diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,29587 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 4927,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.7567567567567575e-06,
+      "loss": 8.3828,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 8.4141,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0270270270270273e-05,
+      "loss": 8.0742,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-05,
+      "loss": 6.5039,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3783783783783784e-05,
+      "loss": 5.9336,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0540540540540545e-05,
+      "loss": 5.7773,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.72972972972973e-05,
+      "loss": 5.5625,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-05,
+      "loss": 5.418,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.0810810810810814e-05,
+      "loss": 5.3477,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.756756756756757e-05,
+      "loss": 5.2227,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.432432432432433e-05,
+      "loss": 5.1172,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.108108108108109e-05,
+      "loss": 5.0039,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.783783783783784e-05,
+      "loss": 4.8867,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.45945945945946e-05,
+      "loss": 4.793,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010135135135135136,
+      "loss": 4.7617,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010810810810810812,
+      "loss": 4.668,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011486486486486487,
+      "loss": 4.5,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012162162162162163,
+      "loss": 4.5703,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012837837837837836,
+      "loss": 4.4297,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013513513513513514,
+      "loss": 4.3594,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014189189189189188,
+      "loss": 4.3672,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014864864864864866,
+      "loss": 4.2012,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001554054054054054,
+      "loss": 4.1562,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016216216216216218,
+      "loss": 4.082,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016891891891891893,
+      "loss": 4.0234,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017567567567567568,
+      "loss": 3.9902,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018243243243243242,
+      "loss": 3.8652,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001891891891891892,
+      "loss": 3.8164,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019594594594594594,
+      "loss": 3.6934,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020270270270270272,
+      "loss": 3.5957,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020945945945945947,
+      "loss": 3.7227,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021621621621621624,
+      "loss": 3.5996,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000222972972972973,
+      "loss": 3.5293,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022972972972972974,
+      "loss": 3.5391,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023648648648648648,
+      "loss": 3.5,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024324324324324326,
+      "loss": 3.2871,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025,
+      "loss": 3.4434,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002567567567567567,
+      "loss": 3.3789,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002635135135135135,
+      "loss": 3.3965,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002702702702702703,
+      "loss": 3.3652,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027702702702702705,
+      "loss": 3.3418,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028378378378378377,
+      "loss": 3.2324,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029054054054054054,
+      "loss": 3.1895,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002972972972972973,
+      "loss": 3.1895,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030405405405405404,
+      "loss": 3.2168,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003108108108108108,
+      "loss": 3.1504,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031756756756756753,
+      "loss": 3.2188,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032432432432432436,
+      "loss": 3.207,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003310810810810811,
+      "loss": 3.1289,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00033783783783783786,
+      "loss": 3.0703,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003445945945945946,
+      "loss": 3.0898,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035135135135135135,
+      "loss": 3.1211,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003581081081081081,
+      "loss": 3.0371,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036486486486486485,
+      "loss": 3.1016,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003716216216216216,
+      "loss": 3.0625,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003783783783783784,
+      "loss": 3.0195,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038513513513513517,
+      "loss": 3.0039,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003918918918918919,
+      "loss": 2.9277,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039864864864864866,
+      "loss": 2.9668,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040540540540540544,
+      "loss": 2.998,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00041216216216216216,
+      "loss": 3.0215,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00041891891891891893,
+      "loss": 3.084,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042567567567567565,
+      "loss": 2.9883,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004324324324324325,
+      "loss": 2.8887,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004391891891891892,
+      "loss": 3.0293,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000445945945945946,
+      "loss": 2.8965,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004527027027027027,
+      "loss": 2.8457,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045945945945945947,
+      "loss": 2.9082,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046621621621621625,
+      "loss": 2.9062,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047297297297297297,
+      "loss": 2.998,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047972972972972974,
+      "loss": 2.918,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004864864864864865,
+      "loss": 2.8926,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004932432432432432,
+      "loss": 2.8945,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 2.8789,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005067567567567568,
+      "loss": 2.8145,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005135135135135135,
+      "loss": 2.8945,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005202702702702703,
+      "loss": 2.8828,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000527027027027027,
+      "loss": 2.8711,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005337837837837838,
+      "loss": 2.8926,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005405405405405405,
+      "loss": 2.877,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005472972972972973,
+      "loss": 2.9023,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005540540540540541,
+      "loss": 2.9434,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005608108108108109,
+      "loss": 2.8262,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005675675675675675,
+      "loss": 2.916,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005743243243243243,
+      "loss": 2.8281,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005810810810810811,
