
+      "loss": 1.2374,
+      "step": 1910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017446064355415957,
+      "loss": 1.3456,
+      "step": 1911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000174274576990172,
+      "loss": 1.2549,
+      "step": 1912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017408853306386579,
+      "loss": 1.3875,
+      "step": 1913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017390251193895472,
+      "loss": 1.413,
+      "step": 1914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017371651377913225,
+      "loss": 1.2291,
+      "step": 1915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017353053874807182,
+      "loss": 1.2189,
+      "step": 1916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017334458700942644,
+      "loss": 1.1537,
+      "step": 1917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017315865872682856,
+      "loss": 1.3019,
+      "step": 1918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017297275406389032,
+      "loss": 1.2471,
+      "step": 1919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017278687318420262,
+      "loss": 1.3551,
+      "step": 1920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017260101625133563,
+      "loss": 1.2943,
+      "step": 1921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001724151834288387,
+      "loss": 1.3282,
+      "step": 1922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001722293748802394,
+      "loss": 1.5019,
+      "step": 1923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017204359076904445,
+      "loss": 1.2648,
+      "step": 1924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017185783125873893,
+      "loss": 1.1127,
+      "step": 1925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017167209651278604,
+      "loss": 1.272,
+      "step": 1926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001714863866946276,
+      "loss": 1.1037,
+      "step": 1927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017130070196768297,
+      "loss": 1.2949,
+      "step": 1928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017111504249534996,
+      "loss": 1.2811,
+      "step": 1929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017092940844100387,
+      "loss": 1.1952,
+      "step": 1930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017074379996799758,
+      "loss": 1.2638,
+      "step": 1931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001705582172396617,
+      "loss": 1.1908,
+      "step": 1932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017037266041930398,
+      "loss": 1.2415,
+      "step": 1933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017018712967020941,
+      "loss": 1.3337,
+      "step": 1934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017000162515564007,
+      "loss": 1.2085,
+      "step": 1935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016981614703883506,
+      "loss": 1.2212,
+      "step": 1936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016963069548301004,
+      "loss": 1.1379,
+      "step": 1937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016944527065135742,
+      "loss": 1.2124,
+      "step": 1938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016925987270704606,
+      "loss": 1.1204,
+      "step": 1939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016907450181322115,
+      "loss": 1.2249,
+      "step": 1940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001688891581330042,
+      "loss": 1.2513,
+      "step": 1941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016870384182949256,
+      "loss": 1.3299,
+      "step": 1942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016851855306575963,
+      "loss": 1.3036,
+      "step": 1943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016833329200485465,
+      "loss": 1.2294,
+      "step": 1944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001681480588098022,
+      "loss": 1.2135,
+      "step": 1945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016796285364360263,
+      "loss": 1.1569,
+      "step": 1946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016777767666923156,
+      "loss": 1.1965,
+      "step": 1947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016759252804963954,
+      "loss": 1.1513,
+      "step": 1948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001674074079477526,
+      "loss": 1.1926,
+      "step": 1949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016722231652647145,
+      "loss": 1.189,
+      "step": 1950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016703725394867145,
+      "loss": 1.114,
+      "step": 1951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016685222037720278,
+      "loss": 1.3933,
+      "step": 1952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016666721597488992,
+      "loss": 1.1315,
+      "step": 1953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016648224090453186,
+      "loss": 1.2984,
+      "step": 1954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016629729532890175,
+      "loss": 1.4407,
+      "step": 1955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001661123794107466,
+      "loss": 1.1905,
+      "step": 1956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016592749331278747,
+      "loss": 1.1534,
+      "step": 1957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016574263719771928,
+      "loss": 1.2697,
+      "step": 1958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016555781122821035,
+      "loss": 1.3054,
+      "step": 1959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016537301556690252,
+      "loss": 1.2623,
+      "step": 1960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001651882503764112,
+      "loss": 1.2823,
+      "step": 1961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016500351581932456,
+      "loss": 1.2938,
+      "step": 1962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016481881205820424,
+      "loss": 1.279,
+      "step": 1963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016463413925558445,
+      "loss": 1.1894,
+      "step": 1964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016444949757397234,
+      "loss": 1.1466,
+      "step": 1965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016426488717584777,
+      "loss": 1.3923,
+      "step": 1966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001640803082236627,
+      "loss": 1.2416,
+      "step": 1967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016389576087984177,
+      "loss": 1.2472,
+      "step": 1968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016371124530678176,
+      "loss": 1.0552,
+      "step": 1969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016352676166685122,
+      "loss": 1.4276,
+      "step": 1970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016334231012239098,
+      "loss": 1.3438,
+      "step": 1971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016315789083571342,
+      "loss": 1.3187,
+      "step": 1972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016297350396910247,
+      "loss": 1.1763,
+      "step": 1973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016278914968481374,
+      "loss": 1.2755,
+      "step": 1974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016260482814507387,
+      "loss": 1.247,
+      "step": 1975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016242053951208104,
+      "loss": 1.2318,
+      "step": 1976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016223628394800425,
+      "loss": 1.215,
+      "step": 1977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016205206161498342,
+      "loss": 1.0413,
+      "step": 1978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016186787267512918,
+      "loss": 1.4229,
+      "step": 1979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016168371729052307,
+      "loss": 1.2017,
+      "step": 1980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016149959562321672,
+      "loss": 1.2663,
+      "step": 1981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001613155078352322,
+      "loss": 1.1215,
+      "step": 1982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016113145408856207,
+      "loss": 1.1153,
+      "step": 1983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001609474345451684,
+      "loss": 1.1303,
+      "step": 1984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016076344936698356,
+      "loss": 1.1184,
+      "step": 1985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016057949871590963,
+      "loss": 1.2001,
+      "step": 1986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016039558275381813,
+      "loss": 1.2126,
+      "step": 1987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016021170164255027,
+      "loss": 1.2566,
+      "step": 1988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016002785554391642,
+      "loss": 1.2464,
+      "step": 1989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015984404461969623,
+      "loss": 1.3024,
+      "step": 1990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001596602690316384,
+      "loss": 1.2382,
+      "step": 1991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001594765289414604,
+      "loss": 1.2314,
+      "step": 1992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015929282451084876,
+      "loss": 1.2229,
+      "step": 1993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015910915590145835,
+      "loss": 1.3042,
+      "step": 1994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001589255232749126,
+      "loss": 1.3198,
+      "step": 1995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015874192679280323,
+      "loss": 1.4086,
+      "step": 1996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001585583666166904,
+      "loss": 1.1759,
+      "step": 1997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015837484290810196,
+      "loss": 1.2343,
+      "step": 1998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015819135582853395,
+      "loss": 1.2502,
+      "step": 1999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015800790553945003,
+      "loss": 1.2231,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3449,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 5.997684734757437e+17,
+  "trial_name": null,
+  "trial_params": null
+}