diff --git "a/checkpoint-3000/trainer_state.json" "b/checkpoint-3000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3000/trainer_state.json"
@@ -0,0 +1,18019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9639328460117278,
+  "eval_steps": 500,
+  "global_step": 3000,
+  "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": 1.2657,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.3437,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 1.36,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.3141,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.3516,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 1.316,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.4411,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.358,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 1.2083,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.3071,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.2787,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 1.14,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.3849,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.263,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.1749,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.1589,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.1352,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-05,
+      "loss": 1.3085,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6e-05,
+      "loss": 1.156,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-05,
+      "loss": 1.0312,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4e-05,
+      "loss": 1.1446,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.0777,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.1365,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6e-05,
+      "loss": 0.9902,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001,
+      "loss": 0.9993,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 1.1102,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 1.128,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 0.9163,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000116,
+      "loss": 0.9935,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012,
+      "loss": 1.0898,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000124,
+      "loss": 1.0969,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.0162,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000132,
+      "loss": 1.033,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 0.9363,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014,
+      "loss": 1.0791,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000144,
+      "loss": 0.9156,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000148,
+      "loss": 0.9933,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000152,
+      "loss": 0.9657,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 1.0068,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016,
+      "loss": 0.998,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000164,
+      "loss": 0.9114,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000168,
+      "loss": 0.9493,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000172,
+      "loss": 0.9821,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.1043,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018,
+      "loss": 0.9675,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 0.9996,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000188,
+      "loss": 0.9619,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000192,
+      "loss": 1.0249,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000196,
+      "loss": 0.9913,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7959,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 1.0425,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00020800000000000001,
+      "loss": 1.0633,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00021200000000000003,
+      "loss": 0.9647,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00021600000000000002,
+      "loss": 0.9763,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00022000000000000003,
+      "loss": 1.0168,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00022400000000000002,
+      "loss": 0.9852,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 1.0007,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000232,
+      "loss": 1.0332,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000236,
+      "loss": 0.9753,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00024,
+      "loss": 1.0567,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000244,
+      "loss": 0.8954,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000248,
+      "loss": 1.0207,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000252,
+      "loss": 0.9293,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00025600000000000004,
+      "loss": 0.9731,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 0.9343,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000264,
+      "loss": 0.9203,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000268,
+      "loss": 0.8022,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 1.0727,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000276,
+      "loss": 0.9156,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00028,
+      "loss": 1.0,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000284,
+      "loss": 0.9119,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000288,
+      "loss": 0.9978,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000292,
+      "loss": 0.8991,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000296,
+      "loss": 0.9723,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 1.0487,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000304,
+      "loss": 1.022,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000308,
+      "loss": 1.1011,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00031200000000000005,
+      "loss": 0.9025,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00031600000000000004,
+      "loss": 0.9647,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00032,
+      "loss": 1.0676,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000324,
+      "loss": 0.9855,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000328,
+      "loss": 0.9698,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000332,
+      "loss": 1.0027,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000336,
+      "loss": 0.9025,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00034,
+      "loss": 0.8639,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000344,
+      "loss": 0.8467,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000348,
+      "loss": 0.9356,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00035200000000000005,
+      "loss": 0.9875,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00035600000000000003,
+      "loss": 0.848,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00036,
+      "loss": 0.936,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000364,
+      "loss": 1.0434,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00036800000000000005,
+      "loss": 0.9038,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00037200000000000004,
+      "loss": 0.956,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000376,
+      "loss": 1.0062,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00038,
+      "loss": 1.0327,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000384,
+      "loss": 0.9512,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000388,
+      "loss": 0.8612,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000392,
+      "loss": 0.9073,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 0.8613,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 0.9321,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039999989120980106,
+      "loss": 0.9222,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039999956483932256,
+      "loss": 0.9403,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999990208889196,
+      "loss": 0.8481,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999982593591839,
+      "loss": 0.9945,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399997280250944,
+      "loss": 1.025,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399996083565265,
+      "loss": 0.95,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999946693034488,
+      "loss": 0.9418,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399993037467034,
+      "loss": 0.9057,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039999118805779583,
+      "loss": 1.0409,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999891210777463,
+      "loss": 0.9333,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039998683652913406,
+      "loss": 0.9392,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999843344144445,
+      "loss": 0.9495,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039998161473639974,
+      "loss": 0.8723,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999786774979584,
+      "loss": 0.8309,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039997552270231595,
+      "loss": 0.919,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039997215035290456,
+      "loss": 0.8769,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999685604533929,
+      "loss": 0.9288,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039996475300768647,
+      "loss": 0.8767,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999607280199275,
+      "loss": 0.7836,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999564854944946,
+      "loss": 0.9204,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999520254360034,
+      "loss": 0.9527,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039994734784930594,
+      "loss": 0.9527,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039994245273949094,
+      "loss": 0.8504,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999373401118838,
+      "loss": 0.945,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039993200997204666,
+      "loss": 1.025,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999264623257781,
+      "loss": 0.9359,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999206971791134,
+      "loss": 1.0323,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039991471453832447,
+      "loss": 0.9431,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039990851440991986,
+      "loss": 0.8771,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039990209680064475,
+      "loss": 0.8938,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039989546171748086,
+      "loss": 0.8502,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039988860916764644,
+      "loss": 0.8991,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039988153915859645,
+      "loss": 0.8344,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039987425169802233,
+      "loss": 0.9809,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998667467938522,
+      "loss": 0.9492,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039985902445425055,
+      "loss": 0.9401,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039985108468761863,
+      "loss": 0.9753,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039984292750259407,
+      "loss": 0.9621,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039983455290805106,
+      "loss": 0.9769,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998259609131004,
+      "loss": 0.9319,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003998171515270893,
+      "loss": 0.82,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003998081247596016,
+      "loss": 0.9276,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003997988806204574,
+      "loss": 0.8548,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039978941911971344,
+      "loss": 0.8974,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039977974026766304,
+      "loss": 0.9384,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039976984407483565,
+      "loss": 0.9008,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003997597305519975,
+      "loss": 0.8802,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039974939971015107,
+      "loss": 0.8109,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003997388515605353,
+      "loss": 0.9472,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003997280861146256,
+      "loss": 0.8003,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039971710338413357,
+      "loss": 0.8599,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039970590338100746,
+      "loss": 0.9683,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039969448611743176,
+      "loss": 0.9864,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039968285160582736,
+      "loss": 0.9497,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003996709998588514,
+      "loss": 0.9325,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039965893088939747,
+      "loss": 0.9167,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039964664471059543,
+      "loss": 0.9643,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039963414133581133,
+      "loss": 0.9718,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039962142077864783,
+      "loss": 0.8662,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039960848305294343,
+      "loss": 0.9711,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039959532817277323,
+      "loss": 1.0001,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995819561524485,
+      "loss": 0.8947,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995683670065165,
+      "loss": 0.8309,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039955456074976107,
+      "loss": 0.852,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000399540537397202,
+      "loss": 0.9198,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995262969640952,
+      "loss": 0.7994,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995118394659331,
+      "loss": 0.9199,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003994971649184438,
+      "loss": 0.9513,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039948227333759203,
+      "loss": 0.8567,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003994671647395782,
+      "loss": 0.8006,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039945183914083895,
+      "loss": 0.8287,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003994362965580472,
+      "loss": 0.8466,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003994205370081116,
+      "loss": 0.8329,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039940456050817705,
+      "loss": 1.0046,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003993883670756244,
+      "loss": 0.8684,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039937195672807055,
+      "loss": 0.8472,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039935532948336827,
+      "loss": 0.9505,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003993384853596065,
+      "loss": 0.9022,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039932142437510993,
+      "loss": 0.7937,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039930414654843925,
+      "loss": 0.9626,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039928665189839095,
+      "loss": 0.8674,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003992689404439977,
+      "loss": 0.8539,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039925101220452763,
+      "loss": 0.8264,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000399232867199485,
+      "loss": 0.9836,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039921450544860973,
+      "loss": 0.9411,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003991959269718777,
+      "loss": 0.8775,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003991771317895004,
+      "loss": 0.9446,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039915811992192514,
+      "loss": 0.9181,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039913889138983505,
+      "loss": 0.8666,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003991194462141488,
+      "loss": 0.9672,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039909978441602094,
+      "loss": 0.9331,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039907990601684146,
+      "loss": 0.9452,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039905981103823614,
+      "loss": 0.9052,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003990394995020664,
+      "loss": 0.9412,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039901897143042916,
+      "loss": 0.8547,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039899822684565697,
+      "loss": 1.0161,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003989772657703179,
+      "loss": 0.9392,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003989560882272155,
+      "loss": 0.8965,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039893469423938894,
+      "loss": 1.0206,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039891308383011275,
+      "loss": 0.9311,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003988912570228969,
+      "loss": 0.8172,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003988692138414869,
+      "loss": 0.859,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003988469543098634,
+      "loss": 1.0104,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003988244784522428,
+      "loss": 1.0001,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039880178629307655,
+      "loss": 0.927,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039877887785705145,
+      "loss": 0.9264,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039875575316908964,
+      "loss": 0.932,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003987324122543486,
+      "loss": 0.8801,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003987088551382209,
+      "loss": 0.9196,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003986850818463343,
+      "loss": 0.963,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003986610924045519,
+      "loss": 0.9475,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003986368868389719,
+      "loss": 0.8879,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039861246517592747,
+      "loss": 0.9186,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039858782744198705,
+      "loss": 0.8936,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003985629736639541,
+      "loss": 0.8012,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039853790386886704,
+      "loss": 0.793,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003985126180839994,
+      "loss": 0.8714,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039848711633685963,
+      "loss": 0.9624,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039846139865519107,
+      "loss": 0.829,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039843546506697206,
+      "loss": 0.9879,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003984093156004159,
+      "loss": 0.7905,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003983829502839705,
+      "loss": 0.784,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039835636914631887,
+      "loss": 0.9276,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039832957221637856,
+      "loss": 0.9115,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003983025595233021,
+      "loss": 0.8191,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003982753310964766,
+      "loss": 0.7928,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039824788696552393,
+      "loss": 0.9718,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003982202271603006,
+      "loss": 0.8687,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003981923517108979,
+      "loss": 0.848,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003981642606476414,
+      "loss": 0.9673,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039813595400109154,
+      "loss": 0.9139,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003981074318020431,
+      "loss": 0.8629,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039807869408152544,
+      "loss": 0.8691,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039804974087080246,
+      "loss": 1.0033,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003980205722013724,
+      "loss": 0.8981,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039799118810496785,
+      "loss": 0.9327,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003979615886135559,
+      "loss": 0.8811,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003979317737593378,
+      "loss": 0.9356,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039790174357474934,
+      "loss": 0.9039,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039787149809246023,
+      "loss": 0.9237,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039784103734537475,
+      "loss": 0.8911,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003978103613666311,
+      "loss": 0.9065,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003977794701896018,
+      "loss": 0.8854,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003977483638478934,
+      "loss": 0.8741,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039771704237534657,
+      "loss": 0.9292,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000397685505806036,
+      "loss": 0.9336,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039765375417427027,
+      "loss": 0.8789,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003976217875145923,
+      "loss": 0.911,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039758960586177845,
+      "loss": 0.9081,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003975572092508393,
+      "loss": 0.8847,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003975245977170191,
+      "loss": 0.845,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003974917712957962,
+      "loss": 0.9535,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039745873002288237,
+      "loss": 0.8964,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039742547393422327,
+      "loss": 0.8121,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039739200306599835,
+      "loss": 1.0468,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003973583174546206,
+      "loss": 0.947,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003973244171367366,
+      "loss": 0.9131,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039729030214922666,
+      "loss": 0.9067,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039725597252920447,
+      "loss": 0.9515,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039722142831401735,
+      "loss": 0.8074,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000397186669541246,
+      "loss": 1.0034,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039715169624870457,
+      "loss": 0.8533,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039711650847444054,
+      "loss": 0.8817,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039708110625673483,
+      "loss": 0.8399,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003970454896341015,
+      "loss": 0.9647,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039700965864528803,
+      "loss": 0.8725,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003969736133292749,
+      "loss": 0.8182,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039693735372527605,
+      "loss": 0.9308,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003969008798727382,
+      "loss": 0.9354,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039686419181134147,
+      "loss": 0.9201,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039682728958099876,
+      "loss": 0.9911,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003967901732218562,
+      "loss": 1.007,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039675284277429257,
+      "loss": 0.7949,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039671529827891997,
+      "loss": 0.9236,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003966775397765829,
+      "loss": 0.9022,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003966395673083591,
+      "loss": 0.8306,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003966013809155588,
+      "loss": 0.8714,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003965629806397251,
+      "loss": 0.8972,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039652436652263374,
+      "loss": 0.8911,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000396485538606293,
+      "loss": 0.7929,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039644649693294386,
+      "loss": 0.8207,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039640724154506,
+      "loss": 0.7907,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039636777248534724,
+      "loss": 0.9995,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039632808979674414,
+      "loss": 0.8193,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039628819352242157,
+      "loss": 0.8342,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039624808370578273,
+      "loss": 0.9908,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003962077603904632,
+      "loss": 0.9465,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039616722362033086,
+      "loss": 0.9834,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003961264734394856,
+      "loss": 0.8794,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003960855098922597,
+      "loss": 0.9583,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003960443330232176,
+      "loss": 0.8139,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039600294287715543,
+      "loss": 0.9616,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039596133949910186,
+      "loss": 0.7973,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039591952293431716,
+      "loss": 0.8786,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039587749322829365,
+      "loss": 0.9212,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039583525042675554,
+      "loss": 0.9779,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039579279457565897,
+      "loss": 0.942,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039575012572119165,
+      "loss": 0.8292,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003957072439097731,
+      "loss": 0.9525,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039566414918805454,
+      "loss": 0.897,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039562084160291877,
+      "loss": 0.8055,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003955773212014803,
+      "loss": 0.8881,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000395533588031085,
+      "loss": 0.9634,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039548964213931027,
+      "loss": 0.9648,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003954454835739649,
+      "loss": 0.8019,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003954011123830891,
+      "loss": 0.9155,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039535652861495446,
+      "loss": 0.9295,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039531173231806364,
+      "loss": 0.9069,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003952667235411507,
+      "loss": 0.9526,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003952215023331807,
+      "loss": 0.855,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039517606874335,
+      "loss": 0.7897,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039513042282108575,
+      "loss": 0.9163,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039508456461604633,
+      "loss": 0.8563,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039503849417812096,
+      "loss": 0.983,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003949922115574297,
+      "loss": 0.8935,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039494571680432364,
+      "loss": 0.9759,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039489900996938436,
+      "loss": 0.908,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003948520911034244,
+      "loss": 0.822,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039480496025748687,
+      "loss": 0.8766,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003947576174828456,
+      "loss": 0.9377,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039471006283100466,
+      "loss": 0.8959,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003946622963536991,
+      "loss": 0.7954,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039461431810289407,
+      "loss": 0.9065,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003945661281307852,
+      "loss": 0.962,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003945177264897984,
+      "loss": 0.8344,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039446911323259,
+      "loss": 0.8187,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003944202884120464,
+      "loss": 1.0666,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039437125208128425,
+      "loss": 0.8967,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003943220042936503,
+      "loss": 0.8107,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039427254510272124,
+      "loss": 0.8605,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039422287456230393,
+      "loss": 0.92,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003941729927264349,
+      "loss": 0.8836,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003941228996493808,
+      "loss": 0.9853,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000394072595385638,
+      "loss": 0.9218,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003940220799899326,
+      "loss": 0.8688,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003939713535172203,
+      "loss": 0.8721,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003939204160226867,
+      "loss": 0.8698,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003938692675617466,
+      "loss": 0.8577,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039381790819004473,
+      "loss": 0.9867,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039376633796345484,
+      "loss": 0.9041,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003937145569380805,
+      "loss": 0.7354,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003936625651702542,
+      "loss": 0.8855,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039361036271653803,
+      "loss": 0.9028,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039355794963372306,
+      "loss": 0.8015,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003935053259788296,
+      "loss": 0.8947,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000393452491809107,
+      "loss": 0.8689,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003933994471820337,
+      "loss": 0.8185,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039334619215531714,
+      "loss": 0.8066,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039329272678689336,
+      "loss": 0.8773,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003932390511349276,
+      "loss": 0.9042,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039318516525781367,
+      "loss": 0.8677,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003931310692141741,
+      "loss": 0.9295,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039307676306286015,
+      "loss": 0.775,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039302224686295155,
+      "loss": 0.8986,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039296752067375653,
+      "loss": 0.9004,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039291258455481187,
+      "loss": 0.8638,
+      "step": 356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003928574385658827,
+      "loss": 0.8749,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003928020827669625,
+      "loss": 0.8102,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039274651721827274,
+      "loss": 0.8012,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039269074198026354,
+      "loss": 0.8959,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003926347571136128,
+      "loss": 0.7712,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003925785626792265,
+      "loss": 0.8142,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003925221587382387,
+      "loss": 0.9226,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003924655453520115,
+      "loss": 0.9043,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003924087225821345,
+      "loss": 0.783,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003923516904904254,
+      "loss": 0.7943,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003922944491389296,
+      "loss": 0.864,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039223699858992,
+      "loss": 0.9989,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039217933890589726,
+      "loss": 0.8635,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003921214701495893,
+      "loss": 0.9517,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003920633923839518,
+      "loss": 0.8266,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039200510567216753,
+      "loss": 0.8487,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003919466100776469,
+      "loss": 0.824,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003918879056640273,
+      "loss": 1.0438,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003918289924951734,
+      "loss": 0.8513,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003917698706351768,
+      "loss": 0.798,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003917105401483565,
+      "loss": 0.9453,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003916510010992582,
+      "loss": 0.7973,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039159125355265446,
+      "loss": 0.9476,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039153129757354483,
+      "loss": 0.9031,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003914711332271555,
+      "loss": 0.803,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003914107605789394,
+      "loss": 0.9528,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039135017969457617,
+      "loss": 0.8449,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039128939063997166,
+      "loss": 0.924,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039122839348125856,
+      "loss": 0.8911,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039116718828479573,
+      "loss": 0.9219,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039110577511716847,
+      "loss": 0.8366,
+      "step": 387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039104415404518823,
+      "loss": 0.8744,
+      "step": 388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003909823251358927,
+      "loss": 0.8698,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039092028845654576,
+      "loss": 0.9401,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003908580440746372,
+      "loss": 0.7745,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003907955920578827,
+      "loss": 0.8802,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039073293247422406,
+      "loss": 0.9421,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039067006539182875,
+      "loss": 0.8955,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039060699087908993,
+      "loss": 0.9476,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003905437090046265,
+      "loss": 0.8878,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039048021983728305,
+      "loss": 0.7741,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003904165234461294,
+      "loss": 0.7793,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039035261990046116,
+      "loss": 0.7884,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039028850926979894,
+      "loss": 0.9254,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039022419162388895,
+      "loss": 0.9179,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039015966703270246,
+      "loss": 0.8275,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00039009493556643586,
+      "loss": 0.8671,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003900299972955107,
+      "loss": 0.9008,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003899648522905734,
+      "loss": 0.9103,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038989950062249536,
+      "loss": 0.9569,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038983394236237283,
+      "loss": 0.9073,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038976817758152666,
+      "loss": 0.9446,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038970220635150263,
+      "loss": 0.8786,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038963602874407093,
+      "loss": 0.9294,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003895696448312262,
+      "loss": 0.9477,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003895030546851878,
+      "loss": 0.8019,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038943625837839914,
+      "loss": 0.7865,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038936925598352814,
+      "loss": 0.848,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038930204757346677,
+      "loss": 0.8365,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003892346332213313,
+      "loss": 0.8079,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003891670130004618,
+      "loss": 0.8427,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003890991869844226,
+      "loss": 0.8727,
+      "step": 418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038903115524700164,
+      "loss": 0.8346,
+      "step": 419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003889629178622108,
+      "loss": 0.8733,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003888944749042857,
+      "loss": 0.8903,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038882582644768555,
+      "loss": 0.8878,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003887569725670931,
+      "loss": 0.804,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003886879133374147,
+      "loss": 0.8073,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038861864883377997,
+      "loss": 0.7739,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000388549179131542,
+      "loss": 0.8039,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003884795043062768,
+      "loss": 0.8864,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000388409624433784,
+      "loss": 0.8783,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003883395395900859,
+      "loss": 0.9202,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000388269249851428,
+      "loss": 0.915,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038819875529427864,
+      "loss": 0.8437,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038812805599532904,
+      "loss": 0.9702,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000388057152031493,
+      "loss": 0.9099,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038798604347990706,
+      "loss": 0.7732,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003879147304179305,
+      "loss": 0.854,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038784321292314485,
+      "loss": 0.9645,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038777149107335415,
+      "loss": 0.8328,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038769956494658477,
+      "loss": 0.9106,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038762743462108525,
+      "loss": 0.9311,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003875551001753263,
+      "loss": 0.8557,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038748256168800077,