+      "loss": 2.832,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005878378378378379,
+      "loss": 2.8262,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005945945945945946,
+      "loss": 2.7988,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006013513513513513,
+      "loss": 2.7969,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006081081081081081,
+      "loss": 2.8555,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000614864864864865,
+      "loss": 2.8887,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006216216216216216,
+      "loss": 2.8184,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006283783783783784,
+      "loss": 2.8516,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006351351351351351,
+      "loss": 2.9199,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000641891891891892,
+      "loss": 2.8496,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006486486486486487,
+      "loss": 2.8203,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006554054054054054,
+      "loss": 2.8359,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006621621621621622,
+      "loss": 2.8086,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006689189189189189,
+      "loss": 2.8613,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006756756756756757,
+      "loss": 2.9062,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006824324324324325,
+      "loss": 2.8594,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006891891891891892,
+      "loss": 2.8477,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006959459459459459,
+      "loss": 2.7422,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007027027027027027,
+      "loss": 2.8516,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007094594594594595,
+      "loss": 2.7891,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007162162162162163,
+      "loss": 2.8535,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000722972972972973,
+      "loss": 2.793,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007297297297297297,
+      "loss": 2.7754,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007364864864864866,
+      "loss": 2.7773,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007432432432432432,
+      "loss": 2.7773,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00075,
+      "loss": 2.8105,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007567567567567568,
+      "loss": 2.7266,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007635135135135135,
+      "loss": 2.7441,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007702702702702703,
+      "loss": 2.7598,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000777027027027027,
+      "loss": 2.7578,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007837837837837838,
+      "loss": 2.7891,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007905405405405406,
+      "loss": 2.7363,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007972972972972973,
+      "loss": 2.7109,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008040540540540541,
+      "loss": 2.6582,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008108108108108109,
+      "loss": 2.6191,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008175675675675675,
+      "loss": 2.7969,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008243243243243243,
+      "loss": 2.6816,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008310810810810811,
+      "loss": 2.6738,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008378378378378379,
+      "loss": 2.7422,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008445945945945946,
+      "loss": 2.7754,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008513513513513513,
+      "loss": 2.7754,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008581081081081081,
+      "loss": 2.7363,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000864864864864865,
+      "loss": 2.7129,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008716216216216216,
+      "loss": 2.7617,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008783783783783784,
+      "loss": 2.6523,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008851351351351351,
+      "loss": 2.6895,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000891891891891892,
+      "loss": 2.7559,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008986486486486487,
+      "loss": 2.6621,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009054054054054054,
+      "loss": 2.7109,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009121621621621622,
+      "loss": 2.7344,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009189189189189189,
+      "loss": 2.6172,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009256756756756757,
+      "loss": 2.6523,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009324324324324325,
+      "loss": 2.6113,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009391891891891892,
+      "loss": 2.7227,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009459459459459459,
+      "loss": 2.6836,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009527027027027027,
+      "loss": 2.6348,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009594594594594595,
+      "loss": 2.6953,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009662162162162163,
+      "loss": 2.6699,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000972972972972973,
+      "loss": 2.7266,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009797297297297297,
+      "loss": 2.6309,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009864864864864865,
+      "loss": 2.6719,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009932432432432432,
+      "loss": 2.6602,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 2.5957,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999998919647012,
+      "loss": 2.6289,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999995678588516,
+      "loss": 2.6777,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999999027682591,
+      "loss": 2.5996,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999998271436153,
+      "loss": 2.6289,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999972991198646,
+      "loss": 2.6875,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999961107341458,
+      "loss": 2.6055,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999947062795098,
+      "loss": 2.7168,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999930857565642,
+      "loss": 2.6406,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999912491660088,
+      "loss": 2.6055,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999891965086374,
+      "loss": 2.5625,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999986927785337,
+      "loss": 2.6055,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999844429970884,
+      "loss": 2.5684,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999817421449649,
+      "loss": 2.6328,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999788252301337,
+      "loss": 2.5801,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999756922538553,
+      "loss": 2.6172,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999972343217484,
+      "loss": 2.5938,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999687781224664,
+      "loss": 2.5703,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999649969703436,
+      "loss": 2.6836,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999609997627496,
+      "loss": 2.5938,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999567865014117,
+      "loss": 2.5781,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999523571881503,
+      "loss": 2.5469,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999477118248798,
+      "loss": 2.5684,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999428504136078,
+      "loss": 2.6465,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999377729564348,
+      "loss": 2.5332,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999324794555552,
+      "loss": 2.5508,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999926969913256,
+      "loss": 2.5645,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999921244331919,
+      "loss": 2.5801,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999153027140178,
+      "loss": 2.5137,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999091450621203,
+      "loss": 2.5273,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999027713788872,
+      "loss": 2.5508,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998961816670732,
+      "loss": 2.6016,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998893759295257,
+      "loss": 2.5508,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998823541691858,
+      "loss": 2.5371,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998751163890882,
+      "loss": 2.502,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00099986766259236,
+      "loss": 2.5781,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999859992782223,
+      "loss": 2.4941,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998521069619912,
+      "loss": 2.4473,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998440051350724,
+      "loss": 2.5234,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998356873049678,
+      "loss": 2.5625,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999827153475272,
+      "loss": 2.4727,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999818403649673,
+      "loss": 2.4863,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998094378319514,
+      "loss": 2.5352,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999800256025982,
+      "loss": 2.5449,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997908582357324,
+      "loss": 2.6094,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997812444652644,
+      "loss": 2.5039,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999771414718732,
+      "loss": 2.5547,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999761369000383,
+      "loss": 2.5469,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997511073145586,
+      "loss": 2.5371,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997406296656938,
+      "loss": 2.5039,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997299360583157,
+      "loss": 2.5449,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999719026497046,
+      "loss": 2.5,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997079009865988,
+      "loss": 2.459,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999696559531782,
+      "loss": 2.4961,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996850021374967,
+      "loss": 2.541,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996732288087375,
+      "loss": 2.5117,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996612395505922,
+      "loss": 2.4629,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996490343682413,
+      "loss": 2.4219,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00099963661326696,
+      "loss": 2.4922,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999623976252115,
+      "loss": 2.5449,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996111233291681,
+      "loss": 2.4512,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999598054503673,
+      "loss": 2.5371,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995847697812778,
+      "loss": 2.4766,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999571269167723,
+      "loss": 2.4707,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995575526688426,
+      "loss": 2.541,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995436202905649,
+      "loss": 2.4922,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995294720389096,
+      "loss": 2.5156,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995151079199916,
+      "loss": 2.502,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995005279400176,
+      "loss": 2.4258,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994857321052885,
+      "loss": 2.4902,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994707204221984,
+      "loss": 2.4473,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999455492897234,
+      "loss": 2.4785,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999440049536976,
+      "loss": 2.459,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999424390348098,
+      "loss": 2.5645,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994085153373674,
+      "loss": 2.4902,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999392424511644,
+      "loss": 2.4883,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993761178778812,
+      "loss": 2.4316,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993595954431262,
+      "loss": 2.5195,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993428572145186,
+      "loss": 2.3906,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999325903199292,
+      "loss": 2.4023,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999308733404773,
+      "loss": 2.4102,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999291347838381,
+      "loss": 2.5625,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992737465076293,
+      "loss": 2.4707,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999255929420124,
+      "loss": 2.4961,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992378965835646,
+      "loss": 2.4102,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999219648005744,
+      "loss": 2.459,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992011836945482,
+      "loss": 2.4434,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999182503657956,
+      "loss": 2.4551,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991636079040402,
+      "loss": 2.3945,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991444964409664,
+      "loss": 2.4277,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991251692769933,
+      "loss": 2.4668,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999105626420473,
+      "loss": 2.5176,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990858678798507,
+      "loss": 2.4609,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999065893663665,
+      "loss": 2.3457,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990457037805476,
+      "loss": 2.3809,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990252982392234,
+      "loss": 2.4883,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990046770485103,
+      "loss": 2.4648,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989838402173197,
+      "loss": 2.4141,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998962787754656,
+      "loss": 2.4395,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998941519669617,
+      "loss": 2.4766,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989200359713931,
+      "loss": 2.4668,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988983366692689,
+      "loss": 2.4316,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988764217726208,
+      "loss": 2.3633,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099885429129092,
+      "loss": 2.4277,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988319452337293,
+      "loss": 2.4297,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988093836107057,
+      "loss": 2.4238,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998786606431599,
+      "loss": 2.4219,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998763613706252,
+      "loss": 2.4277,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987404054446008,
+      "loss": 2.3789,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987169816566748,
+      "loss": 2.4551,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998693342352596,
+      "loss": 2.4258,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986694875425807,
+      "loss": 2.3887,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986454172369369,
+      "loss": 2.3789,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986211314460664,
+      "loss": 2.3809,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985966301804643,
+      "loss": 2.3555,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985719134507185,