+      "loss": 0.7491,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038740981923802336,
+      "loss": 0.8508,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003873368729045308,
+      "loss": 0.9311,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003872637227668815,
+      "loss": 0.8909,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038719036890465556,
+      "loss": 0.8551,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038711681139765497,
+      "loss": 0.8305,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000387043050325903,
+      "loss": 0.8161,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003869690857696444,
+      "loss": 0.8339,
+      "step": 448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038689491780934544,
+      "loss": 0.8997,
+      "step": 449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038682054652569364,
+      "loss": 0.9392,
+      "step": 450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038674597199959753,
+      "loss": 0.992,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000386671194312187,
+      "loss": 0.7871,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003865962135448128,
+      "loss": 0.8712,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038652102977904665,
+      "loss": 0.8069,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003864456430966811,
+      "loss": 0.8363,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038637005357972955,
+      "loss": 0.9135,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038629426131042585,
+      "loss": 0.8358,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003862182663712247,
+      "loss": 0.8965,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003861420688448011,
+      "loss": 0.85,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003860656688140504,
+      "loss": 0.7916,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038598906636208856,
+      "loss": 0.8736,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038591226157225135,
+      "loss": 0.886,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038583525452809484,
+      "loss": 0.8305,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003857580453133953,
+      "loss": 0.8673,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038568063401214867,
+      "loss": 0.8373,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038560302070857086,
+      "loss": 0.7912,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038552520548709767,
+      "loss": 1.0658,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003854471884323842,
+      "loss": 0.8741,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003853689696293056,
+      "loss": 0.9008,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003852905491629561,
+      "loss": 0.8815,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003852119271186496,
+      "loss": 0.8425,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038513310358191904,
+      "loss": 0.9865,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038505407863851683,
+      "loss": 0.876,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003849748523744143,
+      "loss": 0.8976,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003848954248758018,
+      "loss": 0.9194,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038481579622908884,
+      "loss": 0.9721,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038473596652090344,
+      "loss": 0.858,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038465593583809254,
+      "loss": 0.8634,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003845757042677216,
+      "loss": 0.9856,
+      "step": 479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003844952718970749,
+      "loss": 0.8541,
+      "step": 480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003844146388136548,
+      "loss": 0.9375,
+      "step": 481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003843338051051822,
+      "loss": 0.8544,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003842527708595964,
+      "loss": 0.8664,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003841715361650545,
+      "loss": 0.8911,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003840901011099321,
+      "loss": 1.0164,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003840084657828224,
+      "loss": 0.8909,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003839266302725367,
+      "loss": 0.8598,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038384459466810403,
+      "loss": 0.9048,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003837623590587711,
+      "loss": 0.9209,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038367992353400215,
+      "loss": 0.8624,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038359728818347893,
+      "loss": 0.845,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038351445309710064,
+      "loss": 0.9102,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038343141836498376,
+      "loss": 0.7573,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038334818407746187,
+      "loss": 0.7679,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038326475032508583,
+      "loss": 0.8423,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038318111719862327,
+      "loss": 1.08,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003830972847890588,
+      "loss": 0.8998,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003830132531875941,
+      "loss": 0.9221,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000382929022485647,
+      "loss": 0.9048,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038284459277485245,
+      "loss": 0.9558,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038275996414706176,
+      "loss": 0.8278,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003826751366943424,
+      "loss": 0.8083,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003825901105089783,
+      "loss": 0.8273,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003825048856834699,
+      "loss": 0.8761,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038241946231053317,
+      "loss": 0.8983,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003823338404831005,
+      "loss": 0.8438,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038224802029432006,
+      "loss": 0.8881,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038216200183755573,
+      "loss": 0.8629,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003820757852063872,
+      "loss": 0.8392,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038198937049460966,
+      "loss": 0.8125,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038190275779623396,
+      "loss": 0.8714,
+      "step": 511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038181594720548617,
+      "loss": 0.8679,
+      "step": 512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038172893881680773,
+      "loss": 0.8586,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038164173272485513,
+      "loss": 0.7672,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003815543290245001,
+      "loss": 0.8364,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038146672781082946,
+      "loss": 0.8618,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003813789291791445,
+      "loss": 0.7789,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038129093322496167,
+      "loss": 0.8062,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003812027400440119,
+      "loss": 0.813,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003811143497322408,
+      "loss": 0.7999,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003810257623858083,
+      "loss": 0.7777,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038093697810108863,
+      "loss": 0.8994,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038084799697467065,
+      "loss": 0.9804,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003807588191033569,
+      "loss": 0.972,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003806694445841643,
+      "loss": 0.7448,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003805798735143234,
+      "loss": 0.8814,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038049010599127894,
+      "loss": 0.8725,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038040014211268905,
+      "loss": 0.8859,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003803099819764257,
+      "loss": 0.8942,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003802196256805742,
+      "loss": 0.8659,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038012907332343336,
+      "loss": 0.8377,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003800383250035153,
+      "loss": 0.8606,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003799473808195453,
+      "loss": 0.9519,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037985624087046174,
+      "loss": 0.8865,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037976490525541587,
+      "loss": 0.775,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003796733740737719,
+      "loss": 0.7932,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003795816474251069,
+      "loss": 0.8236,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037948972540921035,
+      "loss": 0.789,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003793976081260845,
+      "loss": 0.7518,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003793052956759438,
+      "loss": 0.8992,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003792127881592153,
+      "loss": 0.8724,
+      "step": 541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003791200856765379,
+      "loss": 0.9242,
+      "step": 542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037902718832876305,
+      "loss": 0.924,
+      "step": 543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037893409621695377,
+      "loss": 0.8329,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037884080944238527,
+      "loss": 0.9247,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037874732810654435,
+      "loss": 0.9149,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037865365231112955,
+      "loss": 0.8474,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037855978215805105,
+      "loss": 0.8038,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003784657177494303,
+      "loss": 0.9424,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003783714591876001,
+      "loss": 0.8447,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003782770065751046,
+      "loss": 0.9013,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000378182360014699,
+      "loss": 0.7183,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037808751960934946,
+      "loss": 0.838,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000377992485462233,
+      "loss": 0.761,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003778972576767375,
+      "loss": 0.9029,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037780183635646145,
+      "loss": 0.8397,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003777062216052139,
+      "loss": 0.8368,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037761041352701435,
+      "loss": 0.7481,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003775144122260925,
+      "loss": 0.8826,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037741821780688854,
+      "loss": 0.8586,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003773218303740524,
+      "loss": 0.8049,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037722525003244426,
+      "loss": 0.885,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037712847688713396,
+      "loss": 0.8859,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003770315110434013,
+      "loss": 0.8624,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003769343526067356,
+      "loss": 0.7726,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037683700168283566,
+      "loss": 0.987,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037673945837760975,
+      "loss": 0.956,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003766417227971755,
+      "loss": 0.8685,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003765437950478596,
+      "loss": 0.8198,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037644567523619784,
+      "loss": 0.8599,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037634736346893493,
+      "loss": 0.8393,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037624885985302443,
+      "loss": 0.763,
+      "step": 572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037615016449562877,
+      "loss": 0.7665,
+      "step": 573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037605127750411866,
+      "loss": 0.7949,
+      "step": 574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003759521989860734,
+      "loss": 0.8983,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037585292904928095,
+      "loss": 0.8462,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003757534678017371,
+      "loss": 0.8263,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037565381535164594,
+      "loss": 0.8677,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003755539718074196,
+      "loss": 0.905,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003754539372776781,
+      "loss": 0.8563,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037535371187124916,
+      "loss": 0.7549,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037525329569716823,
+      "loss": 0.7977,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003751526888646782,
+      "loss": 0.8718,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003750518914832296,
+      "loss": 0.8308,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003749509036624799,
+      "loss": 0.7667,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003749509036624799,
+      "loss": 0.8372,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037484972551229414,
+      "loss": 0.8181,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037474835714274403,
+      "loss": 0.8221,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003746467986641086,
+      "loss": 0.7982,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037454505018687343,
+      "loss": 0.985,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003744431118217309,
+      "loss": 0.9419,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037434098367958,
+      "loss": 0.8054,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037423866587152604,
+      "loss": 0.6858,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003741361585088809,
+      "loss": 0.7536,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003740334617031624,
+      "loss": 0.7969,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003739305755660948,
+      "loss": 0.8488,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003738275002096079,
+      "loss": 0.7857,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003737242357458378,
+      "loss": 0.7643,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037362078228712594,
+      "loss": 0.9017,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037351713994601964,
+      "loss": 0.8742,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003734133088352716,
+      "loss": 0.7811,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003733092890678398,
+      "loss": 0.8087,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037320508075688776,
+      "loss": 0.8569,
+      "step": 603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003731006840157837,
+      "loss": 0.7588,
+      "step": 604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003729960989581011,
+      "loss": 0.8922,
+      "step": 605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003728913256976184,
+      "loss": 0.8906,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037278636434831834,
+      "loss": 0.9138,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037268121502438884,
+      "loss": 0.8418,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000372575877840222,
+      "loss": 0.7998,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003724703529104143,
+      "loss": 0.7894,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003723646403497665,
+      "loss": 0.8701,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003722587402732836,
+      "loss": 0.8259,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003721526527961745,
+      "loss": 0.8895,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003720463780338518,
+      "loss": 0.8377,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003719399161019323,
+      "loss": 0.8336,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003718332671162359,
+      "loss": 0.89,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037172643119278635,
+      "loss": 0.8688,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003716194084478107,
+      "loss": 0.8439,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037151219899773913,
+      "loss": 0.9474,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003714048029592051,
+      "loss": 0.8335,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037129722044904485,
+      "loss": 0.7712,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037118945158429775,
+      "loss": 0.8899,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003710814964822057,
+      "loss": 0.8085,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037097335526021326,
+      "loss": 0.932,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003708650280359675,
+      "loss": 0.9699,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037075651492731777,
+      "loss": 0.9081,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003706478160523157,
+      "loss": 0.8561,
+      "step": 627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037053893152921516,
+      "loss": 0.8584,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003704298614764716,
+      "loss": 0.8502,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037032060601274285,
+      "loss": 0.9321,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037021116525688783,
+      "loss": 0.8667,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037010153932796755,
+      "loss": 0.8387,
+      "step": 632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036999172834524416,
+      "loss": 0.9069,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003698817324281814,
+      "loss": 0.9113,
+      "step": 634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003697715516964439,
+      "loss": 0.8418,
+      "step": 635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036966118626989753,
+      "loss": 0.8736,
+      "step": 636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003695506362686092,
+      "loss": 0.844,
+      "step": 637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036943990181284624,
+      "loss": 0.9095,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036932898302307704,
+      "loss": 0.8292,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036921788001997033,
+      "loss": 0.8187,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003691065929243953,
+      "loss": 0.8798,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003689951218574214,
+      "loss": 0.8831,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036888346694031825,
+      "loss": 0.8461,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003687716282945554,
+      "loss": 0.8471,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003686596060418023,
+      "loss": 0.8344,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036854740030392834,
+      "loss": 0.8508,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003684350112030022,
+      "loss": 0.8821,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036832243886129233,
+      "loss": 0.9081,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036820968340126633,
+      "loss": 0.8737,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003680967449455911,
+      "loss": 0.821,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036798362361713266,
+      "loss": 0.7257,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036787031953895586,
+      "loss": 0.9577,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036775683283432446,
+      "loss": 0.8178,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003676431636267009,
+      "loss": 0.8339,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036752931203974605,
+      "loss": 0.7418,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036741527819731935,
+      "loss": 0.8474,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003673010622234785,
+      "loss": 0.7201,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036718666424247914,
+      "loss": 0.7666,
+      "step": 658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003670720843787752,
+      "loss": 0.827,
+      "step": 659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003669573227570182,
+      "loss": 0.8094,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036684237950205774,
+      "loss": 0.8275,
+      "step": 661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036672725473894057,
+      "loss": 0.8146,
+      "step": 662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036661194859291134,
+      "loss": 0.8276,
+      "step": 663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003664964611894117,
+      "loss": 0.874,
+      "step": 664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003663807926540808,
+      "loss": 0.8029,
+      "step": 665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003662649431127545,
+      "loss": 0.9547,
+      "step": 666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003661489126914658,
+      "loss": 0.8731,
+      "step": 667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036603270151644446,
+      "loss": 0.8343,
+      "step": 668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003659163097141168,
+      "loss": 0.9127,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036579973741110574,
+      "loss": 0.816,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003656829847342305,
+      "loss": 0.7588,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003655660518105065,
+      "loss": 0.852,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003654489387671454,
+      "loss": 0.9752,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036533164573155464,
+      "loss": 0.8811,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036521417283133756,
+      "loss": 0.878,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003650965201942932,
+      "loss": 0.8771,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000364978687948416,
+      "loss": 0.8717,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000364860676221896,
+      "loss": 0.7815,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036474248514311837,
+      "loss": 0.8158,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036462411484066337,
+      "loss": 0.8017,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003645055654433064,
+      "loss": 0.8559,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003643868370800174,
+      "loss": 1.1111,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003642679298799614,
+      "loss": 0.8341,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003641488439724976,
+      "loss": 0.8022,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003640295794871799,
+      "loss": 0.784,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003639101365537563,
+      "loss": 0.7322,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036379051530216917,
+      "loss": 0.8684,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003636707158625544,
+      "loss": 0.8278,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003635507383652424,
+      "loss": 0.799,
+      "step": 689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036343058294075674,
+      "loss": 0.7488,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036331024971981464,
+      "loss": 0.8708,
+      "step": 691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003631897388333271,
+      "loss": 0.8592,
+      "step": 692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036306905041239793,
+      "loss": 0.8426,
+      "step": 693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036294818458832437,
+      "loss": 0.8637,
+      "step": 694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036282714149259675,
+      "loss": 0.8659,
+      "step": 695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036270592125689787,
+      "loss": 0.8341,
+      "step": 696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003625845240131035,
+      "loss": 0.9353,
+      "step": 697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003624629498932821,
+      "loss": 0.8689,
+      "step": 698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003623411990296943,
+      "loss": 0.9254,
+      "step": 699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000362219271554793,
+      "loss": 0.8996,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036209716760122356,
+      "loss": 0.85,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000361974887301823,
+      "loss": 0.747,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003618524307896202,
+      "loss": 0.9207,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000361729798197836,
+      "loss": 0.8674,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003616069896598826,
+      "loss": 0.8255,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036148400530936365,
+      "loss": 0.8453,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000361360845280074,
+      "loss": 0.8795,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003612375097059998,
+      "loss": 0.8662,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003611139987213179,
+      "loss": 0.8051,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036099031246039637,
+      "loss": 0.8703,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036086645105779365,
+      "loss": 0.9089,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00036074241464825876,
+      "loss": 0.794,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003606182033667311,
+      "loss": 0.8949,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003604938173483405,
+      "loss": 0.8288,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003603692567284068,
+      "loss": 0.8011,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003602445216424396,
+      "loss": 0.8918,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003601196122261385,
+      "loss": 0.8629,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035999452861539265,
+      "loss": 0.8362,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003598692709462809,
+      "loss": 0.8489,
+      "step": 719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035974383935507117,
+      "loss": 0.9985,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035961823397822083,
+      "loss": 0.8823,
+      "step": 721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003594924549523762,
+      "loss": 0.9075,
+      "step": 722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003593665024143725,
+      "loss": 0.8649,
+      "step": 723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003592403765012338,
+      "loss": 0.8092,
+      "step": 724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035911407735017266,
+      "loss": 0.7945,
+      "step": 725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003589876050985902,
+      "loss": 0.788,
+      "step": 726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035886095988407594,
+      "loss": 0.8422,
+      "step": 727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003587341418444073,
+      "loss": 0.8105,
+      "step": 728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035860715111755,
+      "loss": 0.8472,
+      "step": 729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003584799878416574,
+      "loss": 0.837,
+      "step": 730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035835265215507076,
+      "loss": 0.8994,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035822514419631883,
+      "loss": 0.9595,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003580974641041177,
+      "loss": 0.691,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003579696120173709,
+      "loss": 0.916,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003578415880751689,
+      "loss": 0.9035,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003577133924167892,
+      "loss": 0.8385,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003575850251816961,
+      "loss": 0.9465,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003574564865095406,
+      "loss": 0.8935,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003573277765401602,
+      "loss": 0.9041,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035719889541357866,
+      "loss": 0.8029,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003570698432700061,
+      "loss": 0.9223,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003569406202498385,
+      "loss": 0.8984,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000356811226493658,
+      "loss": 0.8721,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003566816621422322,
+      "loss": 0.9213,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035655192733651436,
+      "loss": 0.9111,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035642202221764337,
+      "loss": 0.8711,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003562919469269432,
+      "loss": 0.8008,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035616170160592307,
+      "loss": 0.8749,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035603128639627706,
+      "loss": 0.8195,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003559007014398842,
+      "loss": 0.7713,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035576994687880803,
+      "loss": 0.774,
+      "step": 751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035563902285529674,
+      "loss": 0.9666,
+      "step": 752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003555079295117829,
+      "loss": 0.9097,
+      "step": 753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003553766669908831,
+      "loss": 0.8152,
+      "step": 754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003552452354353982,
+      "loss": 0.8376,
+      "step": 755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035511363498831285,
+      "loss": 0.8726,
+      "step": 756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035498186579279535,
+      "loss": 0.8028,
+      "step": 757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035484992799219775,
+      "loss": 0.8549,
+      "step": 758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003547178217300554,
+      "loss": 0.807,
+      "step": 759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000354585547150087,
+      "loss": 0.945,
+      "step": 760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035445310439619434,
+      "loss": 0.824,
+      "step": 761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003543204936124621,
+      "loss": 0.8177,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003541877149431578,
+      "loss": 0.8912,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035405476853273167,
+      "loss": 0.8163,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035392165452581646,
+      "loss": 0.8177,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000353788373067227,
+      "loss": 0.9633,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035365492430196056,
+      "loss": 0.7769,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035352130837519623,
+      "loss": 0.957,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003533875254322951,
+      "loss": 0.7852,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003532535756187999,
+      "loss": 0.8522,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003531194590804349,
+      "loss": 0.9567,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003529851759631058,
+      "loss": 0.7786,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035285072641289927,
+      "loss": 0.905,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003527161105760835,
+      "loss": 0.7657,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035258132859910716,
+      "loss": 0.8252,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003524463806285999,
+      "loss": 0.8044,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035231126681137183,
+      "loss": 0.7189,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035217598729441363,
+      "loss": 0.7879,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003520405422248961,
+      "loss": 0.8305,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035190493175017023,
+      "loss": 0.8327,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035176915601776683,
+      "loss": 0.9649,
+      "step": 781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003516332151753967,
+      "loss": 0.8298,
+      "step": 782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003514971093709501,
+      "loss": 0.7862,
+      "step": 783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003513608387524968,
+      "loss": 0.9092,
+      "step": 784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035122440346828593,
+      "loss": 0.8341,
+      "step": 785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035108780366674566,
+      "loss": 0.7885,
+      "step": 786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003509510394964832,
+      "loss": 0.8698,
+      "step": 787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003508141111062845,
+      "loss": 0.8872,
+      "step": 788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035067701864511435,
+      "loss": 0.8214,
+      "step": 789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035053976226211587,
+      "loss": 0.8639,
+      "step": 790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003504023421066105,
+      "loss": 0.9633,
+      "step": 791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003502647583280979,
+      "loss": 0.871,
+      "step": 792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003501270110762558,
+      "loss": 0.8796,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034998910050093964,
+      "loss": 0.9565,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034985102675218264,
+      "loss": 0.8753,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003497127899801956,
+      "loss": 0.8454,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034957439033536647,
+      "loss": 0.7163,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003494358279682605,
+      "loss": 0.7594,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034929710302962,
+      "loss": 0.9273,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003491582156703641,
+      "loss": 0.8454,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003490191660415886,
+      "loss": 0.9674,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034887995429456586,
+      "loss": 0.9816,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034874058058074467,
+      "loss": 0.8607,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003486010450517499,
+      "loss": 0.8103,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034846134785938273,
+      "loss": 0.967,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003483214891556198,
+      "loss": 0.7579,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003481814690926137,
+      "loss": 0.8765,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034804128782269254,
+      "loss": 0.752,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003479009454983599,
+      "loss": 0.8758,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034776044227229443,
+      "loss": 0.8623,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034761977829734977,
+      "loss": 0.876,
+      "step": 811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034747895372655473,
+      "loss": 0.9562,
+      "step": 812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003473379687131124,
+      "loss": 0.8418,
+      "step": 813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034719682341040087,
+      "loss": 0.7452,
+      "step": 814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003470555179719723,
+      "loss": 0.8602,
+      "step": 815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003469140525515531,
+      "loss": 0.9122,
+      "step": 816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003467724273030439,
+      "loss": 0.8434,
+      "step": 817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034663064238051906,
+      "loss": 0.8704,
+      "step": 818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034648869793822663,
+      "loss": 0.9878,
+      "step": 819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034634659413058824,
+      "loss": 0.8218,
+      "step": 820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000346204331112199,
+      "loss": 0.8904,
+      "step": 821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034606190903782705,
+      "loss": 0.8496,
+      "step": 822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003459193280624137,
+      "loss": 0.9083,
+      "step": 823
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034577658834107297,
+      "loss": 0.8706,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003456336900290919,
+      "loss": 1.0112,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003454906332819296,
+      "loss": 0.8652,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000345347418255218,
+      "loss": 0.7367,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003452040451047608,
+      "loss": 0.807,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003450605139865341,
+      "loss": 0.6758,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034491682505668564,
+      "loss": 0.7894,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034477297847153494,
+      "loss": 0.7763,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003446289743875729,
+      "loss": 0.9043,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003444848129614619,
+      "loss": 0.8893,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003443404943500354,
+      "loss": 0.7544,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034419601871029797,
+      "loss": 0.8653,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003440513861994249,
+      "loss": 0.7641,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003439065969747622,
+      "loss": 0.9232,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003437616511938264,
+      "loss": 0.911,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034361654901430416,
+      "loss": 0.852,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034347129059405254,
+      "loss": 0.9509,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034332587609109855,
+      "loss": 0.9502,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034318030566363877,
+      "loss": 1.0095,
+      "step": 842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003430345794700396,
+      "loss": 0.7783,
+      "step": 843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034288869766883683,
+      "loss": 0.8203,
+      "step": 844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034274266041873564,
+      "loss": 0.7332,
+      "step": 845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003425964678786102,
+      "loss": 0.787,
+      "step": 846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003424501202075036,