+      "loss": 2.4219,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985469812675103,
+      "loss": 2.4043,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985218336416137,
+      "loss": 2.4082,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998496470583896,
+      "loss": 2.459,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984708921053178,
+      "loss": 2.4297,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984450982169326,
+      "loss": 2.4922,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984190889298868,
+      "loss": 2.4043,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009983928642554203,
+      "loss": 2.4238,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983664242048658,
+      "loss": 2.4062,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983397687896491,
+      "loss": 2.4199,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983128980212892,
+      "loss": 2.4453,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998285811911398,
+      "loss": 2.457,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982585104716805,
+      "loss": 2.4492,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982309937139344,
+      "loss": 2.4512,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982032616500517,
+      "loss": 2.459,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981753142920158,
+      "loss": 2.4121,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981471516519044,
+      "loss": 2.4258,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981187737418873,
+      "loss": 2.4316,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980901805742282,
+      "loss": 2.4824,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998061372161283,
+      "loss": 2.4023,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980323485155012,
+      "loss": 2.3711,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980031096494253,
+      "loss": 2.3418,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979736555756902,
+      "loss": 2.3652,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979439863070244,
+      "loss": 2.4023,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979141018562495,
+      "loss": 2.4434,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978840022362792,
+      "loss": 2.4824,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978536874601213,
+      "loss": 2.4648,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978231575408759,
+      "loss": 2.332,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997792412491736,
+      "loss": 2.4219,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977614523259883,
+      "loss": 2.334,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977302770570115,
+      "loss": 2.3633,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976988866982782,
+      "loss": 2.4824,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997667281263353,
+      "loss": 2.4375,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997635460765894,
+      "loss": 2.4434,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976034252196524,
+      "loss": 2.3633,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975711746384717,
+      "loss": 2.334,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975387090362892,
+      "loss": 2.3633,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997506028427134,
+      "loss": 2.4043,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974731328251294,
+      "loss": 2.4551,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974400222444904,
+      "loss": 2.4473,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974066966995257,
+      "loss": 2.3496,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973731562046366,
+      "loss": 2.4082,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973394007743175,
+      "loss": 2.3398,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973054304231552,
+      "loss": 2.4727,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099727124516583,
+      "loss": 2.4121,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972368450171145,
+      "loss": 2.4414,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972022299918745,
+      "loss": 2.3867,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971674001050686,
+      "loss": 2.4062,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971323553717483,
+      "loss": 2.3535,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970970958070576,
+      "loss": 2.3496,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997061621426234,
+      "loss": 2.3711,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970259322446072,
+      "loss": 2.4238,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969900282776,
+      "loss": 2.3379,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996953909540728,
+      "loss": 2.3867,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969175760495997,
+      "loss": 2.3867,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996881027819916,
+      "loss": 2.375,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968442648674713,
+      "loss": 2.4141,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996807287208152,
+      "loss": 2.4316,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967700948579378,
+      "loss": 2.459,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967326878329015,
+      "loss": 2.4336,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966950661492073,
+      "loss": 2.3457,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966572298231139,
+      "loss": 2.3555,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966191788709714,
+      "loss": 2.3379,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965809133092236,
+      "loss": 2.3789,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965424331544064,
+      "loss": 2.3633,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965037384231487,
+      "loss": 2.3398,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996464829132172,
+      "loss": 2.3887,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964257052982907,
+      "loss": 2.4199,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963863669384119,
+      "loss": 2.4512,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996346814069535,
+      "loss": 2.3711,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963070467087528,
+      "loss": 2.4238,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962670648732503,
+      "loss": 2.3984,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962268685803053,
+      "loss": 2.3594,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961864578472883,
+      "loss": 2.3809,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961458326916623,
+      "loss": 2.3086,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996104993130983,
+      "loss": 2.3711,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960639391828994,
+      "loss": 2.3672,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960226708651521,
+      "loss": 2.4043,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995981188195575,
+      "loss": 2.3828,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959394911920943,
+      "loss": 2.334,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958975798727294,
+      "loss": 2.3262,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958554542555916,
+      "loss": 2.3984,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995813114358885,
+      "loss": 2.3945,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957705602009067,
+      "loss": 2.4023,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995727791800046,
+      "loss": 2.3379,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956848091747848,
+      "loss": 2.3613,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995641612343698,
+      "loss": 2.4043,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955982013254523,
+      "loss": 2.457,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955545761388073,
+      "loss": 2.3184,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995510736802616,