+      "loss": 0.7788,
+      "step": 847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003423036175646279,
+      "loss": 0.9261,
+      "step": 848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003421569601093635,
+      "loss": 0.8367,
+      "step": 849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034201014800125943,
+      "loss": 0.8294,
+      "step": 850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003418631814000327,
+      "loss": 0.7939,
+      "step": 851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034171606046556885,
+      "loss": 0.7923,
+      "step": 852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034156878535792083,
+      "loss": 0.6938,
+      "step": 853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003414213562373095,
+      "loss": 0.7506,
+      "step": 854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003412737732641234,
+      "loss": 0.8668,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034112603659891837,
+      "loss": 0.8236,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034097814640241734,
+      "loss": 0.9373,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003408301028355103,
+      "loss": 0.7648,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003406819060592543,
+      "loss": 0.8274,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003405335562348728,
+      "loss": 0.8275,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034038505352375575,
+      "loss": 0.7064,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00034023639808745986,
+      "loss": 0.8462,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003400875900877074,
+      "loss": 0.8809,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000339938629686387,
+      "loss": 0.8646,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000339789517045553,
+      "loss": 0.8116,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003396402523274252,
+      "loss": 0.7636,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033949083569438914,
+      "loss": 0.7475,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033934126730899547,
+      "loss": 0.7363,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003391915473339598,
+      "loss": 0.8759,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003390416759321629,
+      "loss": 0.7858,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033889165326665016,
+      "loss": 0.7589,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033874147950063146,
+      "loss": 0.8158,
+      "step": 872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033859115479748127,
+      "loss": 0.8137,
+      "step": 873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000338440679320738,
+      "loss": 0.8563,
+      "step": 874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003382900532341043,
+      "loss": 0.9272,
+      "step": 875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003381392767014465,
+      "loss": 0.9314,
+      "step": 876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003379883498867948,
+      "loss": 0.9395,
+      "step": 877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003378372729543427,
+      "loss": 0.8853,
+      "step": 878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003376860460684472,
+      "loss": 0.8548,
+      "step": 879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033753466939362816,
+      "loss": 0.8496,
+      "step": 880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003373831430945687,
+      "loss": 0.6175,
+      "step": 881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003372314673361146,
+      "loss": 0.9498,
+      "step": 882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003370796422832742,
+      "loss": 0.6941,
+      "step": 883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003369276681012181,
+      "loss": 0.8087,
+      "step": 884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033677554495527957,
+      "loss": 0.842,
+      "step": 885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033662327301095356,
+      "loss": 0.9118,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000336470852433897,
+      "loss": 0.8112,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003363182833899286,
+      "loss": 0.743,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003361655660450285,
+      "loss": 0.8734,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033601270056533817,
+      "loss": 0.9156,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033585968711716035,
+      "loss": 0.8937,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003357065258669586,
+      "loss": 0.7123,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033555321698135735,
+      "loss": 0.7208,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003353997606271417,
+      "loss": 0.867,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033524615697125715,
+      "loss": 0.7774,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033509240618080925,
+      "loss": 0.869,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003349385084230639,
+      "loss": 0.8403,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033478446386544695,
+      "loss": 0.7981,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003346302726755434,
+      "loss": 0.8904,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003344759350210985,
+      "loss": 0.8726,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003343214510700163,
+      "loss": 0.9383,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003341668209903603,
+      "loss": 0.9004,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033401204495035275,
+      "loss": 0.8046,
+      "step": 903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033385712311837494,
+      "loss": 0.9042,
+      "step": 904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003337020556629666,
+      "loss": 0.8001,
+      "step": 905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003335468427528259,
+      "loss": 0.8293,
+      "step": 906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003333914845568092,
+      "loss": 0.907,
+      "step": 907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003332359812439312,
+      "loss": 0.8235,
+      "step": 908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033308033298336406,
+      "loss": 0.8405,
+      "step": 909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003329245399444379,
+      "loss": 0.8134,
+      "step": 910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003327686022966403,
+      "loss": 0.7416,
+      "step": 911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033261252020961614,
+      "loss": 0.7594,
+      "step": 912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003324562938531674,
+      "loss": 0.7302,
+      "step": 913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033229992339725307,
+      "loss": 0.8645,
+      "step": 914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033214340901198885,
+      "loss": 0.9696,
+      "step": 915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033198675086764713,
+      "loss": 0.883,
+      "step": 916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003318299491346565,
+      "loss": 0.8972,
+      "step": 917
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033167300398360196,
+      "loss": 0.6976,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003315159155852244,
+      "loss": 0.9773,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003313586841104207,
+      "loss": 0.8069,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003312013097302431,
+      "loss": 0.892,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003310437926158997,
+      "loss": 0.8575,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003308861329387536,
+      "loss": 0.7721,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033072833087032313,
+      "loss": 0.8745,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003305703865822814,
+      "loss": 0.8907,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033041230024645626,
+      "loss": 0.6967,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00033025407203483034,
+      "loss": 0.8958,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003300957021195402,
+      "loss": 0.7812,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032993719067287705,
+      "loss": 0.8936,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003297785378672855,
+      "loss": 0.9179,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032961974387536453,
+      "loss": 0.7174,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003294608088698663,
+      "loss": 0.7054,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032930173302369653,
+      "loss": 0.7867,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003291425165099142,
+      "loss": 0.7851,
+      "step": 934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032898315950173114,
+      "loss": 0.7648,
+      "step": 935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032882366217251236,
+      "loss": 0.7821,
+      "step": 936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032866402469577515,
+      "loss": 0.8156,
+      "step": 937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003285042472451895,
+      "loss": 0.9293,
+      "step": 938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003283443299945776,
+      "loss": 0.7455,
+      "step": 939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003281842731179138,
+      "loss": 0.8548,
+      "step": 940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032802407678932425,
+      "loss": 0.8553,
+      "step": 941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032786374118308687,
+      "loss": 0.7802,
+      "step": 942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003277032664736311,
+      "loss": 0.8425,
+      "step": 943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003275426528355376,
+      "loss": 0.8487,
+      "step": 944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003273819004435384,
+      "loss": 0.8559,
+      "step": 945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032722100947251624,
+      "loss": 0.7486,
+      "step": 946
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003270599800975048,
+      "loss": 0.8574,
+      "step": 947
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032689881249368827,
+      "loss": 0.8228,
+      "step": 948
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000326737506836401,
+      "loss": 0.7638,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003265760633011281,
+      "loss": 0.8845,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000326414482063504,
+      "loss": 0.8748,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003262527632993134,
+      "loss": 0.9087,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003260909071844904,
+      "loss": 0.7983,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003259289138951187,
+      "loss": 0.881,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000325766783607431,
+      "loss": 0.8324,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003256045164978092,
+      "loss": 0.8184,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003254421127427841,
+      "loss": 0.8551,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003252795725190348,
+      "loss": 0.7543,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032511689600338945,
+      "loss": 0.8992,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000324954083372824,
+      "loss": 0.8429,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003247911348044626,
+      "loss": 0.7733,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003246280504755775,
+      "loss": 0.7624,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003244648305635883,
+      "loss": 0.7916,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032430147524606227,
+      "loss": 0.921,
+      "step": 964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032413798470071404,
+      "loss": 0.9054,
+      "step": 965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003239743591054053,
+      "loss": 0.7536,
+      "step": 966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003238105986381446,
+      "loss": 0.9032,
+      "step": 967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032364670347708737,
+      "loss": 0.7539,
+      "step": 968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003234826738005354,
+      "loss": 0.8976,
+      "step": 969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032331850978693696,
+      "loss": 0.8338,
+      "step": 970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003231542116148864,
+      "loss": 0.8754,
+      "step": 971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032298977946312396,
+      "loss": 0.8339,
+      "step": 972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032282521351053586,
+      "loss": 0.802,
+      "step": 973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003226605139361535,
+      "loss": 0.7888,
+      "step": 974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032249568091915413,
+      "loss": 0.9323,
+      "step": 975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032233071463885966,
+      "loss": 0.94,
+      "step": 976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003221656152747374,
+      "loss": 0.9197,
+      "step": 977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003220003830063993,
+      "loss": 0.8297,
+      "step": 978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032183501801360176,
+      "loss": 0.7927,
+      "step": 979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003216695204762457,
+      "loss": 0.9197,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032150389057437634,
+      "loss": 0.8055,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032133812848818266,
+      "loss": 0.7326,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032117223439799764,
+      "loss": 0.7013,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003210062084842977,
+      "loss": 0.8653,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003208400509277028,
+      "loss": 0.8677,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003206737619089761,
+      "loss": 0.9365,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032050734160902377,
+      "loss": 0.804,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032034079020889475,
+      "loss": 0.7305,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032017410788978066,
+      "loss": 0.859,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00032000729483301545,
+      "loss": 0.8098,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031984035122007554,
+      "loss": 0.7783,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000319673277232579,
+      "loss": 0.9755,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031950607305228617,
+      "loss": 0.7174,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031933873886109867,
+      "loss": 0.9318,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003191712748410598,
+      "loss": 0.8358,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031900368117435384,
+      "loss": 0.8101,
+      "step": 996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003188359580433065,
+      "loss": 0.8867,
+      "step": 997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031866810563038387,
+      "loss": 0.8272,
+      "step": 998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003185001241181931,
+      "loss": 0.7853,
+      "step": 999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003183320136894815,
+      "loss": 0.8275,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031816377452713687,
+      "loss": 0.7602,
+      "step": 1001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003179954068141868,
+      "loss": 0.8157,
+      "step": 1002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031782691073379887,
+      "loss": 0.8219,
+      "step": 1003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003176582864692804,
+      "loss": 0.8085,
+      "step": 1004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000317489534204078,
+      "loss": 0.7868,
+      "step": 1005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003173206541217776,
+      "loss": 0.8237,
+      "step": 1006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031715164640610415,
+      "loss": 0.8728,
+      "step": 1007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003169825112409215,
+      "loss": 0.9293,
+      "step": 1008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003168132488102321,
+      "loss": 0.9206,
+      "step": 1009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000316643859298177,
+      "loss": 0.8193,
+      "step": 1010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031647434288903533,
+      "loss": 0.8653,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031630469976722425,
+      "loss": 0.8366,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003161349301172989,
+      "loss": 0.8048,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031596503412395206,
+      "loss": 0.8373,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003157950119720139,
+      "loss": 0.9527,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003156248638464518,
+      "loss": 0.7577,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003154545899323703,
+      "loss": 0.8823,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031528419041501073,
+      "loss": 0.7875,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031511366547975106,
+      "loss": 0.8558,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003149430153121057,
+      "loss": 0.8018,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031477224009772534,
+      "loss": 0.8858,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003146013400223966,
+      "loss": 0.7805,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031443031527204207,
+      "loss": 0.8765,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031425916603271985,
+      "loss": 0.7516,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031408789249062367,
+      "loss": 0.7372,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031391649483208223,
+      "loss": 0.7388,
+      "step": 1026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031374497324355944,
+      "loss": 0.9391,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000313573327911654,
+      "loss": 0.7591,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003134015590230991,
+      "loss": 0.8702,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031322966676476254,
+      "loss": 0.873,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031305765132364627,
+      "loss": 0.8523,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003128855128868862,
+      "loss": 0.7795,
+      "step": 1032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003127132516417521,
+      "loss": 0.8504,
+      "step": 1033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003125408677756473,
+      "loss": 0.8674,
+      "step": 1034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031236836147610853,
+      "loss": 0.7847,
+      "step": 1035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031219573293080576,
+      "loss": 0.856,
+      "step": 1036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031202298232754186,
+      "loss": 0.8397,
+      "step": 1037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031185010985425267,
+      "loss": 0.7814,
+      "step": 1038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003116771156990065,
+      "loss": 1.054,
+      "step": 1039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031150400005000394,
+      "loss": 0.8827,
+      "step": 1040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031133076309557785,
+      "loss": 0.8494,
+      "step": 1041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031115740502419317,
+      "loss": 0.7716,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003109839260244464,
+      "loss": 0.7804,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031081032628506566,
+      "loss": 0.7986,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003106366059949106,
+      "loss": 0.7989,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003104627653429716,
+      "loss": 0.7971,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003102888045183706,
+      "loss": 0.799,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00031011472371035966,
+      "loss": 0.8702,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030994052310832177,
+      "loss": 0.8774,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030976620290177004,
+      "loss": 0.8574,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030959176328034785,
+      "loss": 0.7147,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003094172044338283,
+      "loss": 0.781,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030924252655211435,
+      "loss": 0.8605,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030906772982523844,
+      "loss": 0.7882,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030889281444336226,
+      "loss": 0.8603,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030871778059677664,
+      "loss": 0.7435,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030854262847590123,
+      "loss": 0.7544,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030836735827128437,
+      "loss": 0.7514,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003081919701736028,
+      "loss": 0.8427,
+      "step": 1059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030801646437366163,
+      "loss": 0.8223,
+      "step": 1060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030784084106239397,
+      "loss": 0.9055,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030766510043086083,
+      "loss": 0.9104,
+      "step": 1062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003074892426702507,
+      "loss": 0.81,
+      "step": 1063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030731326797187963,
+      "loss": 0.7791,
+      "step": 1064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030713717652719084,
+      "loss": 0.8033,
+      "step": 1065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003069609685277546,
+      "loss": 0.7803,
+      "step": 1066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030678464416526783,
+      "loss": 0.7818,
+      "step": 1067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003066082036315542,
+      "loss": 0.9618,
+      "step": 1068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003064316471185638,
+      "loss": 0.7529,
+      "step": 1069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030625497481837283,
+      "loss": 0.8316,
+      "step": 1070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003060781869231833,
+      "loss": 0.9487,
+      "step": 1071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003059012836253231,
+      "loss": 0.9375,
+      "step": 1072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030572426511724587,
+      "loss": 0.8702,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030554713159153024,
+      "loss": 0.7666,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003053698832408801,
+      "loss": 0.8935,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003051925202581245,
+      "loss": 0.8688,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003050150428362168,
+      "loss": 0.7945,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000304837451168235,
+      "loss": 0.9126,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003046597454473815,
+      "loss": 0.9033,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030448192586698275,
+      "loss": 0.8133,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000304303992620489,
+      "loss": 0.8234,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003041259459014741,
+      "loss": 0.7002,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030394778590363557,
+      "loss": 0.7676,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000303769512820794,
+      "loss": 0.8905,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003035911268468929,
+      "loss": 0.9124,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030341262817599877,
+      "loss": 0.7407,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030323401700230076,
+      "loss": 0.7126,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030305529352011033,
+      "loss": 0.8166,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000302876457923861,
+      "loss": 0.8768,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030269751040810846,
+      "loss": 0.7921,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030251845116753,
+      "loss": 0.8717,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003023392803969246,
+      "loss": 0.8155,
+      "step": 1092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030215999829121237,
+      "loss": 0.8716,
+      "step": 1093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030198060504543485,
+      "loss": 0.8935,
+      "step": 1094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003018011008547542,
+      "loss": 0.7692,
+      "step": 1095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030162148591445333,
+      "loss": 0.8295,
+      "step": 1096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003014417604199358,
+      "loss": 0.8026,
+      "step": 1097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003012619245667253,
+      "loss": 0.936,
+      "step": 1098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003010819785504656,
+      "loss": 0.8255,
+      "step": 1099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030090192256692033,
+      "loss": 0.8143,
+      "step": 1100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003007217568119728,
+      "loss": 0.8212,
+      "step": 1101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030054148148162565,
+      "loss": 0.971,
+      "step": 1102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030036109677200064,
+      "loss": 0.8691,
+      "step": 1103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030018060287933894,
+      "loss": 0.8678,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 0.8454,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029981928833046215,
+      "loss": 0.9318,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002996384680673219,
+      "loss": 0.8975,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029945753940729405,
+      "loss": 0.7644,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002992765025472112,
+      "loss": 0.7729,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002990953576840237,
+      "loss": 0.8538,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029891410501479943,
+      "loss": 0.8392,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029873274473672353,
+      "loss": 0.8529,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029855127704709824,
+      "loss": 0.9645,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029836970214334253,
+      "loss": 0.8126,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002981880202229922,
+      "loss": 0.784,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002980062314836993,
+      "loss": 0.7994,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029782433612323225,
+      "loss": 0.7294,
+      "step": 1117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002976423343394752,
+      "loss": 0.8602,
+      "step": 1118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002974602263304284,
+      "loss": 0.8044,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029727801229420753,
+      "loss": 0.8751,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970956924290435,
+      "loss": 0.9211,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002969132669332826,
+      "loss": 0.8308,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029673073600538575,
+      "loss": 0.8556,
+      "step": 1123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029654809984392875,
+      "loss": 0.8402,
+      "step": 1124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002963653586476018,
+      "loss": 0.7698,
+      "step": 1125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029618251261520954,
+      "loss": 0.9267,
+      "step": 1126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002959995619456704,
+      "loss": 0.7853,
+      "step": 1127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029581650683801683,
+      "loss": 0.8162,
+      "step": 1128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029563334749139493,
+      "loss": 0.8833,
+      "step": 1129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029545008410506395,
+      "loss": 0.8235,
+      "step": 1130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029526671687839657,
+      "loss": 0.7709,
+      "step": 1131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002950832460108784,
+      "loss": 0.8247,
+      "step": 1132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002948996717021077,
+      "loss": 0.9014,
+      "step": 1133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029471599415179547,
+      "loss": 0.9221,
+      "step": 1134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002945322135597646,
+      "loss": 0.7462,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002943483301259507,
+      "loss": 0.8751,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002941643440504006,
+      "loss": 0.7828,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029398025553327336,
+      "loss": 0.8799,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002937960647748391,
+      "loss": 0.8926,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002936117719754794,
+      "loss": 0.784,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029342737733568675,
+      "loss": 0.8444,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029324288105606446,
+      "loss": 0.8456,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002930582833373262,
+      "loss": 0.7699,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002928735843802965,
+      "loss": 0.8118,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029268878438590956,
+      "loss": 0.8885,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002925038835552097,
+      "loss": 0.828,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029231888208935085,
+      "loss": 0.7081,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029213378018959655,
+      "loss": 0.8814,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029194857805731945,
+      "loss": 0.7413,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029176327589400133,
+      "loss": 0.7792,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002915778739012329,
+      "loss": 0.8231,
+      "step": 1151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002913923722807132,
+      "loss": 0.8934,
+      "step": 1152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002912067712342499,
+      "loss": 0.8898,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002910210709637586,
+      "loss": 0.8349,
+      "step": 1154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029083527167126326,
+      "loss": 0.8843,
+      "step": 1155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029064937355889516,
+      "loss": 0.8321,
+      "step": 1156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029046337682889315,
+      "loss": 0.7978,
+      "step": 1157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002902772816836035,
+      "loss": 0.9815,
+      "step": 1158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029009108832547955,
+      "loss": 0.7475,
+      "step": 1159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002899047969570814,
+      "loss": 0.8197,
+      "step": 1160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002897184077810757,
+      "loss": 0.818,
+      "step": 1161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002895319210002357,
+      "loss": 0.7047,
+      "step": 1162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002893453368174408,
+      "loss": 0.8619,
+      "step": 1163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028915865543567614,
+      "loss": 0.7857,
+      "step": 1164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028897187705803287,
+      "loss": 0.8098,
+      "step": 1165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002887850018877076,
+      "loss": 0.7907,
+      "step": 1166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002885980301280021,
+      "loss": 0.8615,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028841096198232335,
+      "loss": 0.8535,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002882237976541832,
+      "loss": 0.7688,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000288036537347198,
+      "loss": 0.8443,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028784918126508873,
+      "loss": 0.8012,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002876617296116804,
+      "loss": 0.933,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000287474182590902,
+      "loss": 0.9006,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028728654040678636,
+      "loss": 0.9055,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002870988032634697,
+      "loss": 0.8341,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002869109713651917,
+      "loss": 0.8838,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002867230449162951,
+      "loss": 0.9542,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002865350241212254,
+      "loss": 0.79,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002863469091845307,
+      "loss": 0.9195,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002861587003108618,
+      "loss": 0.8383,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028597039770497147,
+      "loss": 0.7865,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002857820015717144,
+      "loss": 0.8209,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028559351211604705,
+      "loss": 0.7438,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028540492954302774,
+      "loss": 0.7784,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028521625405781556,
+      "loss": 0.7923,
+      "step": 1185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002850274858656711,
+      "loss": 0.893,
+      "step": 1186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002848386251719556,
+      "loss": 0.8473,
+      "step": 1187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000284649672182131,
+      "loss": 0.8231,
+      "step": 1188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002844606271017596,
+      "loss": 0.8538,
+      "step": 1189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000284271490136504,
+      "loss": 0.6931,
+      "step": 1190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002840822614921266,
+      "loss": 0.8764,
+      "step": 1191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002838929413744896,
+      "loss": 0.8028,
+      "step": 1192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028370352998955485,
+      "loss": 0.8448,
+      "step": 1193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002835140275433832,
+      "loss": 0.8188,
+      "step": 1194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000283324434242135,
+      "loss": 0.7779,
+      "step": 1195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028313475029206886,
+      "loss": 0.7734,
+      "step": 1196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002829449758995425,
+      "loss": 0.7393,
+      "step": 1197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002827551112710119,
+      "loss": 0.7134,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002825651566130311,
+      "loss": 0.8682,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002823751121322522,
+      "loss": 0.855,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028218497803542483,
+      "loss": 0.8512,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002819947545293964,
+      "loss": 0.8537,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002818044418211114,
+      "loss": 0.8571,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028161404011761135,
+      "loss": 0.787,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002814235496260348,
+      "loss": 0.7282,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002812329705536165,
+      "loss": 0.8109,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000281042303107688,
+      "loss": 0.7445,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028085154749567666,
+      "loss": 0.8194,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000280660703925106,
+      "loss": 0.802,
+      "step": 1209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000280469772603595,
+      "loss": 0.8686,
+      "step": 1210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002802787537388584,
+      "loss": 0.661,
+      "step": 1211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00028008764753870586,
+      "loss": 0.8107,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002798964542110421,
+      "loss": 0.8587,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000279705173963867,
+      "loss": 0.8315,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027951380700527457,
+      "loss": 0.7553,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002793223535434533,
+      "loss": 0.8377,
+      "step": 1216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002791308137866857,
+      "loss": 0.6913,
+      "step": 1217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027893918794334846,
+      "loss": 0.8209,
+      "step": 1218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002787474762219116,
+      "loss": 0.6961,
+      "step": 1219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002785556788309386,
+      "loss": 0.9444,
+      "step": 1220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002783637959790864,
+      "loss": 0.8084,
+      "step": 1221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027817182787510464,
+      "loss": 0.8558,
+      "step": 1222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027797977472783584,
+      "loss": 0.7882,
+      "step": 1223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027778763674621493,
+      "loss": 0.8436,
+      "step": 1224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027759541413926927,
+      "loss": 0.8484,
+      "step": 1225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002774031071161182,
+      "loss": 0.8702,
+      "step": 1226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027721071588597285,
+      "loss": 0.9079,
+      "step": 1227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002770182406581362,
+      "loss": 0.8536,
+      "step": 1228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002768256816420022,
+      "loss": 0.7895,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027663303904705637,
+      "loss": 0.8394,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002764403130828748,
+      "loss": 0.8754,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002762475039591247,
+      "loss": 0.7562,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027605461188556324,
+      "loss": 0.7641,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027586163707203826,
+      "loss": 0.7786,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002756685797284873,
+      "loss": 0.8149,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027547544006493795,
+      "loss": 0.7559,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002752822182915073,
+      "loss": 0.7657,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027508891461840146,
+      "loss": 0.8771,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002748955292559161,
+      "loss": 0.8152,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002747020624144355,
+      "loss": 0.9096,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002745085143044325,
+      "loss": 0.6346,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002743148851364685,
+      "loss": 0.7072,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002741211751211932,
+      "loss": 0.8562,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027392738446934404,
+      "loss": 0.8177,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002737335133917462,
+      "loss": 0.6827,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027353956209931246,
+      "loss": 0.795,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002733455308030429,
+      "loss": 0.8273,
+      "step": 1247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002731514197140243,
+      "loss": 0.8112,
+      "step": 1248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002729572290434307,
+      "loss": 0.8588,
+      "step": 1249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027276295900252254,
+      "loss": 0.8459,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002725686098026465,
+      "loss": 0.7849,
+      "step": 1251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002723741816552355,
+      "loss": 0.9183,
+      "step": 1252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027217967477180825,
+      "loss": 0.8296,
+      "step": 1253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002719850893639692,
+      "loss": 0.7532,
+      "step": 1254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027179042564340827,
+      "loss": 0.8733,
+      "step": 1255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027159568382190044,
+      "loss": 0.8899,
+      "step": 1256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027140086411130577,
+      "loss": 0.8526,
+      "step": 1257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000271205966723569,
+      "loss": 0.7626,
+      "step": 1258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027101099187071926,
+      "loss": 0.8488,
+      "step": 1259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002708159397648703,
+      "loss": 0.8439,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027062081061821953,
+      "loss": 0.6047,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002704256046430484,
+      "loss": 0.8215,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027023032205172187,
+      "loss": 0.7354,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00027003496305668827,
+      "loss": 0.7371,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000269839527870479,
+      "loss": 0.8954,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002696440167057084,
+      "loss": 0.8199,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026944842977507353,
+      "loss": 0.7683,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026925276729135376,
+      "loss": 0.7715,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002690570294674106,
+      "loss": 0.7595,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002688612165161877,
+      "loss": 0.7667,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002686653286507104,
+      "loss": 0.8561,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002684693660840855,
+      "loss": 0.7591,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026827332902950094,
+      "loss": 0.7741,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026807721770022593,
+      "loss": 0.8932,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002678810323096103,
+      "loss": 0.715,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026768477307108465,
+      "loss": 0.7344,
+      "step": 1276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002674884401981597,
+      "loss": 0.8801,
+      "step": 1277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026729203390442633,
+      "loss": 0.8469,
+      "step": 1278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026709555440355547,
+      "loss": 0.8776,
+      "step": 1279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026689900190929745,
+      "loss": 0.7765,
+      "step": 1280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002667023766354822,
+      "loss": 0.7857,
+      "step": 1281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002665056787960187,
+      "loss": 0.8666,
+      "step": 1282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026630890860489487,
+      "loss": 0.7811,
+      "step": 1283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026611206627617746,
+      "loss": 0.929,
+      "step": 1284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026591515202401163,
+      "loss": 0.7348,
+      "step": 1285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026571816606262087,
+      "loss": 0.8063,
+      "step": 1286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002655211086063064,
+      "loss": 0.8517,
+      "step": 1287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026532397986944755,
+      "loss": 0.8968,
+      "step": 1288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026512678006650105,
+      "loss": 0.826,
+      "step": 1289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000264929509412001,
+      "loss": 0.7708,
+      "step": 1290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002647321681205583,
+      "loss": 0.8591,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026453475640686127,
+      "loss": 0.6878,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002643372744856744,
+      "loss": 0.8772,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002641397225718385,
+      "loss": 0.8754,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002639421008802709,
+      "loss": 0.7222,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002637444096259645,
+      "loss": 0.8451,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026354664902398806,
+      "loss": 0.7148,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002633488192894857,
+      "loss": 0.8482,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026315092063767685,
+      "loss": 0.7856,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026295295328385577,
+      "loss": 0.791,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002627549174433916,
+      "loss": 0.8363,
+      "step": 1301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002625568133317278,
+      "loss": 0.7884,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026235864116438246,
+      "loss": 0.8141,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002621604011569473,
+      "loss": 0.8124,
+      "step": 1304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002619620935250879,
+      "loss": 0.8183,
+      "step": 1305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026176371848454387,
+      "loss": 0.7737,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026156527625112756,
+      "loss": 0.8225,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026136676704072474,
+      "loss": 0.8089,
+      "step": 1308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026116819106929395,
+      "loss": 0.7792,
+      "step": 1309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026096954855286643,
+      "loss": 0.7672,
+      "step": 1310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002607708397075457,
+      "loss": 0.8447,
+      "step": 1311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026057206474950755,
+      "loss": 0.818,
+      "step": 1312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002603732238949997,
+      "loss": 0.9028,
+      "step": 1313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026017431736034146,
+      "loss": 0.73,
+      "step": 1314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025997534536192365,
+      "loss": 0.7847,
+      "step": 1315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002597763081162083,
+      "loss": 0.7655,
+      "step": 1316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025957720583972844,
+      "loss": 0.7659,
+      "step": 1317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002593780387490878,
+      "loss": 0.7874,
+      "step": 1318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002591788070609606,
+      "loss": 0.6857,
+      "step": 1319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002589795109920916,
+      "loss": 0.8471,
+      "step": 1320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025878015075929514,
+      "loss": 0.7182,
+      "step": 1321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002585807265794558,
+      "loss": 0.868,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002583812386695274,
+      "loss": 0.7865,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025818168724653335,
+      "loss": 0.6483,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000257982072527566,
+      "loss": 0.7781,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002577823947297866,
+      "loss": 0.8011,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025758265407042495,
+      "loss": 0.8884,
+      "step": 1327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025738285076677946,
+      "loss": 0.72,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025718298503621645,
+      "loss": 0.8049,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002569830570961702,
+      "loss": 0.802,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002567830671641428,
+      "loss": 0.7597,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002565830154577037,
+      "loss": 0.7524,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002563829021944895,
+      "loss": 0.7396,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002561827275922038,
+      "loss": 0.8087,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000255982491868617,
+      "loss": 0.7759,
+      "step": 1335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002557821952415659,
+      "loss": 0.8478,
+      "step": 1336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002555818379289536,
+      "loss": 0.9058,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002553814201487493,
+      "loss": 0.8304,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025518094211898784,
+      "loss": 0.844,
+      "step": 1339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002549804040577697,
+      "loss": 0.7615,
+      "step": 1340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025477980618326046,
+      "loss": 0.7622,
+      "step": 1341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002545791487136912,
+      "loss": 0.9199,
+      "step": 1342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002543784318673575,
+      "loss": 0.7811,
+      "step": 1343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002541776558626195,
+      "loss": 0.8951,
+      "step": 1344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002539768209179019,
+      "loss": 0.817,
+      "step": 1345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025377592725169353,
+      "loss": 0.8549,
+      "step": 1346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002535749750825469,
+      "loss": 0.905,
+      "step": 1347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025337396462907815,
+      "loss": 0.8297,
+      "step": 1348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025317289610996723,
+      "loss": 0.8099,
+      "step": 1349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025297176974395683,
+      "loss": 0.8205,
+      "step": 1350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002527705857498527,
+      "loss": 0.6529,
+      "step": 1351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025256934434652337,
+      "loss": 0.7679,
+      "step": 1352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002523680457528998,
+      "loss": 0.8152,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025216669018797503,
+      "loss": 0.7571,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025196527787080416,
+      "loss": 0.8246,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025176380902050413,
+      "loss": 0.8021,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002515622838562534,
+      "loss": 0.9041,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002513607025972914,
+      "loss": 0.9309,
+      "step": 1358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025115906546291876,
+      "loss": 0.8437,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025095737267249714,
+      "loss": 0.8432,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025075562444544844,
+      "loss": 0.7801,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025055382100125487,
+      "loss": 0.75,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025035196255945885,
+      "loss": 0.8321,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002501500493396626,
+      "loss": 0.8357,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024994808156152793,
+      "loss": 0.7905,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002497460594447759,
+      "loss": 0.7766,
+      "step": 1366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002495439832091869,
+      "loss": 0.8259,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002493418530746,
+      "loss": 0.9454,
+      "step": 1368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024913966926091286,
+      "loss": 0.8783,
+      "step": 1369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002489374319880819,
+      "loss": 0.82,
+      "step": 1370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002487351414761213,
+      "loss": 0.7648,
+      "step": 1371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002485327979451033,
+      "loss": 0.7604,
+      "step": 1372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024833040161515785,
+      "loss": 0.9296,
+      "step": 1373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002481279527064723,
+      "loss": 0.8537,
+      "step": 1374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024792545143929135,
+      "loss": 0.873,
+      "step": 1375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024772289803391635,
+      "loss": 0.8363,
+      "step": 1376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002475202927107057,
+      "loss": 0.8233,
+      "step": 1377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002473176356900741,
+      "loss": 0.8654,
+      "step": 1378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002471149271924924,
+      "loss": 0.826,
+      "step": 1379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002469121674384877,
+      "loss": 0.8285,
+      "step": 1380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002467093566486427,
+      "loss": 0.8148,
+      "step": 1381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024650649504359576,
+      "loss": 0.9133,
+      "step": 1382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024630358284404025,
+      "loss": 0.7515,
+      "step": 1383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002461006202707249,
+      "loss": 0.7141,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024589760754445297,
+      "loss": 0.8947,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002456945448860825,
+      "loss": 0.812,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002454914325165258,
+      "loss": 0.7709,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002452882706567491,
+      "loss": 0.8503,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002450850595277727,
+      "loss": 0.8723,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002448817993506703,
+      "loss": 0.9074,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024467849034656914,
+      "loss": 0.7424,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024447513273664944,
+      "loss": 0.8135,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024427172674214433,
+      "loss": 0.801,
+      "step": 1393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002440682725843396,
+      "loss": 0.8348,
+      "step": 1394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024386477048457352,
+      "loss": 0.7546,
+      "step": 1395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002436612206642363,
+      "loss": 0.8318,
+      "step": 1396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002434576233447703,
+      "loss": 0.7767,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002432539787476694,
+      "loss": 0.7848,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024305028709447894,
+      "loss": 0.7776,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024284654860679556,
+      "loss": 0.8982,
+      "step": 1400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002426427635062666,
+      "loss": 0.7583,
+      "step": 1401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024243893201459046,
+      "loss": 0.7043,
+      "step": 1402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024223505435351576,
+      "loss": 0.8667,
+      "step": 1403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024203113074484139,
+      "loss": 0.7717,
+      "step": 1404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024182716141041627,
+      "loss": 0.7789,
+      "step": 1405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024162314657213903,
+      "loss": 0.7296,
+      "step": 1406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024141908645195783,
+      "loss": 0.8057,
+      "step": 1407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024121498127187003,
+      "loss": 0.6596,
+      "step": 1408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002410108312539222,
+      "loss": 0.835,
+      "step": 1409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002408066366202094,
+      "loss": 0.7188,
+      "step": 1410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024060239759287542,
+      "loss": 0.8117,
+      "step": 1411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024039811439411236,
+      "loss": 0.7803,
+      "step": 1412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024019378724616028,
+      "loss": 0.8046,
+      "step": 1413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023998941637130708,
+      "loss": 0.8879,
+      "step": 1414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023978500199188822,
+      "loss": 0.9018,
+      "step": 1415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023958054433028658,
+      "loss": 0.7964,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023937604360893202,
+      "loss": 0.728,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023917150005030123,
+      "loss": 0.7437,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023896691387691768,
+      "loss": 0.9238,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023876228531135102,
+      "loss": 0.814,
+      "step": 1420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023855761457621698,
+      "loss": 0.8089,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023835290189417734,
+      "loss": 0.8004,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023814814748793946,
+      "loss": 0.7277,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000237943351580256,
+      "loss": 0.8786,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023773851439392484,
+      "loss": 0.7797,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023753363615178886,
+      "loss": 0.7497,
+      "step": 1426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023732871707673545,
+      "loss": 0.8385,
+      "step": 1427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002371237573916964,
+      "loss": 0.7799,
+      "step": 1428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023691875731964785,
+      "loss": 0.862,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002367137170836098,
+      "loss": 0.8429,
+      "step": 1430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023650863690664585,
+      "loss": 0.8117,
+      "step": 1431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023630351701186316,
+      "loss": 0.8043,
+      "step": 1432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023609835762241214,
+      "loss": 0.9067,
+      "step": 1433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023589315896148602,
+      "loss": 0.8076,
+      "step": 1434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002356879212523209,
+      "loss": 0.8996,
+      "step": 1435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023548264471819522,
+      "loss": 0.7879,
+      "step": 1436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023527732958242972,
+      "loss": 0.7498,
+      "step": 1437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023507197606838722,
+      "loss": 0.8444,
+      "step": 1438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023486658439947217,
+      "loss": 0.9029,
+      "step": 1439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002346611547991306,
+      "loss": 0.7395,
+      "step": 1440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023445568749084976,
+      "loss": 0.8959,
+      "step": 1441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023425018269815798,
+      "loss": 0.8395,
+      "step": 1442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023404464064462425,
+      "loss": 0.8406,
+      "step": 1443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002338390615538583,
+      "loss": 0.8222,
+      "step": 1444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002336334456495099,
+      "loss": 0.9294,
+      "step": 1445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023342779315526914,
+      "loss": 0.766,
+      "step": 1446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002332221042948656,
+      "loss": 0.7716,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023301637929206876,
+      "loss": 0.7466,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023281061837068722,
+      "loss": 0.6366,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002326048217545686,
+      "loss": 0.8176,
+      "step": 1450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002323989896675996,
+      "loss": 0.941,
+      "step": 1451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023219312233370526,
+      "loss": 0.7878,
+      "step": 1452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002319872199768491,
+      "loss": 0.8154,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023178128282103267,
+      "loss": 0.8614,
+      "step": 1454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023157531109029546,
+      "loss": 0.7817,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002313693050087145,
+      "loss": 0.6855,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023116326480040415,
+      "loss": 0.7142,
+      "step": 1457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002309571906895161,
+      "loss": 0.901,
+      "step": 1458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023075108290023866,
+      "loss": 0.8425,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023054494165679698,
+      "loss": 0.861,
+      "step": 1460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002303387671834525,
+      "loss": 0.7317,
+      "step": 1461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002301325597045028,
+      "loss": 0.737,
+      "step": 1462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002299263194442815,
+      "loss": 0.7566,
+      "step": 1463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022972004662715772,
+      "loss": 0.778,
+      "step": 1464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022951374147753612,
+      "loss": 0.7502,
+      "step": 1465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022930740421985642,
+      "loss": 1.0207,
+      "step": 1466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022910103507859338,
+      "loss": 0.8894,
+      "step": 1467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022889463427825635,
+      "loss": 0.7747,
+      "step": 1468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022868820204338924,
+      "loss": 0.8235,
+      "step": 1469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022848173859857002,
+      "loss": 0.8582,
+      "step": 1470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002282752441684107,
+      "loss": 0.7006,
+      "step": 1471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002280687189775571,
+      "loss": 0.7917,
+      "step": 1472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022786216325068818,
+      "loss": 0.7856,
+      "step": 1473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002276555772125165,
+      "loss": 0.8872,
+      "step": 1474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022744896108778728,
+      "loss": 0.8301,
+      "step": 1475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002272423151012787,
+      "loss": 0.8535,
+      "step": 1476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022703563947780136,
+      "loss": 0.8565,
+      "step": 1477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000226828934442198,
+      "loss": 0.7949,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022662220021934352,
+      "loss": 0.8309,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022641543703414446,
+      "loss": 0.7413,
+      "step": 1480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022620864511153888,
+      "loss": 0.8349,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002260018246764961,
+      "loss": 0.8186,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002257949759540166,
+      "loss": 0.9376,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022558809916913144,
+      "loss": 0.7483,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022538119454690222,
+      "loss": 0.7852,
+      "step": 1485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022517426231242102,
+      "loss": 0.6877,
+      "step": 1486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022496730269080977,
+      "loss": 0.8403,
+      "step": 1487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002247603159072202,
+      "loss": 0.8573,
+      "step": 1488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002245533021868337,
+      "loss": 0.7986,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022434626175486092,
+      "loss": 0.8699,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022413919483654158,
+      "loss": 0.7706,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022393210165714406,
+      "loss": 0.7099,
+      "step": 1492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022372498244196566,
+      "loss": 0.7017,
+      "step": 1493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002235178374163316,
+      "loss": 0.7534,
+      "step": 1494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022331066680559541,
+      "loss": 0.7933,
+      "step": 1495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002231034708351384,
+      "loss": 0.7733,
+      "step": 1496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022289624973036954,
+      "loss": 0.8583,
+      "step": 1497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022268900371672507,
+      "loss": 0.8467,
+      "step": 1498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022248173301966828,
+      "loss": 0.818,
+      "step": 1499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022227443786468945,
+      "loss": 0.7388,
+      "step": 1500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022206711847730533,
+      "loss": 0.8622,
+      "step": 1501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002218597750830591,
+      "loss": 0.8335,
+      "step": 1502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002216524079075201,
+      "loss": 0.7232,
+      "step": 1503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022144501717628344,
+      "loss": 0.8736,
+      "step": 1504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022123760311496994,
+      "loss": 0.7952,
+      "step": 1505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022103016594922573,
+      "loss": 0.8455,
+      "step": 1506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022082270590472218,
+      "loss": 0.7298,
+      "step": 1507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022061522320715544,
+      "loss": 0.6802,
+      "step": 1508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022040771808224635,
+      "loss": 0.7038,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00022020019075574018,
+      "loss": 0.8333,
+      "step": 1510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021999264145340628,
+      "loss": 0.7265,
+      "step": 1511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021978507040103798,
+      "loss": 0.7554,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002195774778244522,
+      "loss": 0.843,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021936986394948936,
+      "loss": 0.9085,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000219162229002013,
+      "loss": 0.7808,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002189545732079096,
+      "loss": 0.7746,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002187468967930883,
+      "loss": 0.9266,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021853919998348064,
+      "loss": 0.8618,
+      "step": 1518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021833148300504047,
+      "loss": 0.7558,
+      "step": 1519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021812374608374348,
+      "loss": 0.6906,
+      "step": 1520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021791598944558702,
+      "loss": 0.7529,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021770821331659006,
+      "loss": 0.8236,
+      "step": 1522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002175004179227926,
+      "loss": 0.7675,
+      "step": 1523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021729260349025563,
+      "loss": 0.8469,
+      "step": 1524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002170847702450609,
+      "loss": 0.729,
+      "step": 1525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021687691841331067,
+      "loss": 0.7497,
+      "step": 1526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021666904822112732,
+      "loss": 0.6322,
+      "step": 1527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021646115989465322,
+      "loss": 0.8131,
+      "step": 1528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021625325366005056,
+      "loss": 0.7856,
+      "step": 1529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021604532974350086,
+      "loss": 0.9164,
+      "step": 1530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002158373883712051,
+      "loss": 0.835,
+      "step": 1531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021562942976938293,
+      "loss": 0.8374,
+      "step": 1532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021542145416427304,
+      "loss": 0.7724,
+      "step": 1533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002152134617821325,
+      "loss": 0.8587,
+      "step": 1534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002150054528492366,
+      "loss": 0.8003,
+      "step": 1535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002147974275918787,
+      "loss": 0.7291,
+      "step": 1536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021458938623636993,
+      "loss": 0.7748,
+      "step": 1537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021438132900903885,
+      "loss": 0.8508,
+      "step": 1538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021417325613623124,
+      "loss": 0.8028,
+      "step": 1539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021396516784431016,
+      "loss": 0.6825,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021375706435965519,
+      "loss": 0.6851,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021354894590866247,
+      "loss": 0.8531,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021334081271774462,
+      "loss": 0.7176,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021313266501333,
+      "loss": 0.7469,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002129245030218631,
+      "loss": 0.7978,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002127163269698035,
+      "loss": 0.7695,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002125081370836266,
+      "loss": 0.8152,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021229993358982244,
+      "loss": 0.8913,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021209171671489604,
+      "loss": 0.8515,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021188348668536703,
+      "loss": 0.8801,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021167524372776918,
+      "loss": 0.8311,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002114669880686505,
+      "loss": 0.7091,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021125871993457257,
+      "loss": 0.8446,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002110504395521109,
+      "loss": 0.7112,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021084214714785405,
+      "loss": 0.7434,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021063384294840373,
+      "loss": 0.8711,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021042552718037456,
+      "loss": 0.7273,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002102172000703936,
+      "loss": 0.7523,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021000886184510033,
+      "loss": 0.7453,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002098005127311464,
+      "loss": 0.7736,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002095921529551951,
+      "loss": 0.8002,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002093837827439216,
+      "loss": 0.801,
+      "step": 1562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020917540232401208,
+      "loss": 0.8126,
+      "step": 1563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002089670119221642,
+      "loss": 0.7585,
+      "step": 1564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002087586117650862,
+      "loss": 0.658,
+      "step": 1565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020855020207949707,
+      "loss": 0.7763,
+      "step": 1566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020834178309212596,
+      "loss": 0.742,
+      "step": 1567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020813335502971257,
+      "loss": 0.6853,
+      "step": 1568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000207924918119006,
+      "loss": 0.8136,
+      "step": 1569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020771647258676525,
+      "loss": 0.8252,
+      "step": 1570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020750801865975863,
+      "loss": 0.7103,
+      "step": 1571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020729955656476352,
+      "loss": 0.8445,
+      "step": 1572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020709108652856636,
+      "loss": 0.8702,
+      "step": 1573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020688260877796202,
+      "loss": 0.8597,
+      "step": 1574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020667412353975393,
+      "loss": 0.7747,
+      "step": 1575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020646563104075356,
+      "loss": 0.8153,
+      "step": 1576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020625713150778033,
+      "loss": 0.7816,
+      "step": 1577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020604862516766123,
+      "loss": 0.6706,
+      "step": 1578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020584011224723082,
+      "loss": 0.6397,
+      "step": 1579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020563159297333066,
+      "loss": 0.8451,
+      "step": 1580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020542306757280926,
+      "loss": 0.824,
+      "step": 1581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020521453627252197,
+      "loss": 0.8315,
+      "step": 1582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002050059992993302,
+      "loss": 0.834,
+      "step": 1583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002047974568801019,
+      "loss": 0.8242,
+      "step": 1584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020458890924171064,
+      "loss": 0.7767,
+      "step": 1585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002043803566110359,
+      "loss": 0.8137,
+      "step": 1586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020417179921496256,
+      "loss": 0.7905,
+      "step": 1587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002039632372803805,
+      "loss": 0.8561,
+      "step": 1588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002037546710341847,
+      "loss": 0.835,
+      "step": 1589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020354610070327488,
+      "loss": 0.7648,
+      "step": 1590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020333752651455499,
+      "loss": 0.8348,
+      "step": 1591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020312894869493336,
+      "loss": 0.8747,
+      "step": 1592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020292036747132227,
+      "loss": 0.8079,
+      "step": 1593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020271178307063758,
+      "loss": 0.823,
+      "step": 1594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020250319571979867,
+      "loss": 0.8166,
+      "step": 1595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020229460564572817,
+      "loss": 0.8083,
+      "step": 1596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020208601307535165,
+      "loss": 0.7555,
+      "step": 1597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020187741823559732,
+      "loss": 0.8243,
+      "step": 1598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020166882135339592,
+      "loss": 0.7161,
+      "step": 1599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002014602226556805,
+      "loss": 0.7037,
+      "step": 1600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020125162236938598,
+      "loss": 0.8165,
+      "step": 1601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020104302072144892,
+      "loss": 0.7295,
+      "step": 1602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020083441793880758,
+      "loss": 0.8318,
+      "step": 1603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002006258142484013,
+      "loss": 0.8506,
+      "step": 1604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020041720987717048,
+      "loss": 0.7114,
+      "step": 1605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020020860505205615,
+      "loss": 0.765,
+      "step": 1606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002,
+      "loss": 0.8039,
+      "step": 1607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019979139494794387,
+      "loss": 0.7185,
+      "step": 1608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019958279012282954,
+      "loss": 0.7756,
+      "step": 1609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019937418575159872,
+      "loss": 0.8273,
+      "step": 1610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019916558206119247,
+      "loss": 0.6699,
+      "step": 1611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019895697927855112,
+      "loss": 0.9322,
+      "step": 1612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019874837763061406,
+      "loss": 0.7485,
+      "step": 1613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019853977734431952,
+      "loss": 0.7863,
+      "step": 1614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001983311786466041,
+      "loss": 0.954,
+      "step": 1615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019812258176440272,
+      "loss": 0.7622,
+      "step": 1616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019791398692464842,
+      "loss": 0.8082,
+      "step": 1617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001977053943542719,
+      "loss": 0.8896,
+      "step": 1618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019749680428020138,
+      "loss": 0.9032,
+      "step": 1619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019728821692936247,
+      "loss": 0.8202,
+      "step": 1620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019707963252867778,
+      "loss": 0.7785,
+      "step": 1621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019687105130506666,
+      "loss": 0.8617,
+      "step": 1622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019666247348544503,
+      "loss": 0.7849,
+      "step": 1623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019645389929672517,
+      "loss": 0.8126,
+      "step": 1624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019624532896581534,
+      "loss": 0.7492,
+      "step": 1625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019603676271961958,
+      "loss": 0.7439,
+      "step": 1626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019582820078503746,
+      "loss": 0.7455,
+      "step": 1627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001956196433889641,
+      "loss": 0.8445,
+      "step": 1628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001954110907582894,
+      "loss": 0.7642,
+      "step": 1629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019520254311989813,
+      "loss": 0.8229,
+      "step": 1630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001949940007006698,
+      "loss": 0.8314,
+      "step": 1631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001947854637274781,
+      "loss": 0.731,
+      "step": 1632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019457693242719076,
+      "loss": 0.7503,
+      "step": 1633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019436840702666936,
+      "loss": 0.8317,
+      "step": 1634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019415988775276923,
+      "loss": 0.6619,
+      "step": 1635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019395137483233882,
+      "loss": 0.7907,
+      "step": 1636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001937428684922197,
+      "loss": 0.6837,
+      "step": 1637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001935343689592465,
+      "loss": 0.7548,
+      "step": 1638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019332587646024612,
+      "loss": 0.8726,
+      "step": 1639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019311739122203803,
+      "loss": 0.9105,
+      "step": 1640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019290891347143368,
+      "loss": 0.8083,
+      "step": 1641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001927004434352365,
+      "loss": 0.8153,
+      "step": 1642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019249198134024145,
+      "loss": 0.8885,
+      "step": 1643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001922835274132348,
+      "loss": 0.7479,
+      "step": 1644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019207508188099403,
+      "loss": 0.8869,
+      "step": 1645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019186664497028747,
+      "loss": 0.8251,
+      "step": 1646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019165821690787405,
+      "loss": 0.7607,
+      "step": 1647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000191449797920503,
+      "loss": 0.7468,
+      "step": 1648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019124138823491382,
+      "loss": 0.7786,
+      "step": 1649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019103298807783583,
+      "loss": 0.8229,
+      "step": 1650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019082459767598794,
+      "loss": 0.7271,
+      "step": 1651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019061621725607845,
+      "loss": 0.8413,
+      "step": 1652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019040784704480492,
+      "loss": 0.8139,
+      "step": 1653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019019948726885368,
+      "loss": 0.6997,
+      "step": 1654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001899911381548997,
+      "loss": 0.8176,
+      "step": 1655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018978279992960646,
+      "loss": 0.7341,
+      "step": 1656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001895744728196255,
+      "loss": 0.7737,
+      "step": 1657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001893661570515963,
+      "loss": 0.8599,
+      "step": 1658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018915785285214597,
+      "loss": 0.7832,
+      "step": 1659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018894956044788913,
+      "loss": 0.8195,
+      "step": 1660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018874128006542747,
+      "loss": 0.7961,
+      "step": 1661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018853301193134958,
+      "loss": 0.8451,
+      "step": 1662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018832475627223084,
+      "loss": 0.802,
+      "step": 1663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000188116513314633,
+      "loss": 0.9099,
+      "step": 1664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018790828328510398,
+      "loss": 0.7606,
+      "step": 1665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018770006641017758,
+      "loss": 0.7824,
+      "step": 1666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018749186291637345,
+      "loss": 0.7934,
+      "step": 1667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018728367303019652,
+      "loss": 0.7466,
+      "step": 1668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018707549697813698,
+      "loss": 0.7388,
+      "step": 1669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018686733498667,
+      "loss": 0.7934,
+      "step": 1670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018665918728225545,
+      "loss": 0.7256,
+      "step": 1671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018645105409133755,
+      "loss": 0.6988,
+      "step": 1672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018624293564034486,
+      "loss": 0.7414,
+      "step": 1673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018603483215568988,
+      "loss": 0.8365,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001858267438637688,
+      "loss": 0.7443,
+      "step": 1675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018561867099096122,
+      "loss": 0.7268,
+      "step": 1676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018541061376363012,
+      "loss": 0.7836,
+      "step": 1677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018520257240812132,
+      "loss": 0.7445,
+      "step": 1678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018499454715076343,
+      "loss": 0.846,
+      "step": 1679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018478653821786753,
+      "loss": 0.6905,
+      "step": 1680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000184578545835727,
+      "loss": 0.806,
+      "step": 1681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018437057023061712,
+      "loss": 0.6979,
+      "step": 1682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018416261162879495,
+      "loss": 0.8408,
+      "step": 1683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018395467025649915,
+      "loss": 0.8509,
+      "step": 1684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018374674633994952,
+      "loss": 0.7777,
+      "step": 1685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018353884010534682,
+      "loss": 0.7476,
+      "step": 1686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018333095177887272,
+      "loss": 0.832,
+      "step": 1687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018312308158668938,
+      "loss": 0.8384,
+      "step": 1688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018291522975493916,
+      "loss": 0.8898,
+      "step": 1689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018270739650974441,
+      "loss": 0.8352,
+      "step": 1690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018249958207720748,
+      "loss": 0.9178,
+      "step": 1691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018229178668341002,
+      "loss": 0.7701,
+      "step": 1692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018208401055441303,
+      "loss": 0.7638,
+      "step": 1693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018187625391625657,
+      "loss": 0.8145,
+      "step": 1694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018166851699495957,
+      "loss": 0.8134,
+      "step": 1695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018146080001651938,
+      "loss": 0.7895,
+      "step": 1696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018125310320691175,
+      "loss": 0.65,
+      "step": 1697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018104542679209045,
+      "loss": 0.8936,
+      "step": 1698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018083777099798704,
+      "loss": 0.8745,
+      "step": 1699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018063013605051068,
+      "loss": 0.7993,
+      "step": 1700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018042252217554783,
+      "loss": 0.8205,
+      "step": 1701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001802149295989621,
+      "loss": 0.7547,
+      "step": 1702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001800073585465938,
+      "loss": 0.7228,
+      "step": 1703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017979980924425986,
+      "loss": 0.6904,
+      "step": 1704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001795922819177537,
+      "loss": 0.7916,
+      "step": 1705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017938477679284464,
+      "loss": 0.7323,
+      "step": 1706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001791772940952779,
+      "loss": 0.7904,
+      "step": 1707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017896983405077431,
+      "loss": 0.8089,
+      "step": 1708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017876239688503014,
+      "loss": 0.6921,
+      "step": 1709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017855498282371666,
+      "loss": 0.7507,
+      "step": 1710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017834759209247998,
+      "loss": 0.8818,
+      "step": 1711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017814022491694095,
+      "loss": 0.776,
+      "step": 1712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017793288152269477,
+      "loss": 0.7692,
+      "step": 1713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017772556213531065,
+      "loss": 0.7386,
+      "step": 1714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001775182669803318,
+      "loss": 0.7954,
+      "step": 1715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000177310996283275,
+      "loss": 0.7893,
+      "step": 1716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017710375026963054,
+      "loss": 0.8639,
+      "step": 1717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017689652916486162,
+      "loss": 0.6907,
+      "step": 1718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017668933319440469,
+      "loss": 0.8047,
+      "step": 1719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017648216258366853,
+      "loss": 0.7488,
+      "step": 1720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017627501755803447,
+      "loss": 0.8056,
+      "step": 1721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017606789834285596,
+      "loss": 0.8852,
+      "step": 1722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017586080516345852,
+      "loss": 0.7733,
+      "step": 1723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017565373824513915,
+      "loss": 0.7706,
+      "step": 1724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017544669781316633,
+      "loss": 0.772,
+      "step": 1725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017523968409277987,
+      "loss": 0.8197,
+      "step": 1726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017503269730919033,
+      "loss": 0.6471,
+      "step": 1727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017482573768757908,
+      "loss": 0.8577,
+      "step": 1728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001746188054530978,
+      "loss": 0.7513,
+      "step": 1729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017441190083086863,
+      "loss": 0.7461,
+      "step": 1730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017420502404598347,
+      "loss": 0.7179,
+      "step": 1731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017399817532350392,
+      "loss": 0.7415,
+      "step": 1732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001737913548884612,
+      "loss": 0.6967,
+      "step": 1733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017358456296585564,
+      "loss": 0.7643,
+      "step": 1734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017337779978065658,
+      "loss": 0.8377,
+      "step": 1735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017317106555780204,
+      "loss": 0.7948,
+      "step": 1736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017296436052219874,
+      "loss": 0.8747,
+      "step": 1737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017275768489872137,
+      "loss": 0.6976,
+      "step": 1738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017255103891221282,
+      "loss": 0.7814,
+      "step": 1739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017234442278748363,
+      "loss": 0.8099,
+      "step": 1740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001721378367493119,
+      "loss": 0.8866,
+      "step": 1741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017193128102244303,
+      "loss": 0.7333,
+      "step": 1742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017172475583158932,
+      "loss": 0.7419,
+      "step": 1743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017151826140143002,
+      "loss": 0.7874,
+      "step": 1744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017131179795661086,
+      "loss": 0.8612,
+      "step": 1745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017110536572174375,
+      "loss": 0.7411,
+      "step": 1746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001708989649214067,
+      "loss": 0.6847,
+      "step": 1747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017069259578014366,
+      "loss": 0.8613,
+      "step": 1748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017048625852246398,
+      "loss": 0.8613,
+      "step": 1749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017027995337284233,
+      "loss": 0.7202,
+      "step": 1750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017007368055571857,
+      "loss": 0.7933,
+      "step": 1751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016986744029549725,
+      "loss": 0.8188,
+      "step": 1752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016966123281654763,
+      "loss": 0.7905,
+      "step": 1753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001694550583432031,
+      "loss": 0.7936,
+      "step": 1754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016924891709976144,
+      "loss": 0.6362,
+      "step": 1755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016904280931048402,
+      "loss": 0.7489,
+      "step": 1756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001688367351995959,
+      "loss": 0.7361,
+      "step": 1757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00016863069499128562,
+      "loss": 0.7455,
+      "step": 1758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016842468890970464,
+      "loss": 0.7597,
+      "step": 1759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016821871717896743,
+      "loss": 0.7759,
+      "step": 1760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016801278002315099,
+      "loss": 0.8304,
+      "step": 1761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001678068776662948,
+      "loss": 0.8309,
+      "step": 1762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016760101033240049,
+      "loss": 0.7846,
+      "step": 1763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016739517824543144,
+      "loss": 0.9083,
+      "step": 1764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001671893816293129,
+      "loss": 0.9347,
+      "step": 1765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016698362070793134,
+      "loss": 0.822,
+      "step": 1766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016677789570513453,
+      "loss": 0.8056,
+      "step": 1767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000166572206844731,
+      "loss": 0.8217,
+      "step": 1768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016636655435049017,
+      "loss": 0.6312,
+      "step": 1769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016616093844614184,
+      "loss": 0.783,
+      "step": 1770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016595535935537577,
+      "loss": 0.8428,
+      "step": 1771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001657498173018421,
+      "loss": 0.8217,
+      "step": 1772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001655443125091503,
+      "loss": 0.7633,
+      "step": 1773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016533884520086945,
+      "loss": 0.7935,
+      "step": 1774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016513341560052788,
+      "loss": 0.768,
+      "step": 1775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016492802393161285,
+      "loss": 0.8008,
+      "step": 1776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016472267041757035,
+      "loss": 0.8355,
+      "step": 1777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016451735528180485,
+      "loss": 0.6755,
+      "step": 1778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016431207874767918,
+      "loss": 0.7598,
+      "step": 1779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016410684103851402,
+      "loss": 0.7293,
+      "step": 1780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016390164237758793,
+      "loss": 0.7001,
+      "step": 1781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016369648298813686,
+      "loss": 0.8488,
+      "step": 1782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016349136309335422,
+      "loss": 0.7636,
+      "step": 1783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001632862829163903,
+      "loss": 0.8001,
+      "step": 1784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001630812426803522,
+      "loss": 0.6985,
+      "step": 1785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016287624260830365,
+      "loss": 0.7736,
+      "step": 1786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016267128292326465,
+      "loss": 0.7735,
+      "step": 1787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00016246636384821122,
+      "loss": 0.742,
+      "step": 1788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001622614856060752,
+      "loss": 0.7824,
+      "step": 1789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016205664841974408,
+      "loss": 0.8066,
+      "step": 1790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016185185251206067,
+      "loss": 0.7574,
+      "step": 1791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016164709810582273,
+      "loss": 0.7511,
+      "step": 1792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016144238542378312,
+      "loss": 0.8003,
+      "step": 1793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001612377146886491,
+      "loss": 0.8905,
+      "step": 1794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001610330861230824,
+      "loss": 0.8297,
+      "step": 1795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016082849994969876,
+      "loss": 0.6713,
+      "step": 1796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016062395639106794,
+      "loss": 0.7632,
+      "step": 1797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001604194556697134,
+      "loss": 0.7787,
+      "step": 1798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016021499800811177,
+      "loss": 0.669,
+      "step": 1799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00016001058362869296,
+      "loss": 0.6587,
+      "step": 1800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015980621275383974,
+      "loss": 0.8034,
+      "step": 1801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015960188560588763,
+      "loss": 0.6769,
+      "step": 1802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015939760240712457,
+      "loss": 0.7338,
+      "step": 1803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001591933633797906,
+      "loss": 0.8175,
+      "step": 1804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015898916874607783,
+      "loss": 0.711,
+      "step": 1805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015878501872812996,
+      "loss": 0.8065,
+      "step": 1806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015858091354804222,
+      "loss": 0.8306,
+      "step": 1807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015837685342786096,
+      "loss": 0.7091,
+      "step": 1808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015817283858958375,
+      "loss": 0.6772,
+      "step": 1809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015796886925515863,
+      "loss": 0.6708,
+      "step": 1810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015776494564648423,
+      "loss": 0.7695,
+      "step": 1811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001575610679854095,
+      "loss": 0.7892,
+      "step": 1812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015735723649373337,
+      "loss": 0.8739,
+      "step": 1813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001571534513932045,
+      "loss": 0.7084,
+      "step": 1814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015694971290552105,
+      "loss": 0.8328,
+      "step": 1815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015674602125233063,
+      "loss": 0.8048,
+      "step": 1816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00015654237665522973,
+      "loss": 0.7228,
+      "step": 1817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001563387793357637,
+      "loss": 0.7546,
+      "step": 1818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001561352295154265,
+      "loss": 0.8437,
+      "step": 1819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001559317274156604,
+      "loss": 0.7485,
+      "step": 1820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015572827325785568,
+      "loss": 0.7562,
+      "step": 1821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015552486726335055,
+      "loss": 0.773,
+      "step": 1822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015532150965343085,
+      "loss": 0.9349,
+      "step": 1823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015511820064932969,
+      "loss": 0.7626,
+      "step": 1824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015491494047222733,
+      "loss": 0.6344,
+      "step": 1825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001547117293432509,
+      "loss": 0.669,
+      "step": 1826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015450856748347423,
+      "loss": 0.8472,
+      "step": 1827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001543054551139175,
+      "loss": 0.8574,
+      "step": 1828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015410239245554702,
+      "loss": 0.6973,
+      "step": 1829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015389937972927516,
+      "loss": 0.8998,
+      "step": 1830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001536964171559598,
+      "loss": 0.8161,
+      "step": 1831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015349350495640432,
+      "loss": 0.6984,
+      "step": 1832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015329064335135729,
+      "loss": 0.8321,
+      "step": 1833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015308783256151232,
+      "loss": 0.8287,
+      "step": 1834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015288507280750765,
+      "loss": 0.6726,
+      "step": 1835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015268236430992596,
+      "loss": 0.7857,
+      "step": 1836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015247970728929431,
+      "loss": 0.7587,
+      "step": 1837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001522771019660836,
+      "loss": 0.6711,
+      "step": 1838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001520745485607087,
+      "loss": 0.6842,
+      "step": 1839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015187204729352765,
+      "loss": 0.7693,
+      "step": 1840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015166959838484217,
+      "loss": 0.751,
+      "step": 1841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015146720205489676,
+      "loss": 0.8136,
+      "step": 1842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015126485852387872,
+      "loss": 0.7524,
+      "step": 1843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001510625680119181,
+      "loss": 0.8021,
+      "step": 1844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001508603307390871,
+      "loss": 0.8402,
+      "step": 1845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001506581469254001,
+      "loss": 0.774,
+      "step": 1846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001504560167908131,
+      "loss": 0.6819,
+      "step": 1847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015025394055522408,
+      "loss": 0.7427,
+      "step": 1848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00015005191843847212,
+      "loss": 0.7865,
+      "step": 1849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001498499506603374,
+      "loss": 0.7777,
+      "step": 1850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00014964803744054117,
+      "loss": 0.7994,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014944617899874518,
+      "loss": 0.8142,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001492443755545516,
+      "loss": 0.7358,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014904262732750285,
+      "loss": 0.7551,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001488409345370812,
+      "loss": 0.8057,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014863929740270869,
+      "loss": 0.71,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014843771614374664,
+      "loss": 0.8115,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014823619097949583,
+      "loss": 0.7707,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014803472212919584,
+      "loss": 0.8985,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014783330981202504,
+      "loss": 0.8966,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001476319542471002,
+      "loss": 0.8662,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014743065565347662,
+      "loss": 0.7066,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014722941425014728,
+      "loss": 0.7135,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014702823025604316,
+      "loss": 0.6693,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014682710389003276,
+      "loss": 0.7063,
+      "step": 1865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001466260353709218,
+      "loss": 0.7424,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014642502491745321,
+      "loss": 0.7763,
+      "step": 1867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014622407274830648,
+      "loss": 0.778,
+      "step": 1868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001460231790820981,
+      "loss": 0.6681,
+      "step": 1869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014582234413738055,
+      "loss": 0.8356,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014562156813264254,
+      "loss": 0.7462,
+      "step": 1871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001454208512863088,
+      "loss": 0.7712,
+      "step": 1872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014522019381673954,
+      "loss": 0.7813,
+      "step": 1873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014501959594223038,
+      "loss": 0.8347,
+      "step": 1874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014481905788101215,
+      "loss": 0.8195,
+      "step": 1875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014461857985125068,
+      "loss": 0.7651,
+      "step": 1876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014441816207104636,
+      "loss": 0.7593,
+      "step": 1877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014421780475843408,
+      "loss": 0.9151,
+      "step": 1878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000144017508131383,
+      "loss": 0.9212,
+      "step": 1879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014381727240779619,
+      "loss": 0.7678,
+      "step": 1880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00014361709780551053,
+      "loss": 0.7607,
+      "step": 1881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001434169845422963,
+      "loss": 0.756,
+      "step": 1882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014321693283585716,
+      "loss": 0.6924,
+      "step": 1883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001430169429038298,
+      "loss": 0.8002,
+      "step": 1884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014281701496378355,
+      "loss": 0.7169,
+      "step": 1885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014261714923322053,
+      "loss": 0.8302,
+      "step": 1886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000142417345929575,
+      "loss": 0.7167,
+      "step": 1887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014221760527021344,
+      "loss": 0.788,
+      "step": 1888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014201792747243398,
+      "loss": 0.6728,
+      "step": 1889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014181831275346664,
+      "loss": 0.7272,
+      "step": 1890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001416187613304726,
+      "loss": 0.7204,
+      "step": 1891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014141927342054423,
+      "loss": 0.873,
+      "step": 1892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014121984924070485,
+      "loss": 0.8394,
+      "step": 1893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014102048900790847,
+      "loss": 0.7785,
+      "step": 1894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001408211929390394,
+      "loss": 0.6916,
+      "step": 1895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014062196125091226,
+      "loss": 0.7842,
+      "step": 1896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014042279416027163,
+      "loss": 0.7883,
+      "step": 1897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014022369188379177,
+      "loss": 0.8306,
+      "step": 1898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00014002465463807637,
+      "loss": 0.747,
+      "step": 1899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013982568263965853,
+      "loss": 0.8099,
+      "step": 1900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001396267761050003,
+      "loss": 0.8538,
+      "step": 1901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013942793525049244,
+      "loss": 0.7769,
+      "step": 1902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001392291602924543,
+      "loss": 0.8236,
+      "step": 1903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013903045144713364,
+      "loss": 0.7585,
+      "step": 1904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001388318089307061,
+      "loss": 0.8265,
+      "step": 1905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013863323295927534,
+      "loss": 0.8194,
+      "step": 1906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013843472374887246,
+      "loss": 0.7221,
+      "step": 1907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013823628151545618,
+      "loss": 0.7478,
+      "step": 1908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001380379064749121,
+      "loss": 0.8059,
+      "step": 1909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001378395988430528,
+      "loss": 0.7191,
+      "step": 1910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001376413588356176,
+      "loss": 0.8223,
+      "step": 1911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001374431866682722,
+      "loss": 0.7848,
+      "step": 1912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001372450825566085,
+      "loss": 0.6529,
+      "step": 1913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00013704704671614425,
+      "loss": 0.6519,
+      "step": 1914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001368490793623232,
+      "loss": 0.7064,
+      "step": 1915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013665118071051432,
+      "loss": 0.7093,
+      "step": 1916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013645335097601198,
+      "loss": 0.8445,
+      "step": 1917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013625559037403555,
+      "loss": 0.7643,
+      "step": 1918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013605789911972917,
+      "loss": 0.719,
+      "step": 1919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013586027742816152,
+      "loss": 0.6573,
+      "step": 1920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013566272551432566,
+      "loss": 0.8327,
+      "step": 1921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013546524359313872,
+      "loss": 0.7526,
+      "step": 1922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013526783187944167,
+      "loss": 0.8355,
+      "step": 1923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001350704905879991,
+      "loss": 0.7807,
+      "step": 1924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000134873219933499,
+      "loss": 0.7645,
+      "step": 1925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001346760201305525,
+      "loss": 0.6884,
+      "step": 1926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013447889139369366,
+      "loss": 0.8718,
+      "step": 1927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001342818339373792,
+      "loss": 0.8266,
+      "step": 1928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013408484797598836,
+      "loss": 0.8219,
+      "step": 1929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013388793372382259,
+      "loss": 0.7263,
+      "step": 1930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013369109139510515,
+      "loss": 0.7553,
+      "step": 1931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001334943212039814,
+      "loss": 0.7357,
+      "step": 1932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013329762336451788,
+      "loss": 0.6594,
+      "step": 1933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013310099809070263,
+      "loss": 0.6648,
+      "step": 1934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013290444559644458,
+      "loss": 0.8032,
+      "step": 1935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013270796609557369,
+      "loss": 0.8469,
+      "step": 1936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001325115598018404,
+      "loss": 0.8465,
+      "step": 1937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001323152269289154,
+      "loss": 0.9019,
+      "step": 1938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001321189676903897,
+      "loss": 0.7414,
+      "step": 1939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013192278229977412,
+      "loss": 0.8776,
+      "step": 1940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013172667097049914,
+      "loss": 0.8149,
+      "step": 1941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00013153063391591453,
+      "loss": 0.8749,
+      "step": 1942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001313346713492896,
+      "loss": 0.7275,
+      "step": 1943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001311387834838123,
+      "loss": 0.7389,
+      "step": 1944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001309429705325894,
+      "loss": 0.7799,
+      "step": 1945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001307472327086463,
+      "loss": 0.8711,
+      "step": 1946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001305515702249265,
+      "loss": 0.7159,
+      "step": 1947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001303559832942916,
+      "loss": 0.7764,
+      "step": 1948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00013016047212952104,
+      "loss": 0.6822,
+      "step": 1949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012996503694331178,
+      "loss": 0.7561,
+      "step": 1950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001297696779482782,
+      "loss": 0.8393,
+      "step": 1951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001295743953569516,
+      "loss": 0.7767,
+      "step": 1952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012937918938178052,
+      "loss": 0.8431,
+      "step": 1953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012918406023512978,
+      "loss": 0.7289,
+      "step": 1954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012898900812928079,
+      "loss": 0.7751,
+      "step": 1955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012879403327643106,
+      "loss": 0.7771,
+      "step": 1956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012859913588869428,
+      "loss": 0.7684,
+      "step": 1957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012840431617809958,
+      "loss": 0.8698,
+      "step": 1958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012820957435659172,
+      "loss": 0.8154,
+      "step": 1959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001280149106360308,
+      "loss": 0.8007,
+      "step": 1960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001278203252281918,
+      "loss": 0.8978,
+      "step": 1961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012762581834476456,
+      "loss": 0.7552,
+      "step": 1962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012743139019735352,
+      "loss": 0.8178,
+      "step": 1963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012723704099747748,
+      "loss": 0.7282,
+      "step": 1964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012704277095656932,
+      "loss": 0.724,
+      "step": 1965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001268485802859757,
+      "loss": 0.734,
+      "step": 1966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012665446919695722,
+      "loss": 0.7849,
+      "step": 1967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012646043790068756,
+      "loss": 0.8954,
+      "step": 1968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012626648660825384,
+      "loss": 0.6898,
+      "step": 1969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012607261553065598,
+      "loss": 0.7147,
+      "step": 1970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001258788248788068,
+      "loss": 0.7674,
+      "step": 1971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012568511486353153,
+      "loss": 0.8382,
+      "step": 1972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012549148569556757,
+      "loss": 0.7625,
+      "step": 1973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001252979375855646,
+      "loss": 0.7775,
+      "step": 1974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012510447074408395,
+      "loss": 0.7342,
+      "step": 1975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00012491108538159858,
+      "loss": 0.7572,
+      "step": 1976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012471778170849274,
+      "loss": 0.7195,
+      "step": 1977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012452455993506204,
+      "loss": 0.8503,
+      "step": 1978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012433142027151273,
+      "loss": 0.7265,
+      "step": 1979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001241383629279618,
+      "loss": 0.7178,
+      "step": 1980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001239453881144368,
+      "loss": 0.7758,
+      "step": 1981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001237524960408754,
+      "loss": 0.7486,
+      "step": 1982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012355968691712523,
+      "loss": 0.6596,
+      "step": 1983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012336696095294368,
+      "loss": 0.8075,
+      "step": 1984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012317431835799785,
+      "loss": 0.7528,
+      "step": 1985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001229817593418639,
+      "loss": 0.7661,
+      "step": 1986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001227892841140272,
+      "loss": 0.7071,
+      "step": 1987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012259689288388184,
+      "loss": 0.6794,
+      "step": 1988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012240458586073078,
+      "loss": 0.8315,
+      "step": 1989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012221236325378514,
+      "loss": 0.8007,
+      "step": 1990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012202022527216419,
+      "loss": 0.8108,
+      "step": 1991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012182817212489536,
+      "loss": 0.7695,
+      "step": 1992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012163620402091362,
+      "loss": 0.7752,
+      "step": 1993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012144432116906141,
+      "loss": 0.7886,
+      "step": 1994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012125252377808847,
+      "loss": 0.707,
+      "step": 1995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001210608120566516,
+      "loss": 0.6416,
+      "step": 1996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012086918621331431,
+      "loss": 0.7626,
+      "step": 1997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012067764645654675,
+      "loss": 0.7415,
+      "step": 1998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001204861929947255,
+      "loss": 0.7189,
+      "step": 1999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012029482603613302,
+      "loss": 0.7357,
+      "step": 2000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00012010354578895788,
+      "loss": 0.7909,
+      "step": 2001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011991235246129421,
+      "loss": 0.6961,
+      "step": 2002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011972124626114167,
+      "loss": 0.7272,
+      "step": 2003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000119530227396405,
+      "loss": 0.7409,
+      "step": 2004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011933929607489403,
+      "loss": 0.7532,
+      "step": 2005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011914845250432336,
+      "loss": 0.7915,
+      "step": 2006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011895769689231207,
+      "loss": 0.8186,
+      "step": 2007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011876702944638353,
+      "loss": 0.7606,
+      "step": 2008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011857645037396525,
+      "loss": 0.7402,
+      "step": 2009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011838595988238865,
+      "loss": 0.7321,
+      "step": 2010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011819555817888865,
+      "loss": 0.782,
+      "step": 2011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011800524547060363,
+      "loss": 0.7983,
+      "step": 2012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011781502196457519,
+      "loss": 0.7841,
+      "step": 2013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011762488786774791,
+      "loss": 0.8789,
+      "step": 2014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011743484338696895,
+      "loss": 0.8101,
+      "step": 2015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011724488872898811,
+      "loss": 0.689,
+      "step": 2016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011705502410045749,
+      "loss": 0.6798,
+      "step": 2017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001168652497079312,
+      "loss": 0.7577,
+      "step": 2018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011667556575786513,
+      "loss": 0.8506,
+      "step": 2019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011648597245661679,
+      "loss": 0.7136,
+      "step": 2020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001162964700104452,
+      "loss": 0.7641,
+      "step": 2021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011610705862551043,
+      "loss": 0.7238,
+      "step": 2022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011591773850787345,
+      "loss": 0.7979,
+      "step": 2023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011572850986349601,
+      "loss": 0.6941,
+      "step": 2024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011553937289824044,
+      "loss": 0.8353,
+      "step": 2025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011535032781786901,
+      "loss": 0.7174,
+      "step": 2026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011516137482804446,
+      "loss": 0.7849,
+      "step": 2027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011497251413432895,
+      "loss": 0.7046,
+      "step": 2028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011478374594218446,
+      "loss": 0.7235,
+      "step": 2029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011459507045697235,
+      "loss": 0.6886,
+      "step": 2030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011440648788395297,
+      "loss": 0.6386,
+      "step": 2031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011421799842828567,
+      "loss": 0.8538,
+      "step": 2032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011402960229502863,
+      "loss": 0.9003,
+      "step": 2033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011384129968913822,
+      "loss": 0.7578,
+      "step": 2034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011365309081546927,
+      "loss": 0.8148,
+      "step": 2035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011346497587877469,
+      "loss": 0.704,
+      "step": 2036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011327695508370493,
+      "loss": 0.7431,
+      "step": 2037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011308902863480835,
+      "loss": 0.68,
+      "step": 2038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011290119673653034,
+      "loss": 0.6658,
+      "step": 2039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011271345959321368,
+      "loss": 0.77,
+      "step": 2040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011252581740909809,
+      "loss": 0.718,
+      "step": 2041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011233827038831965,
+      "loss": 0.7781,
+      "step": 2042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011215081873491126,
+      "loss": 0.7968,
+      "step": 2043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011196346265280204,
+      "loss": 0.7186,
+      "step": 2044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011177620234581685,
+      "loss": 0.7415,
+      "step": 2045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011158903801767666,
+      "loss": 0.7577,
+      "step": 2046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011140196987199795,
+      "loss": 0.656,
+      "step": 2047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011121499811229244,
+      "loss": 0.7677,
+      "step": 2048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001110281229419671,
+      "loss": 0.894,
+      "step": 2049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001108413445643239,
+      "loss": 0.769,
+      "step": 2050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011065466318255926,
+      "loss": 0.8013,
+      "step": 2051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011046807899976434,
+      "loss": 0.7766,
+      "step": 2052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011028159221892433,
+      "loss": 0.7058,
+      "step": 2053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011009520304291867,
+      "loss": 0.7082,
+      "step": 2054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010990891167452054,
+      "loss": 0.7934,
+      "step": 2055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010972271831639653,
+      "loss": 0.7574,
+      "step": 2056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010953662317110688,
+      "loss": 0.7616,
+      "step": 2057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010935062644110494,
+      "loss": 0.7762,
+      "step": 2058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010916472832873677,
+      "loss": 0.7709,
+      "step": 2059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010897892903624134,
+      "loss": 0.8338,
+      "step": 2060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001087932287657502,
+      "loss": 0.7634,
+      "step": 2061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010860762771928685,
+      "loss": 0.7831,
+      "step": 2062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010842212609876712,
+      "loss": 0.8538,
+      "step": 2063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010823672410599867,
+      "loss": 0.7299,
+      "step": 2064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010805142194268057,
+      "loss": 0.7855,
+      "step": 2065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010786621981040355,
+      "loss": 0.8392,
+      "step": 2066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010768111791064917,
+      "loss": 0.6805,
+      "step": 2067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010749611644479032,
+      "loss": 0.7646,
+      "step": 2068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010731121561409048,
+      "loss": 0.7952,
+      "step": 2069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001071264156197035,
+      "loss": 0.7299,
+      "step": 2070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010694171666267376,
+      "loss": 0.7545,
+      "step": 2071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010675711894393567,
+      "loss": 0.7879,
+      "step": 2072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001065726226643133,
+      "loss": 0.7548,
+      "step": 2073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001063882280245206,
+      "loss": 0.8099,
+      "step": 2074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010620393522516094,
+      "loss": 0.7943,
+      "step": 2075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010601974446672668,
+      "loss": 0.766,
+      "step": 2076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010583565594959938,
+      "loss": 0.7508,
+      "step": 2077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010565166987404941,
+      "loss": 0.7769,
+      "step": 2078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001054677864402354,
+      "loss": 0.7932,
+      "step": 2079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010528400584820466,
+      "loss": 0.7778,
+      "step": 2080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010510032829789231,
+      "loss": 0.7969,
+      "step": 2081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010491675398912164,
+      "loss": 0.6693,
+      "step": 2082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010473328312160352,
+      "loss": 0.6641,
+      "step": 2083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010454991589493612,
+      "loss": 0.6707,
+      "step": 2084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010436665250860515,
+      "loss": 0.703,
+      "step": 2085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010418349316198323,
+      "loss": 0.7837,
+      "step": 2086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010400043805432966,
+      "loss": 0.7588,
+      "step": 2087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010381748738479051,
+      "loss": 0.7142,
+      "step": 2088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010363464135239826,
+      "loss": 0.7566,
+      "step": 2089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010345190015607134,
+      "loss": 0.8534,
+      "step": 2090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010326926399461431,
+      "loss": 0.8175,
+      "step": 2091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001030867330667175,
+      "loss": 0.8379,
+      "step": 2092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001029043075709565,
+      "loss": 0.7666,
+      "step": 2093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010272198770579256,
+      "loss": 0.8003,
+      "step": 2094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010253977366957163,
+      "loss": 0.733,
+      "step": 2095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010235766566052485,
+      "loss": 0.8517,
+      "step": 2096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010217566387676791,
+      "loss": 0.7585,
+      "step": 2097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010199376851630078,
+      "loss": 0.7575,
+      "step": 2098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010181197977700785,
+      "loss": 0.8311,
+      "step": 2099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010163029785665757,
+      "loss": 0.8775,
+      "step": 2100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010144872295290184,
+      "loss": 0.8072,
+      "step": 2101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010126725526327652,
+      "loss": 0.784,
+      "step": 2102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010108589498520067,
+      "loss": 0.7686,
+      "step": 2103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010090464231597639,
+      "loss": 0.6831,
+      "step": 2104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010072349745278886,
+      "loss": 0.7299,
+      "step": 2105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010054246059270603,
+      "loss": 0.7793,
+      "step": 2106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010036153193267815,
+      "loss": 0.8473,
+      "step": 2107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010018071166953795,
+      "loss": 0.7371,
+      "step": 2108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010000000000000005,
+      "loss": 0.7053,
+      "step": 2109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.981939712066109e-05,
+      "loss": 0.7763,
+      "step": 2110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.963890322799938e-05,
+      "loss": 0.7206,
+      "step": 2111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.945851851837447e-05,
+      "loss": 0.7374,
+      "step": 2112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.927824318802726e-05,
+      "loss": 0.8446,
+      "step": 2113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.909807743307973e-05,
+      "loss": 0.7909,
+      "step": 2114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.891802144953445e-05,
+      "loss": 0.7195,
+      "step": 2115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.873807543327474e-05,
+      "loss": 0.8462,
+      "step": 2116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.855823958006427e-05,
+      "loss": 0.7278,
+      "step": 2117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.837851408554673e-05,
+      "loss": 0.8446,
+      "step": 2118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.819889914524595e-05,
+      "loss": 0.7338,
+      "step": 2119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.801939495456525e-05,
+      "loss": 0.7745,
+      "step": 2120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.784000170878766e-05,
+      "loss": 0.8227,
+      "step": 2121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.766071960307553e-05,
+      "loss": 0.7187,
+      "step": 2122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.748154883247007e-05,
+      "loss": 0.8599,
+      "step": 2123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.730248959189159e-05,
+      "loss": 0.7896,
+      "step": 2124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.712354207613907e-05,
+      "loss": 0.7142,
+      "step": 2125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.694470647988974e-05,
+      "loss": 0.7166,
+      "step": 2126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.676598299769923e-05,
+      "loss": 0.6992,
+      "step": 2127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.658737182400128e-05,
+      "loss": 0.7517,
+      "step": 2128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.64088731531072e-05,
+      "loss": 0.783,
+      "step": 2129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.623048717920608e-05,
+      "loss": 0.8274,
+      "step": 2130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.60522140963645e-05,
+      "loss": 0.7631,
+      "step": 2131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.587405409852592e-05,
+      "loss": 0.5594,
+      "step": 2132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.569600737951112e-05,
+      "loss": 0.7351,
+      "step": 2133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.551807413301733e-05,
+      "loss": 0.8061,
+      "step": 2134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.534025455261855e-05,
+      "loss": 0.671,
+      "step": 2135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.516254883176514e-05,
+      "loss": 0.8225,
+      "step": 2136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.498495716378333e-05,
+      "loss": 0.7581,
+      "step": 2137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.480747974187555e-05,
+      "loss": 0.8013,
+      "step": 2138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.463011675911994e-05,
+      "loss": 0.7154,
+      "step": 2139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.445286840846986e-05,
+      "loss": 0.7151,
+      "step": 2140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.427573488275418e-05,
+      "loss": 0.7445,
+      "step": 2141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.409871637467695e-05,
+      "loss": 0.7218,
+      "step": 2142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.39218130768168e-05,
+      "loss": 0.7554,
+      "step": 2143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.374502518162724e-05,
+      "loss": 0.6584,
+      "step": 2144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.356835288143627e-05,
+      "loss": 0.7891,
+      "step": 2145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.339179636844581e-05,
+      "loss": 0.9167,
+      "step": 2146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.32153558347323e-05,
+      "loss": 0.7209,
+      "step": 2147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.303903147224553e-05,
+      "loss": 0.7365,
+      "step": 2148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.286282347280921e-05,
+      "loss": 0.7995,
+      "step": 2149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.268673202812047e-05,
+      "loss": 0.7415,
+      "step": 2150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.251075732974936e-05,
+      "loss": 0.7079,
+      "step": 2151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.23348995691392e-05,
+      "loss": 0.7299,
+      "step": 2152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.215915893760609e-05,
+      "loss": 0.7684,
+      "step": 2153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.198353562633844e-05,
+      "loss": 0.78,
+      "step": 2154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.180802982639726e-05,
+      "loss": 0.7227,
+      "step": 2155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.163264172871575e-05,
+      "loss": 0.6924,
+      "step": 2156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.145737152409883e-05,
+      "loss": 0.7827,
+      "step": 2157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.128221940322337e-05,
+      "loss": 0.6623,
+      "step": 2158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.11071855566378e-05,
+      "loss": 0.7006,
+      "step": 2159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.093227017476161e-05,
+      "loss": 0.8679,
+      "step": 2160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.075747344788577e-05,
+      "loss": 0.7511,
+      "step": 2161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.05827955661718e-05,
+      "loss": 0.732,
+      "step": 2162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.040823671965224e-05,
+      "loss": 0.7287,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.023379709823006e-05,
+      "loss": 0.7161,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.005947689167829e-05,
+      "loss": 0.792,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.988527628964036e-05,
+      "loss": 0.8636,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.97111954816295e-05,
+      "loss": 0.8679,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.95372346570284e-05,
+      "loss": 0.8083,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.936339400508948e-05,
+      "loss": 0.8217,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.91896737149344e-05,
+      "loss": 0.7105,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.901607397555367e-05,
+      "loss": 0.7122,
+      "step": 2171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.884259497580686e-05,
+      "loss": 0.8017,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.866923690442212e-05,
+      "loss": 0.6792,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.849599994999607e-05,
+      "loss": 0.6699,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.832288430099354e-05,
+      "loss": 0.8044,
+      "step": 2175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.81498901457473e-05,
+      "loss": 0.7479,
+      "step": 2176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.797701767245809e-05,
+      "loss": 0.603,
+      "step": 2177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.780426706919429e-05,
+      "loss": 0.8083,
+      "step": 2178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.763163852389149e-05,
+      "loss": 0.7399,
+      "step": 2179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.745913222435275e-05,
+      "loss": 0.6923,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.72867483582479e-05,
+      "loss": 0.759,
+      "step": 2181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.711448711311377e-05,
+      "loss": 0.8103,
+      "step": 2182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.694234867635374e-05,
+      "loss": 0.733,
+      "step": 2183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.677033323523744e-05,
+      "loss": 0.8951,
+      "step": 2184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.659844097690088e-05,
+      "loss": 0.8446,
+      "step": 2185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.642667208834608e-05,
+      "loss": 0.7458,
+      "step": 2186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.625502675644055e-05,
+      "loss": 0.8104,
+      "step": 2187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.608350516791775e-05,
+      "loss": 0.8716,
+      "step": 2188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.591210750937635e-05,
+      "loss": 0.7066,
+      "step": 2189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.57408339672801e-05,
+      "loss": 0.7055,
+      "step": 2190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.55696847279579e-05,
+      "loss": 0.7227,
+      "step": 2191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.539865997760341e-05,
+      "loss": 0.7484,
+      "step": 2192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.522775990227465e-05,
+      "loss": 0.7159,
+      "step": 2193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.505698468789431e-05,
+      "loss": 0.865,
+      "step": 2194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.488633452024892e-05,
+      "loss": 0.8084,
+      "step": 2195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.471580958498921e-05,
+      "loss": 0.8713,
+      "step": 2196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.45454100676297e-05,
+      "loss": 0.7681,
+      "step": 2197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.43751361535482e-05,
+      "loss": 0.7993,
+      "step": 2198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.420498802798608e-05,
+      "loss": 0.7198,
+      "step": 2199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.403496587604795e-05,
+      "loss": 0.741,
+      "step": 2200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.386506988270109e-05,
+      "loss": 0.7102,
+      "step": 2201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.369530023277574e-05,
+      "loss": 0.8062,
+      "step": 2202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.352565711096472e-05,
+      "loss": 0.6974,
+      "step": 2203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.335614070182299e-05,
+      "loss": 0.7153,
+      "step": 2204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.318675118976791e-05,
+      "loss": 0.818,
+      "step": 2205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.301748875907853e-05,
+      "loss": 0.7777,
+      "step": 2206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.284835359389587e-05,
+      "loss": 0.6768,
+      "step": 2207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.267934587822248e-05,
+      "loss": 0.7808,
+      "step": 2208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.251046579592204e-05,
+      "loss": 0.8128,
+      "step": 2209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.234171353071958e-05,
+      "loss": 0.7281,
+      "step": 2210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.217308926620115e-05,
+      "loss": 0.7772,
+      "step": 2211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.200459318581324e-05,
+      "loss": 0.7216,
+      "step": 2212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.183622547286314e-05,
+      "loss": 0.7617,
+      "step": 2213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.16679863105185e-05,
+      "loss": 0.7828,
+      "step": 2214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.149987588180692e-05,
+      "loss": 0.6622,
+      "step": 2215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.13318943696161e-05,
+      "loss": 0.7367,
+      "step": 2216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.116404195669358e-05,
+      "loss": 0.7316,
+      "step": 2217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.099631882564614e-05,
+      "loss": 0.7583,
+      "step": 2218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.08287251589403e-05,
+      "loss": 0.694,
+      "step": 2219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.066126113890133e-05,
+      "loss": 0.7845,
+      "step": 2220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.049392694771382e-05,
+      "loss": 0.6729,
+      "step": 2221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.032672276742099e-05,
+      "loss": 0.6877,
+      "step": 2222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.015964877992451e-05,
+      "loss": 0.8266,
+      "step": 2223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.999270516698451e-05,
+      "loss": 0.7755,
+      "step": 2224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.982589211021939e-05,
+      "loss": 0.7646,
+      "step": 2225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.965920979110525e-05,
+      "loss": 0.8221,
+      "step": 2226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.94926583909762e-05,
+      "loss": 0.9021,
+      "step": 2227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.932623809102391e-05,
+      "loss": 0.7158,
+      "step": 2228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.915994907229718e-05,
+      "loss": 0.7915,
+      "step": 2229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.899379151570227e-05,
+      "loss": 0.6226,
+      "step": 2230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.88277656020024e-05,
+      "loss": 0.7362,
+      "step": 2231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.866187151181731e-05,
+      "loss": 0.7581,
+      "step": 2232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.849610942562366e-05,
+      "loss": 0.8429,
+      "step": 2233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.833047952375426e-05,
+      "loss": 0.8257,
+      "step": 2234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.816498198639821e-05,
+      "loss": 0.7877,
+      "step": 2235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.799961699360072e-05,
+      "loss": 0.8056,
+      "step": 2236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.783438472526257e-05,
+      "loss": 0.7669,
+      "step": 2237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.766928536114031e-05,
+      "loss": 0.7235,
+      "step": 2238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.750431908084594e-05,
+      "loss": 0.6324,
+      "step": 2239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.733948606384647e-05,
+      "loss": 0.796,
+      "step": 2240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.717478648946415e-05,
+      "loss": 0.7805,
+      "step": 2241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.701022053687603e-05,
+      "loss": 0.919,
+      "step": 2242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.684578838511361e-05,
+      "loss": 0.8261,
+      "step": 2243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.6681490213063e-05,
+      "loss": 0.7703,
+      "step": 2244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.651732619946462e-05,
+      "loss": 0.698,
+      "step": 2245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.635329652291265e-05,
+      "loss": 0.7535,
+      "step": 2246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.618940136185544e-05,
+      "loss": 0.751,
+      "step": 2247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.602564089459476e-05,
+      "loss": 0.6667,
+      "step": 2248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.586201529928598e-05,
+      "loss": 0.6973,
+      "step": 2249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.569852475393779e-05,
+      "loss": 0.7584,
+      "step": 2250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.553516943641175e-05,
+      "loss": 0.6655,
+      "step": 2251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.53719495244225e-05,
+      "loss": 0.7899,
+      "step": 2252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.520886519553738e-05,
+      "loss": 0.7955,
+      "step": 2253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.504591662717599e-05,
+      "loss": 0.7927,
+      "step": 2254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.488310399661051e-05,
+      "loss": 0.6105,
+      "step": 2255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.47204274809652e-05,
+      "loss": 0.7126,
+      "step": 2256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.455788725721598e-05,
+      "loss": 0.7496,
+      "step": 2257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.439548350219077e-05,
+      "loss": 0.7383,
+      "step": 2258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.423321639256902e-05,
+      "loss": 0.706,
+      "step": 2259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.40710861048813e-05,
+      "loss": 0.738,
+      "step": 2260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.390909281550959e-05,
+      "loss": 0.7764,
+      "step": 2261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.37472367006866e-05,
+      "loss": 0.7001,
+      "step": 2262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.358551793649597e-05,
+      "loss": 0.7121,
+      "step": 2263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.342393669887198e-05,
+      "loss": 0.7112,
+      "step": 2264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.326249316359897e-05,
+      "loss": 0.8563,
+      "step": 2265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.310118750631179e-05,
+      "loss": 0.7849,
+      "step": 2266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.294001990249526e-05,
+      "loss": 0.6511,
+      "step": 2267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.277899052748378e-05,
+      "loss": 0.6774,
+      "step": 2268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.261809955646164e-05,
+      "loss": 0.5751,
+      "step": 2269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.245734716446243e-05,
+      "loss": 0.8173,
+      "step": 2270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.229673352636896e-05,
+      "loss": 0.9327,
+      "step": 2271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.213625881691312e-05,
+      "loss": 0.8318,
+      "step": 2272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.197592321067578e-05,
+      "loss": 0.7903,
+      "step": 2273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.181572688208622e-05,
+      "loss": 0.7507,
+      "step": 2274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.165567000542243e-05,
+      "loss": 0.6838,
+      "step": 2275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.149575275481056e-05,
+      "loss": 0.7761,
+      "step": 2276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.13359753042249e-05,
+      "loss": 0.6937,
+      "step": 2277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.117633782748774e-05,
+      "loss": 0.8209,
+      "step": 2278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.101684049826887e-05,
+      "loss": 0.7558,
+      "step": 2279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.085748349008585e-05,
+      "loss": 0.7804,
+      "step": 2280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.069826697630352e-05,
+      "loss": 0.7261,
+      "step": 2281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.053919113013374e-05,
+      "loss": 0.7342,
+      "step": 2282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.038025612463548e-05,
+      "loss": 0.7313,
+      "step": 2283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.022146213271452e-05,
+      "loss": 0.8266,
+      "step": 2284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.006280932712301e-05,
+      "loss": 0.7289,
+      "step": 2285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.990429788045983e-05,
+      "loss": 0.71,
+      "step": 2286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.974592796516974e-05,
+      "loss": 0.7515,
+      "step": 2287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.958769975354373e-05,
+      "loss": 0.8274,
+      "step": 2288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.94296134177187e-05,
+      "loss": 0.8009,
+      "step": 2289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.927166912967693e-05,
+      "loss": 0.7245,
+      "step": 2290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.911386706124638e-05,
+      "loss": 0.7548,
+      "step": 2291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.895620738410034e-05,
+      "loss": 0.7102,
+      "step": 2292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.87986902697569e-05,
+      "loss": 0.7234,
+      "step": 2293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.864131588957934e-05,
+      "loss": 0.8104,
+      "step": 2294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.848408441477563e-05,
+      "loss": 0.7507,
+      "step": 2295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.832699601639807e-05,
+      "loss": 0.7712,
+      "step": 2296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.817005086534349e-05,
+      "loss": 0.8351,
+      "step": 2297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.801324913235292e-05,
+      "loss": 0.7288,
+      "step": 2298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.785659098801114e-05,
+      "loss": 0.739,
+      "step": 2299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.770007660274698e-05,
+      "loss": 0.8032,
+      "step": 2300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.75437061468326e-05,
+      "loss": 0.7044,
+      "step": 2301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.738747979038387e-05,
+      "loss": 0.745,
+      "step": 2302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.723139770335973e-05,
+      "loss": 0.7822,
+      "step": 2303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.707546005556212e-05,
+      "loss": 0.6324,
+      "step": 2304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.691966701663596e-05,
+      "loss": 0.849,
+      "step": 2305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.676401875606888e-05,
+      "loss": 0.8472,
+      "step": 2306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.66085154431908e-05,
+      "loss": 0.7656,
+      "step": 2307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.645315724717414e-05,
+      "loss": 0.8898,
+      "step": 2308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.629794433703346e-05,
+      "loss": 0.701,
+      "step": 2309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.614287688162506e-05,
+      "loss": 0.8077,
+      "step": 2310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.598795504964723e-05,
+      "loss": 0.7322,
+      "step": 2311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.583317900963977e-05,
+      "loss": 0.7283,
+      "step": 2312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.56785489299837e-05,
+      "loss": 0.6928,
+      "step": 2313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.552406497890155e-05,
+      "loss": 0.5485,
+      "step": 2314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.53697273244566e-05,
+      "loss": 0.6742,
+      "step": 2315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.521553613455314e-05,
+      "loss": 0.7358,
+      "step": 2316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.50614915769361e-05,
+      "loss": 0.6445,
+      "step": 2317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.490759381919078e-05,
+      "loss": 0.8179,
+      "step": 2318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.475384302874292e-05,
+      "loss": 0.8205,
+      "step": 2319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.460023937285835e-05,
+      "loss": 0.6919,
+      "step": 2320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.444678301864267e-05,
+      "loss": 0.7496,
+      "step": 2321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.429347413304141e-05,
+      "loss": 0.688,
+      "step": 2322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.414031288283972e-05,
+      "loss": 0.7133,
+      "step": 2323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.398729943466184e-05,
+      "loss": 0.7027,
+      "step": 2324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.383443395497152e-05,
+      "loss": 0.7495,
+      "step": 2325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.368171661007146e-05,
+      "loss": 0.6732,
+      "step": 2326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.352914756610302e-05,
+      "loss": 0.6303,
+      "step": 2327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.337672698904653e-05,
+      "loss": 0.9371,
+      "step": 2328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.322445504472048e-05,
+      "loss": 0.7963,
+      "step": 2329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.30723318987819e-05,
+      "loss": 0.6916,
+      "step": 2330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.292035771672591e-05,
+      "loss": 0.8467,
+      "step": 2331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.276853266388543e-05,
+      "loss": 0.7584,
+      "step": 2332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.261685690543128e-05,
+      "loss": 0.7608,
+      "step": 2333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.246533060637187e-05,
+      "loss": 0.8151,
+      "step": 2334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.231395393155289e-05,
+      "loss": 0.7919,
+      "step": 2335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.216272704565731e-05,
+      "loss": 0.7811,
+      "step": 2336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.201165011320527e-05,
+      "loss": 0.6673,
+      "step": 2337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.186072329855355e-05,
+      "loss": 0.691,
+      "step": 2338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.170994676589576e-05,
+      "loss": 0.8197,
+      "step": 2339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.155932067926207e-05,
+      "loss": 0.7821,
+      "step": 2340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.140884520251878e-05,
+      "loss": 0.68,
+      "step": 2341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.125852049936858e-05,
+      "loss": 0.695,
+      "step": 2342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.110834673334989e-05,
+      "loss": 0.6637,
+      "step": 2343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.09583240678371e-05,
+      "loss": 0.766,
+      "step": 2344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.080845266604023e-05,
+      "loss": 0.7759,
+      "step": 2345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.065873269100457e-05,
+      "loss": 0.7313,
+      "step": 2346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0509164305610844e-05,
+      "loss": 0.7697,
+      "step": 2347
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0359747672574815e-05,
+      "loss": 0.7566,
+      "step": 2348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.0210482954447065e-05,
+      "loss": 0.745,
+      "step": 2349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.0061370313613005e-05,
+      "loss": 0.8086,
+      "step": 2350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.991240991229265e-05,
+      "loss": 0.798,
+      "step": 2351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.97636019125402e-05,
+      "loss": 0.7364,
+      "step": 2352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9614946476244197e-05,
+      "loss": 0.5915,
+      "step": 2353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.946644376512729e-05,
+      "loss": 0.8399,
+      "step": 2354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.931809394074574e-05,
+      "loss": 0.8411,
+      "step": 2355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.916989716448973e-05,
+      "loss": 0.7409,
+      "step": 2356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.902185359758272e-05,
+      "loss": 0.776,
+      "step": 2357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.8873963401081645e-05,
+      "loss": 0.7978,
+      "step": 2358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.872622673587662e-05,
+      "loss": 0.7254,
+      "step": 2359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.857864376269051e-05,
+      "loss": 0.7013,
+      "step": 2360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.843121464207921e-05,
+      "loss": 0.7079,
+      "step": 2361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.828393953443121e-05,
+      "loss": 0.7458,
+      "step": 2362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.813681859996727e-05,
+      "loss": 0.7365,
+      "step": 2363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7989851998740605e-05,
+      "loss": 0.6855,
+      "step": 2364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.784303989063655e-05,
+      "loss": 0.862,
+      "step": 2365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7696382435372146e-05,
+      "loss": 0.8226,
+      "step": 2366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.7549879792496445e-05,
+      "loss": 0.7235,
+      "step": 2367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.740353212138987e-05,
+      "loss": 0.7999,
+      "step": 2368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.725733958126438e-05,
+      "loss": 0.6642,
+      "step": 2369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.711130233116322e-05,
+      "loss": 0.7572,
+      "step": 2370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.696542052996046e-05,
+      "loss": 0.8031,
+      "step": 2371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.681969433636125e-05,
+      "loss": 0.8151,
+      "step": 2372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.66741239089015e-05,
+      "loss": 0.7799,
+      "step": 2373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.6528709405947436e-05,
+      "loss": 0.7265,
+      "step": 2374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.638345098569584e-05,
+      "loss": 0.6018,
+      "step": 2375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.623834880617369e-05,
+      "loss": 0.8466,
+      "step": 2376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.609340302523782e-05,
+      "loss": 0.7479,
+      "step": 2377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.59486138005751e-05,
+      "loss": 0.7232,
+      "step": 2378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.580398128970208e-05,
+      "loss": 0.7334,
+      "step": 2379
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.5659505649964605e-05,
+      "loss": 0.8386,
+      "step": 2380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.5515187038538174e-05,
+      "loss": 0.7448,
+      "step": 2381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.537102561242715e-05,
+      "loss": 0.8319,
+      "step": 2382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.5227021528465085e-05,
+      "loss": 0.7452,
+      "step": 2383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.508317494331441e-05,
+      "loss": 0.784,
+      "step": 2384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4939486013465926e-05,
+      "loss": 0.682,
+      "step": 2385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.479595489523921e-05,
+      "loss": 0.6914,
+      "step": 2386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.465258174478212e-05,
+      "loss": 0.7594,
+      "step": 2387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.450936671807043e-05,
+      "loss": 0.7741,
+      "step": 2388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.436630997090815e-05,
+      "loss": 0.7039,
+      "step": 2389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.422341165892704e-05,
+      "loss": 0.6093,
+      "step": 2390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4080671937586346e-05,
+      "loss": 0.7187,
+      "step": 2391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.393809096217295e-05,
+      "loss": 0.7187,
+      "step": 2392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.379566888780105e-05,
+      "loss": 0.7656,
+      "step": 2393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3653405869411765e-05,
+      "loss": 0.7519,
+      "step": 2394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.351130206177346e-05,
+      "loss": 0.6298,
+      "step": 2395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.336935761948101e-05,
+      "loss": 0.6215,
+      "step": 2396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.3227572696956106e-05,
+      "loss": 0.6595,
+      "step": 2397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.308594744844695e-05,
+      "loss": 0.6982,
+      "step": 2398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2944482028027777e-05,
+      "loss": 0.8867,
+      "step": 2399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.280317658959915e-05,
+      "loss": 0.8048,
+      "step": 2400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.266203128688762e-05,
+      "loss": 0.6889,
+      "step": 2401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.252104627344534e-05,
+      "loss": 0.716,
+      "step": 2402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.23802217026502e-05,
+      "loss": 0.7043,
+      "step": 2403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.223955772770563e-05,
+      "loss": 0.6725,
+      "step": 2404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.2099054501640096e-05,
+      "loss": 0.6876,
+      "step": 2405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1958712177307436e-05,
+      "loss": 0.8085,
+      "step": 2406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1818530907386377e-05,
+      "loss": 0.7201,
+      "step": 2407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.167851084438027e-05,
+      "loss": 0.7661,
+      "step": 2408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.153865214061735e-05,
+      "loss": 0.8382,
+      "step": 2409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.1398954948250086e-05,
+      "loss": 0.6911,
+      "step": 2410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.125941941925536e-05,
+      "loss": 0.7329,
+      "step": 2411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.112004570543423e-05,
+      "loss": 0.7546,
+      "step": 2412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0980833958411514e-05,
+      "loss": 0.6313,
+      "step": 2413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.084178432963598e-05,
+      "loss": 0.7307,
+      "step": 2414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.070289697038009e-05,
+      "loss": 0.8044,
+      "step": 2415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0564172031739554e-05,
+      "loss": 0.8045,
+      "step": 2416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.042560966463357e-05,
+      "loss": 0.6823,
+      "step": 2417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.028721001980445e-05,
+      "loss": 0.8117,
+      "step": 2418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.0148973247817334e-05,
+      "loss": 0.7376,
+      "step": 2419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.001089949906035e-05,
+      "loss": 0.7219,
+      "step": 2420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.987298892374426e-05,
+      "loss": 0.7534,
+      "step": 2421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.973524167190213e-05,
+      "loss": 0.669,
+      "step": 2422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9597657893389614e-05,
+      "loss": 0.7289,
+      "step": 2423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.946023773788422e-05,
+      "loss": 0.8166,
+      "step": 2424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9322981354885666e-05,
+      "loss": 0.826,
+      "step": 2425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.918588889371556e-05,
+      "loss": 0.782,
+      "step": 2426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.904896050351688e-05,
+      "loss": 0.7019,
+      "step": 2427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.891219633325439e-05,
+      "loss": 0.6846,
+      "step": 2428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.877559653171415e-05,
+      "loss": 0.8003,
+      "step": 2429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.863916124750323e-05,
+      "loss": 0.721,
+      "step": 2430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.850289062904994e-05,
+      "loss": 0.774,
+      "step": 2431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.8366784824603375e-05,
+      "loss": 0.8205,
+      "step": 2432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.823084398223321e-05,
+      "loss": 0.7529,
+      "step": 2433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.809506824982981e-05,
+      "loss": 0.6702,
+      "step": 2434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7959457775103936e-05,
+      "loss": 0.7257,
+      "step": 2435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.782401270558636e-05,
+      "loss": 0.7438,
+      "step": 2436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7688733188628186e-05,
+      "loss": 0.7678,
+      "step": 2437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7553619371400126e-05,
+      "loss": 0.7293,
+      "step": 2438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.741867140089284e-05,
+      "loss": 0.6818,
+      "step": 2439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.728388942391657e-05,
+      "loss": 0.8632,
+      "step": 2440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7149273587100754e-05,
+      "loss": 0.7745,
+      "step": 2441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.701482403689428e-05,
+      "loss": 0.7307,
+      "step": 2442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.688054091956515e-05,
+      "loss": 0.6636,
+      "step": 2443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.674642438120014e-05,
+      "loss": 0.7753,
+      "step": 2444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6612474567704924e-05,
+      "loss": 0.8681,
+      "step": 2445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6478691624803826e-05,
+      "loss": 0.6147,
+      "step": 2446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.634507569803952e-05,
+      "loss": 0.6298,
+      "step": 2447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.621162693277308e-05,
+      "loss": 0.6658,
+      "step": 2448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607834547418362e-05,
+      "loss": 0.7404,
+      "step": 2449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.594523146726832e-05,
+      "loss": 0.716,
+      "step": 2450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5812285056842254e-05,
+      "loss": 0.6832,
+      "step": 2451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5679506387537975e-05,
+      "loss": 0.7695,
+      "step": 2452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.554689560380572e-05,
+      "loss": 0.8528,
+      "step": 2453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.541445284991306e-05,
+      "loss": 0.8197,
+      "step": 2454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.528217826994463e-05,
+      "loss": 0.6861,
+      "step": 2455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.515007200780228e-05,
+      "loss": 0.8054,
+      "step": 2456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.501813420720471e-05,
+      "loss": 0.6675,
+      "step": 2457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4886365011687214e-05,
+      "loss": 0.7372,
+      "step": 2458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4754764564601814e-05,
+      "loss": 0.7998,
+      "step": 2459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.462333300911694e-05,
+      "loss": 0.6849,
+      "step": 2460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.449207048821717e-05,
+      "loss": 0.6642,
+      "step": 2461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.436097714470333e-05,
+      "loss": 0.8221,
+      "step": 2462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.423005312119204e-05,
+      "loss": 0.7798,
+      "step": 2463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.409929856011588e-05,
+      "loss": 0.781,
+      "step": 2464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.396871360372301e-05,
+      "loss": 0.8261,
+      "step": 2465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.383829839407698e-05,
+      "loss": 0.7527,
+      "step": 2466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.370805307305681e-05,
+      "loss": 0.7788,
+      "step": 2467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.357797778235668e-05,
+      "loss": 0.7739,
+      "step": 2468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.344807266348567e-05,
+      "loss": 0.6391,
+      "step": 2469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3318337857767866e-05,
+      "loss": 0.707,
+      "step": 2470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3188773506342074e-05,
+      "loss": 0.7503,
+      "step": 2471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.30593797501615e-05,
+      "loss": 0.7252,
+      "step": 2472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2930156729993943e-05,
+      "loss": 0.791,
+      "step": 2473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.280110458642139e-05,
+      "loss": 0.7355,
+      "step": 2474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.267222345983985e-05,
+      "loss": 0.6599,
+      "step": 2475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2543513490459486e-05,
+      "loss": 0.7434,
+      "step": 2476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.241497481830396e-05,
+      "loss": 0.722,
+      "step": 2477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2286607583210856e-05,
+      "loss": 0.7841,
+      "step": 2478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2158411924831185e-05,
+      "loss": 0.7707,
+      "step": 2479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.203038798262915e-05,
+      "loss": 0.7663,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.190253589588229e-05,
+      "loss": 0.7541,
+      "step": 2481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1774855803681234e-05,
+      "loss": 0.7465,
+      "step": 2482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1647347844929255e-05,
+      "loss": 0.8535,
+      "step": 2483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.15200121583426e-05,
+      "loss": 0.7352,
+      "step": 2484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.139284888245008e-05,
+      "loss": 0.6574,
+      "step": 2485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.126585815559272e-05,
+      "loss": 0.7074,
+      "step": 2486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1139040115924085e-05,
+      "loss": 0.7027,
+      "step": 2487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1012394901409824e-05,
+      "loss": 0.7803,
+      "step": 2488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.088592264982738e-05,
+      "loss": 0.7618,
+      "step": 2489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0759623498766296e-05,
+      "loss": 0.741,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0633497585627556e-05,
+      "loss": 0.7734,
+      "step": 2491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.050754504762384e-05,
+      "loss": 0.734,
+      "step": 2492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.038176602177923e-05,
+      "loss": 0.6657,
+      "step": 2493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.025616064492887e-05,
+      "loss": 0.7789,
+      "step": 2494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.013072905371915e-05,
+      "loss": 0.6721,
+      "step": 2495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0005471384607394e-05,
+      "loss": 0.7718,
+      "step": 2496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.988038777386156e-05,
+      "loss": 0.8608,
+      "step": 2497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.975547835756044e-05,
+      "loss": 0.7109,
+      "step": 2498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9630743271593265e-05,
+      "loss": 0.7201,
+      "step": 2499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.950618265165946e-05,
+      "loss": 0.7349,
+      "step": 2500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.938179663326886e-05,
+      "loss": 0.8182,
+      "step": 2501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.925758535174131e-05,
+      "loss": 0.7843,
+      "step": 2502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.91335489422064e-05,
+      "loss": 0.6428,
+      "step": 2503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.900968753960366e-05,
+      "loss": 0.7779,
+      "step": 2504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.8886001278682096e-05,
+      "loss": 0.8912,
+      "step": 2505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.876249029400028e-05,
+      "loss": 0.6446,
+      "step": 2506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8639154719926074e-05,
+      "loss": 0.7135,
+      "step": 2507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.851599469063645e-05,
+      "loss": 0.6328,
+      "step": 2508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8393010340117444e-05,
+      "loss": 0.8034,
+      "step": 2509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8270201802164055e-05,
+      "loss": 0.7994,
+      "step": 2510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.814756921037983e-05,
+      "loss": 0.6839,
+      "step": 2511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.802511269817708e-05,
+      "loss": 0.6865,
+      "step": 2512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.79028323987765e-05,
+      "loss": 0.8033,
+      "step": 2513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7780728445206994e-05,
+      "loss": 0.7571,
+      "step": 2514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7658800970305744e-05,
+      "loss": 0.7148,
+      "step": 2515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.753705010671795e-05,
+      "loss": 0.7347,
+      "step": 2516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.74154759868965e-05,
+      "loss": 0.6944,
+      "step": 2517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.729407874310225e-05,
+      "loss": 0.7061,
+      "step": 2518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7172858507403354e-05,
+      "loss": 0.7231,
+      "step": 2519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.705181541167564e-05,
+      "loss": 0.749,
+      "step": 2520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.693094958760217e-05,
+      "loss": 0.7728,
+      "step": 2521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.681026116667301e-05,
+      "loss": 0.7423,
+      "step": 2522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.668975028018537e-05,
+      "loss": 0.6566,
+      "step": 2523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656941705924337e-05,
+      "loss": 0.6635,
+      "step": 2524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644926163475766e-05,
+      "loss": 0.802,
+      "step": 2525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6329284137445564e-05,
+      "loss": 0.7707,
+      "step": 2526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.620948469783094e-05,
+      "loss": 0.7141,
+      "step": 2527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.608986344624372e-05,
+      "loss": 0.8285,
+      "step": 2528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.597042051282018e-05,
+      "loss": 0.8119,
+      "step": 2529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.585115602750246e-05,
+      "loss": 0.7395,
+      "step": 2530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.573207012003867e-05,
+      "loss": 0.6763,
+      "step": 2531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.5613162919982646e-05,
+      "loss": 0.8342,
+      "step": 2532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.549443455669368e-05,
+      "loss": 0.8135,
+      "step": 2533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.537588515933663e-05,
+      "loss": 0.8046,
+      "step": 2534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.525751485688169e-05,
+      "loss": 0.662,
+      "step": 2535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.513932377810403e-05,
+      "loss": 0.7869,
+      "step": 2536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.502131205158401e-05,
+      "loss": 0.6702,
+      "step": 2537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.49034798057069e-05,
+      "loss": 0.8027,
+      "step": 2538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4785827168662477e-05,
+      "loss": 0.8702,
+      "step": 2539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.466835426844539e-05,
+      "loss": 0.6849,
+      "step": 2540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.455106123285468e-05,
+      "loss": 0.8581,
+      "step": 2541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.443394818949353e-05,
+      "loss": 0.8513,
+      "step": 2542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.43170152657696e-05,
+      "loss": 0.761,
+      "step": 2543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.4200262588894325e-05,
+      "loss": 0.769,
+      "step": 2544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.408369028588323e-05,
+      "loss": 0.7395,
+      "step": 2545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.396729848355562e-05,
+      "loss": 0.664,
+      "step": 2546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3851087308534254e-05,
+      "loss": 0.8049,
+      "step": 2547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.373505688724554e-05,
+      "loss": 0.8225,
+      "step": 2548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3619207345919234e-05,
+      "loss": 0.8929,
+      "step": 2549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3503538810588256e-05,
+      "loss": 0.7008,
+      "step": 2550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3388051407088696e-05,
+      "loss": 0.7453,
+      "step": 2551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.327274526105944e-05,
+      "loss": 0.6943,
+      "step": 2552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.315762049794229e-05,
+      "loss": 0.757,
+      "step": 2553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.304267724298178e-05,
+      "loss": 0.7603,
+      "step": 2554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2927915621224814e-05,
+      "loss": 0.7382,
+      "step": 2555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.281333575752084e-05,
+      "loss": 0.7545,
+      "step": 2556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2698937776521535e-05,
+      "loss": 0.7312,
+      "step": 2557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.258472180268062e-05,
+      "loss": 0.786,
+      "step": 2558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.247068796025394e-05,
+      "loss": 0.8363,
+      "step": 2559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2356836373299136e-05,
+      "loss": 0.6958,
+      "step": 2560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2243167165675544e-05,
+      "loss": 0.8016,
+      "step": 2561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2129680461044184e-05,
+      "loss": 0.7865,
+      "step": 2562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.2016376382867364e-05,
+      "loss": 0.6353,
+      "step": 2563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.190325505440888e-05,
+      "loss": 0.6574,
+      "step": 2564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1790316598733685e-05,
+      "loss": 0.8378,
+      "step": 2565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.167756113870768e-05,
+      "loss": 0.7286,
+      "step": 2566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.156498879699776e-05,
+      "loss": 0.7892,
+      "step": 2567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.145259969607168e-05,
+      "loss": 0.7776,
+      "step": 2568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1340393958197676e-05,
+      "loss": 0.7558,
+      "step": 2569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1228371705444615e-05,
+      "loss": 0.7507,
+      "step": 2570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.111653305968178e-05,
+      "loss": 0.766,
+      "step": 2571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1004878142578595e-05,
+      "loss": 0.6055,
+      "step": 2572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.089340707560469e-05,
+      "loss": 0.6961,
+      "step": 2573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0782119980029666e-05,
+      "loss": 0.7414,
+      "step": 2574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0671016976922965e-05,
+      "loss": 0.7639,
+      "step": 2575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0560098187153793e-05,
+      "loss": 0.8134,
+      "step": 2576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0449363731390845e-05,
+      "loss": 0.6797,
+      "step": 2577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0338813730102435e-05,
+      "loss": 0.7085,
+      "step": 2578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0228448303556135e-05,
+      "loss": 0.7708,
+      "step": 2579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0118267571818636e-05,
+      "loss": 0.8382,
+      "step": 2580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0008271654755814e-05,
+      "loss": 0.7312,
+      "step": 2581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9898460672032504e-05,
+      "loss": 0.732,
+      "step": 2582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.978883474311218e-05,
+      "loss": 0.8608,
+      "step": 2583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9679393987257188e-05,
+      "loss": 0.6173,
+      "step": 2584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9570138523528368e-05,
+      "loss": 0.6616,
+      "step": 2585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.946106847078487e-05,
+      "loss": 0.6787,
+      "step": 2586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9352183947684243e-05,
+      "loss": 0.8153,
+      "step": 2587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9243485072682265e-05,
+      "loss": 0.6658,
+      "step": 2588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.913497196403252e-05,
+      "loss": 0.8054,
+      "step": 2589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.902664473978678e-05,
+      "loss": 0.7597,
+      "step": 2590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8918503517794303e-05,
+      "loss": 0.6924,
+      "step": 2591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.881054841570223e-05,
+      "loss": 0.7906,
+      "step": 2592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.870277955095515e-05,
+      "loss": 0.6481,
+      "step": 2593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8595197040794942e-05,
+      "loss": 0.7108,
+      "step": 2594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.848780100226087e-05,
+      "loss": 0.7391,
+      "step": 2595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.838059155218935e-05,
+      "loss": 0.7083,
+      "step": 2596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.827356880721368e-05,
+      "loss": 0.8054,
+      "step": 2597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8166732883764145e-05,
+      "loss": 0.7458,
+      "step": 2598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8060083898067803e-05,
+      "loss": 0.7395,
+      "step": 2599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7953621966148213e-05,
+      "loss": 0.7152,
+      "step": 2600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.784734720382556e-05,
+      "loss": 0.6637,
+      "step": 2601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.774125972671644e-05,
+      "loss": 0.6999,
+      "step": 2602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7635359650233493e-05,
+      "loss": 0.8484,
+      "step": 2603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7529647089585765e-05,
+      "loss": 0.6216,
+      "step": 2604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.742412215977803e-05,
+      "loss": 0.7074,
+      "step": 2605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7318784975611132e-05,
+      "loss": 0.6861,
+      "step": 2606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.721363565168167e-05,
+      "loss": 0.6837,
+      "step": 2607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.7108674302381664e-05,
+      "loss": 0.7155,
+      "step": 2608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.700390104189887e-05,
+      "loss": 0.7338,
+      "step": 2609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6899315984216333e-05,
+      "loss": 0.615,
+      "step": 2610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.679491924311226e-05,
+      "loss": 0.863,
+      "step": 2611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6690710932160157e-05,
+      "loss": 0.7759,
+      "step": 2612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.658669116472845e-05,
+      "loss": 0.72,
+      "step": 2613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6482860053980374e-05,
+      "loss": 0.9088,
+      "step": 2614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6379217712874105e-05,
+      "loss": 0.6683,
+      "step": 2615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.627576425416225e-05,
+      "loss": 0.9264,
+      "step": 2616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6172499790392092e-05,
+      "loss": 0.7087,
+      "step": 2617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6069424433905255e-05,
+      "loss": 0.701,
+      "step": 2618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5966538296837594e-05,
+      "loss": 0.7582,
+      "step": 2619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5863841491119135e-05,
+      "loss": 0.7278,
+      "step": 2620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5761334128474013e-05,
+      "loss": 0.7395,
+      "step": 2621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5659016320420072e-05,
+      "loss": 0.8499,
+      "step": 2622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5556888178269132e-05,
+      "loss": 0.7636,
+      "step": 2623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.545494981312664e-05,
+      "loss": 0.7299,
+      "step": 2624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.535320133589143e-05,
+      "loss": 0.8033,
+      "step": 2625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5251642857255963e-05,
+      "loss": 0.6575,
+      "step": 2626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5150274487705926e-05,
+      "loss": 0.7163,
+      "step": 2627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5049096337520084e-05,
+      "loss": 0.685,
+      "step": 2628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4948108516770452e-05,
+      "loss": 0.6325,
+      "step": 2629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4847311135321772e-05,
+      "loss": 0.8408,
+      "step": 2630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.474670430283177e-05,
+      "loss": 0.8786,
+      "step": 2631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.464628812875087e-05,
+      "loss": 0.8391,
+      "step": 2632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4546062722321916e-05,
+      "loss": 0.7184,
+      "step": 2633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4446028192580394e-05,
+      "loss": 0.7733,
+      "step": 2634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.43461846483541e-05,
+      "loss": 0.6515,
+      "step": 2635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.424653219826294e-05,
+      "loss": 0.7552,
+      "step": 2636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4147070950719063e-05,
+      "loss": 0.6426,
+      "step": 2637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4047801013926587e-05,
+      "loss": 0.7569,
+      "step": 2638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3948722495881405e-05,
+      "loss": 0.7351,
+      "step": 2639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3849835504371253e-05,
+      "loss": 0.6693,
+      "step": 2640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.375114014697557e-05,
+      "loss": 0.8422,
+      "step": 2641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3652636531065087e-05,
+      "loss": 0.7859,
+      "step": 2642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3554324763802217e-05,
+      "loss": 0.7284,
+      "step": 2643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3456204952140427e-05,
+      "loss": 0.7431,
+      "step": 2644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.335827720282451e-05,
+      "loss": 0.9497,
+      "step": 2645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3260541622390263e-05,
+      "loss": 0.8366,
+      "step": 2646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3162998317164375e-05,
+      "loss": 0.7234,
+      "step": 2647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3065647393264445e-05,
+      "loss": 0.6935,
+      "step": 2648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2968488956598732e-05,
+      "loss": 0.7063,
+      "step": 2649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2871523112866065e-05,
+      "loss": 0.7007,
+      "step": 2650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2774749967555776e-05,
+      "loss": 0.8063,
+      "step": 2651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2678169625947642e-05,
+      "loss": 0.6874,
+      "step": 2652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.258178219311149e-05,
+      "loss": 0.7532,
+      "step": 2653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2485587773907456e-05,
+      "loss": 0.9417,
+      "step": 2654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.238958647298568e-05,
+      "loss": 0.7802,
+      "step": 2655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2293778394786102e-05,
+      "loss": 0.7979,
+      "step": 2656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2198163643538593e-05,
+      "loss": 0.7718,
+      "step": 2657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2102742323262527e-05,
+      "loss": 0.7489,
+      "step": 2658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2007514537767036e-05,
+      "loss": 0.6496,
+      "step": 2659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1912480390650615e-05,
+      "loss": 0.7426,
+      "step": 2660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.181763998530102e-05,
+      "loss": 0.786,
+      "step": 2661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.172299342489539e-05,
+      "loss": 0.688,
+      "step": 2662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1628540812399932e-05,
+      "loss": 0.7598,
+      "step": 2663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.153428225056975e-05,
+      "loss": 0.7187,
+      "step": 2664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.144021784194894e-05,
+      "loss": 0.8298,
+      "step": 2665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1346347688870427e-05,
+      "loss": 0.7964,
+      "step": 2666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1252671893455657e-05,
+      "loss": 0.8179,
+      "step": 2667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.115919055761475e-05,
+      "loss": 0.6362,
+      "step": 2668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1065903783046248e-05,
+      "loss": 0.8421,
+      "step": 2669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0972811671236992e-05,
+      "loss": 0.9378,
+      "step": 2670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0879914323462124e-05,
+      "loss": 0.764,
+      "step": 2671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0787211840784758e-05,
+      "loss": 0.8297,
+      "step": 2672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.069470432405618e-05,
+      "loss": 0.6596,
+      "step": 2673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0602391873915527e-05,
+      "loss": 0.7106,
+      "step": 2674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0510274590789626e-05,
+      "loss": 0.8399,
+      "step": 2675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0418352574893085e-05,
+      "loss": 0.6985,
+      "step": 2676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0326625926228073e-05,
+      "loss": 0.8158,
+      "step": 2677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.023509474458416e-05,
+      "loss": 0.7389,
+      "step": 2678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.0143759129538276e-05,
+      "loss": 0.8003,
+      "step": 2679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.005261918045471e-05,
+      "loss": 0.6648,
+      "step": 2680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9961674996484712e-05,
+      "loss": 0.7807,
+      "step": 2681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.987092667656665e-05,
+      "loss": 0.7351,
+      "step": 2682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9780374319425855e-05,
+      "loss": 0.684,
+      "step": 2683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9690018023574332e-05,
+      "loss": 0.7637,
+      "step": 2684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.959985788731098e-05,
+      "loss": 0.7536,
+      "step": 2685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9509894008721076e-05,
+      "loss": 0.7662,
+      "step": 2686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.942012648567657e-05,
+      "loss": 0.7288,
+      "step": 2687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.933055541583575e-05,
+      "loss": 0.7913,
+      "step": 2688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9241180896643108e-05,
+      "loss": 0.7387,
+      "step": 2689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.915200302532936e-05,
+      "loss": 0.671,
+      "step": 2690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9063021898911358e-05,
+      "loss": 0.8028,
+      "step": 2691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897423761419177e-05,
+      "loss": 0.7407,
+      "step": 2692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8885650267759215e-05,
+      "loss": 0.7259,
+      "step": 2693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.879725995598811e-05,
+      "loss": 0.7757,
+      "step": 2694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8709066775038342e-05,
+      "loss": 0.7903,
+      "step": 2695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8621070820855535e-05,
+      "loss": 0.8074,
+      "step": 2696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8533272189170603e-05,
+      "loss": 0.7514,
+      "step": 2697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8445670975499873e-05,
+      "loss": 0.7722,
+      "step": 2698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8358267275144914e-05,
+      "loss": 0.7032,
+      "step": 2699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8271061183192327e-05,
+      "loss": 0.6955,
+      "step": 2700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8184052794513828e-05,
+      "loss": 0.7279,
+      "step": 2701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8097242203766052e-05,
+      "loss": 0.7581,
+      "step": 2702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.801062950539032e-05,
+      "loss": 0.7845,
+      "step": 2703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7924214793612814e-05,
+      "loss": 0.7954,
+      "step": 2704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7837998162444313e-05,
+      "loss": 0.745,
+      "step": 2705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7751979705679965e-05,
+      "loss": 0.6844,
+      "step": 2706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7666159516899493e-05,
+      "loss": 0.7725,
+      "step": 2707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.758053768946686e-05,
+      "loss": 0.6831,
+      "step": 2708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7495114316530147e-05,
+      "loss": 0.7093,
+      "step": 2709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.740988949102169e-05,
+      "loss": 0.845,
+      "step": 2710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.732486330565768e-05,
+      "loss": 0.79,
+      "step": 2711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7240035852938295e-05,
+      "loss": 0.8422,
+      "step": 2712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.715540722514757e-05,
+      "loss": 0.7023,
+      "step": 2713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7070977514353026e-05,
+      "loss": 0.7613,
+      "step": 2714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6986746812405973e-05,
+      "loss": 0.715,
+      "step": 2715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.69027152109412e-05,
+      "loss": 0.798,
+      "step": 2716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.68188828013768e-05,
+      "loss": 0.6978,
+      "step": 2717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6735249674914198e-05,
+      "loss": 0.7669,
+      "step": 2718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6651815922538128e-05,
+      "loss": 0.8377,
+      "step": 2719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6568581635016244e-05,
+      "loss": 0.6955,
+      "step": 2720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.648554690289934e-05,
+      "loss": 0.8646,
+      "step": 2721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6402711816521087e-05,
+      "loss": 0.7053,
+      "step": 2722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.632007646599787e-05,
+      "loss": 0.583,
+      "step": 2723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6237640941228928e-05,
+      "loss": 0.7284,
+      "step": 2724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6155405331895947e-05,
+      "loss": 0.7219,
+      "step": 2725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6073369727463272e-05,
+      "loss": 0.837,
+      "step": 2726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.599153421717763e-05,
+      "loss": 0.8217,
+      "step": 2727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5909898890067933e-05,
+      "loss": 0.731,
+      "step": 2728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5828463834945494e-05,
+      "loss": 0.7494,
+      "step": 2729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5747229140403673e-05,
+      "loss": 0.6017,
+      "step": 2730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.566619489481782e-05,
+      "loss": 0.7396,
+      "step": 2731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5585361186345237e-05,
+      "loss": 0.7972,
+      "step": 2732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.550472810292516e-05,
+      "loss": 0.7543,
+      "step": 2733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5424295732278394e-05,
+      "loss": 0.7127,
+      "step": 2734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.53440641619075e-05,
+      "loss": 0.6433,
+      "step": 2735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.526403347909662e-05,
+      "loss": 0.7141,
+      "step": 2736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5184203770911187e-05,
+      "loss": 0.7608,
+      "step": 2737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5104575124198206e-05,
+      "loss": 0.8698,
+      "step": 2738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.5025147625585756e-05,
+      "loss": 0.7559,
+      "step": 2739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4945921361483205e-05,
+      "loss": 0.8184,
+      "step": 2740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.486689641808099e-05,
+      "loss": 0.9104,
+      "step": 2741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4788072881350445e-05,
+      "loss": 0.6659,
+      "step": 2742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4709450837043891e-05,
+      "loss": 0.7219,
+      "step": 2743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4631030370694443e-05,
+      "loss": 0.7149,
+      "step": 2744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4552811567615788e-05,
+      "loss": 0.711,
+      "step": 2745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.447479451290239e-05,
+      "loss": 0.8052,
+      "step": 2746
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4396979291429158e-05,
+      "loss": 0.752,
+      "step": 2747
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4319365987851375e-05,
+      "loss": 0.9884,
+      "step": 2748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.424195468660472e-05,
+      "loss": 0.758,
+      "step": 2749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4164745471905184e-05,
+      "loss": 0.8506,
+      "step": 2750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4087738427748709e-05,
+      "loss": 0.7583,
+      "step": 2751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4010933637911506e-05,
+      "loss": 0.8107,
+      "step": 2752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3934331185949578e-05,
+      "loss": 0.6961,
+      "step": 2753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.3857931155198933e-05,
+      "loss": 0.6346,
+      "step": 2754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3781733628775351e-05,
+      "loss": 0.7877,
+      "step": 2755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3705738689574166e-05,
+      "loss": 0.635,
+      "step": 2756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3705738689574166e-05,
+      "loss": 0.6888,
+      "step": 2757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3629946420270512e-05,
+      "loss": 0.6972,
+      "step": 2758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3554356903318944e-05,
+      "loss": 0.6939,
+      "step": 2759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3478970220953414e-05,
+      "loss": 0.8417,
+      "step": 2760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3403786455187229e-05,
+      "loss": 0.7858,
+      "step": 2761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3328805687813029e-05,
+      "loss": 0.7548,
+      "step": 2762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3254028000402475e-05,
+      "loss": 0.705,
+      "step": 2763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.317945347430638e-05,
+      "loss": 0.7373,
+      "step": 2764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.310508219065456e-05,
+      "loss": 0.6849,
+      "step": 2765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3030914230355607e-05,
+      "loss": 0.6639,
+      "step": 2766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2956949674097063e-05,
+      "loss": 0.7004,
+      "step": 2767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2883188602345054e-05,
+      "loss": 0.8325,
+      "step": 2768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2809631095344432e-05,
+      "loss": 0.7406,
+      "step": 2769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2736277233118565e-05,
+      "loss": 0.6547,
+      "step": 2770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.266312709546924e-05,
+      "loss": 0.682,
+      "step": 2771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2590180761976644e-05,
+      "loss": 0.6835,
+      "step": 2772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2517438311999274e-05,
+      "loss": 0.6788,
+      "step": 2773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2444899824673718e-05,
+      "loss": 0.7698,
+      "step": 2774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.237256537891478e-05,
+      "loss": 0.8825,
+      "step": 2775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.230043505341527e-05,
+      "loss": 0.7708,
+      "step": 2776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2228508926645865e-05,
+      "loss": 0.6802,
+      "step": 2777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2156787076855148e-05,
+      "loss": 0.7534,
+      "step": 2778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2085269582069525e-05,
+      "loss": 0.728,
+      "step": 2779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.201395652009294e-05,
+      "loss": 0.7524,
+      "step": 2780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1942847968507088e-05,
+      "loss": 0.7988,
+      "step": 2781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.187194400467102e-05,
+      "loss": 0.6562,
+      "step": 2782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1801244705721348e-05,
+      "loss": 0.7105,
+      "step": 2783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1730750148572011e-05,
+      "loss": 0.7364,
+      "step": 2784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1660460409914087e-05,
+      "loss": 0.6884,
+      "step": 2785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1590375566216006e-05,
+      "loss": 0.7498,
+      "step": 2786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.15204956937232e-05,
+      "loss": 0.717,
+      "step": 2787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1450820868458078e-05,
+      "loss": 0.7154,
+      "step": 2788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1381351166220055e-05,
+      "loss": 0.719,
+      "step": 2789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1312086662585363e-05,
+      "loss": 0.6832,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.124302743290695e-05,
+      "loss": 0.8042,
+      "step": 2791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.117417355231456e-05,
+      "loss": 0.6692,
+      "step": 2792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1105525095714387e-05,
+      "loss": 0.6078,
+      "step": 2793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1037082137789245e-05,
+      "loss": 0.7186,
+      "step": 2794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0968844752998442e-05,
+      "loss": 0.778,
+      "step": 2795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.090081301557746e-05,
+      "loss": 0.5901,
+      "step": 2796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0832986999538186e-05,
+      "loss": 0.7746,
+      "step": 2797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0765366778668751e-05,
+      "loss": 0.7524,
+      "step": 2798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0697952426533243e-05,
+      "loss": 0.8332,
+      "step": 2799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0630744016471882e-05,
+      "loss": 0.746,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0563741621600897e-05,
+      "loss": 0.7511,
+      "step": 2801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0496945314812245e-05,
+      "loss": 0.7337,
+      "step": 2802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0430355168773798e-05,
+      "loss": 0.7544,
+      "step": 2803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0363971255929116e-05,
+      "loss": 0.7786,
+      "step": 2804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0297793648497366e-05,
+      "loss": 0.7934,
+      "step": 2805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0231822418473335e-05,
+      "loss": 0.636,
+      "step": 2806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0166057637627214e-05,
+      "loss": 0.6152,
+      "step": 2807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.0100499377504658e-05,
+      "loss": 0.6337,
+      "step": 2808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.003514770942664e-05,
+      "loss": 0.6237,
+      "step": 2809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.970002704489333e-06,
+      "loss": 0.8105,
+      "step": 2810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.905064433564159e-06,
+      "loss": 0.754,
+      "step": 2811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.840332967297582e-06,
+      "loss": 0.7578,
+      "step": 2812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.775808376111073e-06,
+      "loss": 0.8027,
+      "step": 2813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.711490730201079e-06,
+      "loss": 0.7237,
+      "step": 2814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.647380099538894e-06,
+      "loss": 0.7703,
+      "step": 2815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.583476553870595e-06,
+      "loss": 0.7144,
+      "step": 2816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.519780162716974e-06,
+      "loss": 0.7356,
+      "step": 2817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.456290995373506e-06,
+      "loss": 0.6975,
+      "step": 2818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.393009120910102e-06,
+      "loss": 0.8105,
+      "step": 2819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.32993460817131e-06,
+      "loss": 0.6632,
+      "step": 2820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.267067525775974e-06,
+      "loss": 0.6174,
+      "step": 2821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.20440794211732e-06,
+      "loss": 0.7158,
+      "step": 2822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.141955925362888e-06,
+      "loss": 0.9003,
+      "step": 2823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.079711543454261e-06,
+      "loss": 0.7577,
+      "step": 2824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.017674864107296e-06,
+      "loss": 0.686,
+      "step": 2825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.955845954811826e-06,
+      "loss": 0.8301,
+      "step": 2826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.894224882831581e-06,
+      "loss": 0.5781,
+      "step": 2827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.832811715204291e-06,
+      "loss": 0.7833,
+      "step": 2828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.771606518741472e-06,
+      "loss": 0.6742,
+      "step": 2829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.710609360028343e-06,
+      "loss": 0.8088,
+      "step": 2830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.649820305423851e-06,
+      "loss": 0.8637,
+      "step": 2831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.58923942106058e-06,
+      "loss": 0.8269,
+      "step": 2832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.528866772844502e-06,
+      "loss": 0.684,
+      "step": 2833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.46870242645521e-06,
+      "loss": 0.7355,
+      "step": 2834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.408746447345595e-06,
+      "loss": 0.7199,
+      "step": 2835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.348998900741855e-06,
+      "loss": 0.6295,
+      "step": 2836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.289459851643533e-06,
+      "loss": 0.7145,
+      "step": 2837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.230129364823213e-06,
+      "loss": 0.6621,
+      "step": 2838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.17100750482669e-06,
+      "loss": 0.6925,
+      "step": 2839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.11209433597273e-06,
+      "loss": 0.7408,
+      "step": 2840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.053389922353094e-06,
+      "loss": 0.7056,
+      "step": 2841
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.994894327832448e-06,
+      "loss": 0.6488,
+      "step": 2842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.936607616048242e-06,
+      "loss": 0.7233,
+      "step": 2843
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.87852985041071e-06,
+      "loss": 0.6948,
+      "step": 2844
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.820661094102776e-06,
+      "loss": 0.7529,
+      "step": 2845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.763001410079995e-06,
+      "loss": 0.6462,
+      "step": 2846
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.705550861070387e-06,
+      "loss": 0.7063,
+      "step": 2847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.648309509574603e-06,
+      "loss": 0.6789,
+      "step": 2848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.591277417865561e-06,
+      "loss": 0.7821,
+      "step": 2849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.5344546479885865e-06,
+      "loss": 0.774,
+      "step": 2850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.477841261761342e-06,
+      "loss": 0.604,
+      "step": 2851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.421437320773583e-06,
+      "loss": 0.747,
+      "step": 2852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.365242886387269e-06,
+      "loss": 0.7588,
+      "step": 2853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.309258019736498e-06,
+      "loss": 0.7603,
+      "step": 2854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.253482781727261e-06,
+      "loss": 0.6939,
+      "step": 2855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.197917233037576e-06,
+      "loss": 0.7547,
+      "step": 2856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.142561434117312e-06,
+      "loss": 0.92,
+      "step": 2857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.087415445188139e-06,
+      "loss": 0.8132,
+      "step": 2858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.03247932624349e-06,
+      "loss": 0.9267,
+      "step": 2859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.977753137048493e-06,
+      "loss": 0.8301,
+      "step": 2860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.923236937139854e-06,
+      "loss": 0.683,
+      "step": 2861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.868930785825911e-06,
+      "loss": 0.729,
+      "step": 2862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.81483474218636e-06,
+      "loss": 0.8855,
+      "step": 2863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.760948865072414e-06,
+      "loss": 0.8312,
+      "step": 2864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.707273213106691e-06,
+      "loss": 0.7771,
+      "step": 2865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.653807844682946e-06,
+      "loss": 0.6131,
+      "step": 2866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.600552817966299e-06,
+      "loss": 0.759,
+      "step": 2867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.547508190893048e-06,
+      "loss": 0.7735,
+      "step": 2868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.494674021170455e-06,
+      "loss": 0.7278,
+      "step": 2869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.442050366276986e-06,
+      "loss": 0.828,
+      "step": 2870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.389637283462025e-06,
+      "loss": 0.8339,
+      "step": 2871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.337434829745803e-06,
+      "loss": 0.7305,
+      "step": 2872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.285443061919538e-06,
+      "loss": 0.6491,
+      "step": 2873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.233662036545162e-06,
+      "loss": 0.698,
+      "step": 2874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.182091809955326e-06,
+      "loss": 0.8047,
+      "step": 2875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.130732438253417e-06,
+      "loss": 0.6479,
+      "step": 2876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.079583977313385e-06,
+      "loss": 0.7469,
+      "step": 2877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.028646482779721e-06,
+      "loss": 0.7269,
+      "step": 2878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.9779200100674725e-06,
+      "loss": 0.6885,
+      "step": 2879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.9274046143620306e-06,
+      "loss": 0.7527,
+      "step": 2880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.877100350619213e-06,
+      "loss": 0.8146,
+      "step": 2881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.827007273565155e-06,
+      "loss": 0.6241,
+      "step": 2882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.777125437696152e-06,
+      "loss": 0.6438,
+      "step": 2883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.727454897278772e-06,
+      "loss": 0.8498,
+      "step": 2884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.677995706349748e-06,
+      "loss": 0.719,
+      "step": 2885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.628747918715771e-06,
+      "loss": 0.6997,
+      "step": 2886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.579711587953651e-06,
+      "loss": 0.8262,
+      "step": 2887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.5308867674100485e-06,
+      "loss": 0.7829,
+      "step": 2888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.48227351020163e-06,
+      "loss": 0.7693,
+      "step": 2889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.433871869214869e-06,
+      "loss": 0.7722,
+      "step": 2890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.385681897105954e-06,
+      "loss": 0.6862,
+      "step": 2891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.337703646300906e-06,
+      "loss": 0.8832,
+      "step": 2892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.289937168995329e-06,
+      "loss": 0.6669,
+      "step": 2893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.242382517154476e-06,
+      "loss": 0.7948,
+      "step": 2894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.195039742513142e-06,
+      "loss": 0.7227,
+      "step": 2895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.147908896575637e-06,
+      "loss": 0.6939,
+      "step": 2896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.100990030615682e-06,
+      "loss": 0.9131,
+      "step": 2897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.054283195676424e-06,
+      "loss": 0.7396,
+      "step": 2898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.007788442570327e-06,
+      "loss": 0.7201,
+      "step": 2899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.961505821879109e-06,
+      "loss": 0.7767,
+      "step": 2900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.915435383953737e-06,
+      "loss": 0.7749,
+      "step": 2901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8695771789142975e-06,
+      "loss": 0.7465,
+      "step": 2902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.823931256650061e-06,
+      "loss": 0.7334,
+      "step": 2903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.778497666819326e-06,
+      "loss": 0.737,
+      "step": 2904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.7332764588493335e-06,
+      "loss": 0.7106,
+      "step": 2905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.688267681936376e-06,
+      "loss": 0.6957,
+      "step": 2906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6434713850455945e-06,
+      "loss": 0.6438,
+      "step": 2907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.598887616910919e-06,
+      "loss": 0.8329,
+      "step": 2908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.554516426035149e-06,
+      "loss": 0.7842,
+      "step": 2909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.510357860689807e-06,
+      "loss": 0.7081,
+      "step": 2910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.466411968915063e-06,
+      "loss": 0.7109,
+      "step": 2911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4226787985197195e-06,
+      "loss": 0.6963,
+      "step": 2912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.379158397081251e-06,
+      "loss": 0.7161,
+      "step": 2913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.335850811945519e-06,
+      "loss": 0.717,
+      "step": 2914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.292756090226968e-06,
+      "loss": 0.6916,
+      "step": 2915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.249874278808408e-06,
+      "loss": 0.7101,
+      "step": 2916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.207205424341054e-06,
+      "loss": 0.7827,
+      "step": 2917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.164749573244464e-06,
+      "loss": 0.7805,
+      "step": 2918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.122506771706381e-06,
+      "loss": 0.8507,
+      "step": 2919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.080477065682909e-06,
+      "loss": 0.7335,
+      "step": 2920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.038660500898206e-06,
+      "loss": 0.7126,
+      "step": 2921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.997057122844594e-06,
+      "loss": 0.7367,
+      "step": 2922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9556669767824684e-06,
+      "loss": 0.7535,
+      "step": 2923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9144901077402985e-06,
+      "loss": 0.7832,
+      "step": 2924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8735265605144065e-06,
+      "loss": 0.7158,
+      "step": 2925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.832776379669167e-06,
+      "loss": 0.762,
+      "step": 2926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.792239609536785e-06,
+      "loss": 0.7639,
+      "step": 2927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7519162942172725e-06,
+      "loss": 0.8639,
+      "step": 2928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.71180647757845e-06,
+      "loss": 0.7608,
+      "step": 2929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.6719102032558796e-06,
+      "loss": 0.7382,
+      "step": 2930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.632227514652775e-06,
+      "loss": 0.773,
+      "step": 2931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.592758454940048e-06,
+      "loss": 0.7745,
+      "step": 2932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5535030670561076e-06,
+      "loss": 0.7361,
+      "step": 2933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5144613937070358e-06,
+      "loss": 0.7479,
+      "step": 2934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4756334773663024e-06,
+      "loss": 0.7501,
+      "step": 2935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4370193602748734e-06,
+      "loss": 0.7555,
+      "step": 2936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.398619084441168e-06,
+      "loss": 0.7445,
+      "step": 2937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.36043269164088e-06,
+      "loss": 0.7547,
+      "step": 2938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3224602234170678e-06,
+      "loss": 0.6787,
+      "step": 2939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2847017210800635e-06,
+      "loss": 0.8066,
+      "step": 2940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.247157225707409e-06,
+      "loss": 0.701,
+      "step": 2941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.209826778143854e-06,
+      "loss": 0.7641,
+      "step": 2942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.172710419001268e-06,
+      "loss": 0.7422,
+      "step": 2943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1358081886585734e-06,
+      "loss": 0.7432,
+      "step": 2944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.099120127261812e-06,
+      "loss": 0.7321,
+      "step": 2945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0626462747239905e-06,
+      "loss": 0.6615,
+      "step": 2946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.026386670725079e-06,
+      "loss": 0.7409,
+      "step": 2947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.990341354712012e-06,
+      "loss": 0.7575,
+      "step": 2948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9545103658984886e-06,
+      "loss": 0.773,
+      "step": 2949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.918893743265172e-06,
+      "loss": 0.6237,
+      "step": 2950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8834915255594454e-06,
+      "loss": 0.6966,
+      "step": 2951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8483037512954337e-06,
+      "loss": 0.7088,
+      "step": 2952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8133304587540046e-06,
+      "loss": 0.7964,
+      "step": 2953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.778571685982678e-06,
+      "loss": 0.8558,
+      "step": 2954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.744027470795563e-06,
+      "loss": 0.7191,
+      "step": 2955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7096978507733963e-06,
+      "loss": 0.6743,
+      "step": 2956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.675582863263437e-06,
+      "loss": 0.8065,
+      "step": 2957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6416825453794646e-06,
+      "loss": 0.7158,
+      "step": 2958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6079969340016885e-06,
+      "loss": 0.7354,
+      "step": 2959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.57452606577675e-06,
+      "loss": 0.8222,
+      "step": 2960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.541269977117655e-06,
+      "loss": 0.6176,
+      "step": 2961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.508228704203819e-06,
+      "loss": 0.6959,
+      "step": 2962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4754022829808655e-06,
+      "loss": 0.813,
+      "step": 2963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4427907491607172e-06,
+      "loss": 0.7808,
+      "step": 2964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4103941382215944e-06,
+      "loss": 0.7093,
+      "step": 2965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.378212485407749e-06,
+      "loss": 0.7593,
+      "step": 2966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3462458257297093e-06,
+      "loss": 0.6787,
+      "step": 2967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3144941939640565e-06,
+      "loss": 0.6651,
+      "step": 2968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2829576246534703e-06,
+      "loss": 0.7632,
+      "step": 2969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2516361521066178e-06,
+      "loss": 0.701,
+      "step": 2970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.220529810398242e-06,
+      "loss": 0.7656,
+      "step": 2971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.18963863336894e-06,
+      "loss": 0.7704,
+      "step": 2972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1589626546253185e-06,
+      "loss": 0.7831,
+      "step": 2973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1285019075397928e-06,
+      "loss": 0.7663,
+      "step": 2974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.098256425250722e-06,
+      "loss": 0.6898,
+      "step": 2975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0682262406622077e-06,
+      "loss": 0.706,
+      "step": 2976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.038411386444139e-06,
+      "loss": 0.6948,
+      "step": 2977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.00881189503217e-06,
+      "loss": 0.6968,
+      "step": 2978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.979427798627631e-06,
+      "loss": 0.7704,
+      "step": 2979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.950259129197529e-06,
+      "loss": 0.7231,
+      "step": 2980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9213059184745475e-06,
+      "loss": 0.8018,
+      "step": 2981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8925681979569342e-06,
+      "loss": 0.6557,
+      "step": 2982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8640459989085035e-06,
+      "loss": 0.8314,
+      "step": 2983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.835739352358612e-06,
+      "loss": 0.9106,
+      "step": 2984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8076482891021374e-06,
+      "loss": 0.7794,
+      "step": 2985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7797728396993673e-06,
+      "loss": 0.7908,
+      "step": 2986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7521130344760884e-06,
+      "loss": 0.7695,
+      "step": 2987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7246689035234299e-06,
+      "loss": 0.798,
+      "step": 2988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6974404766979313e-06,
+      "loss": 0.6632,
+      "step": 2989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6704277836214755e-06,
+      "loss": 0.7943,
+      "step": 2990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6436308536811996e-06,
+      "loss": 0.6574,
+      "step": 2991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6170497160295173e-06,
+      "loss": 0.7416,
+      "step": 2992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906843995841636e-06,
+      "loss": 0.6875,
+      "step": 2993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5645349330279502e-06,
+      "loss": 0.7428,
+      "step": 2994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5386013448089654e-06,
+      "loss": 0.7972,
+      "step": 2995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5128836631404187e-06,
+      "loss": 0.6389,
+      "step": 2996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4873819160006185e-06,
+      "loss": 0.7321,
+      "step": 2997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.462096131132973e-06,
+      "loss": 0.7158,
+      "step": 2998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4370263360459213e-06,
+      "loss": 0.9534,
+      "step": 2999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4121725580129585e-06,
+      "loss": 0.7404,
+      "step": 3000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3112,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 7.927776730331218e+17,
+  "trial_name": null,
+  "trial_params": null
+}