+      "loss": 2.3926,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954666833358224,
+      "loss": 2.4277,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995422415757464,
+      "loss": 2.332,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995377934086671,
+      "loss": 2.3223,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953332383426654,
+      "loss": 2.3457,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952883285447623,
+      "loss": 2.3887,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952432047123688,
+      "loss": 2.3926,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995197866864985,
+      "loss": 2.3809,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995152315022203,
+      "loss": 2.3398,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995106549203708,
+      "loss": 2.3438,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995060569429277,
+      "loss": 2.3848,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950143757187799,
+      "loss": 2.3438,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994967968092179,
+      "loss": 2.3008,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949213465695287,
+      "loss": 2.3066,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994874511170976,
+      "loss": 2.3418,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948274619167607,
+      "loss": 2.5137,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947801988272147,
+      "loss": 2.2285,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947327219227621,
+      "loss": 2.4102,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946850312239199,
+      "loss": 2.3301,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946371267512972,
+      "loss": 2.3398,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994589008525595,
+      "loss": 2.3418,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945406765676078,
+      "loss": 2.3398,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944921308982218,
+      "loss": 2.3066,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994443371538415,
+      "loss": 2.291,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943943985092588,
+      "loss": 2.3242,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943452118319164,
+      "loss": 2.2852,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942958115276433,
+      "loss": 2.334,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942461976177874,
+      "loss": 2.3008,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994196370123789,
+      "loss": 2.3145,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941463290671806,
+      "loss": 2.3496,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940960744695868,
+      "loss": 2.3184,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994045606352725,
+      "loss": 2.3535,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939949247384046,
+      "loss": 2.3281,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939440296485268,
+      "loss": 2.3184,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938929211050856,
+      "loss": 2.3984,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938415991301674,
+      "loss": 2.2988,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937900637459502,
+      "loss": 2.291,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937383149747048,
+      "loss": 2.3164,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993686352838794,
+      "loss": 2.2695,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936341773606723,
+      "loss": 2.2578,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935817885628874,
+      "loss": 2.3574,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935291864680783,
+      "loss": 2.3047,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934763710989771,
+      "loss": 2.2793,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993423342478407,
+      "loss": 2.3105,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933701006292837,
+      "loss": 2.3262,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933166455746158,
+      "loss": 2.3613,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932629773375028,
+      "loss": 2.2852,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932090959411375,
+      "loss": 2.3262,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993155001408804,
+      "loss": 2.3848,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931006937638786,
+      "loss": 2.2871,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099304617302983,
+      "loss": 2.375,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929914392302192,
+      "loss": 2.3398,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929364923886986,
+      "loss": 2.2695,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992881332529013,
+      "loss": 2.373,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928259596749993,
+      "loss": 2.3828,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927703738505861,
+      "loss": 2.2715,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992714575079795,
+      "loss": 2.2832,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926585633867385,
+      "loss": 2.2715,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926023387956215,
+      "loss": 2.3086,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925459013307412,
+      "loss": 2.3242,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924892510164863,
+      "loss": 2.3301,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924323878773378,
+      "loss": 2.3105,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992375311937869,
+      "loss": 2.2969,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992318023222744,
+      "loss": 2.3613,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922605217567203,
+      "loss": 2.3242,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922028075646464,
+      "loss": 2.3457,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992144880671463,
+      "loss": 2.2715,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920867411022025,
+      "loss": 2.3359,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920283888819898,
+      "loss": 2.3457,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919698240360409,
+      "loss": 2.2637,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919110465896643,
+      "loss": 2.2676,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918520565682604,
+      "loss": 2.4023,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917928539973206,
+      "loss": 2.3242,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917334389024292,
+      "loss": 2.3164,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916738113092617,
+      "loss": 2.2578,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991613971243586,
+      "loss": 2.3691,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991553918731261,
+      "loss": 2.332,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914936537982382,
+      "loss": 2.3691,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914331764705605,
+      "loss": 2.2793,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913724867743623,
+      "loss": 2.2539,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913115847358708,
+      "loss": 2.3457,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912504703814037,
+      "loss": 2.3398,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991189143737371,
+      "loss": 2.3164,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991127604830275,
+      "loss": 2.3613,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910658536867085,
+      "loss": 2.2715,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910038903333575,
+      "loss": 2.2871,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990941714796998,
+      "loss": 2.3203,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908793271044996,
+      "loss": 2.3457,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990816727282822,
+      "loss": 2.3418,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907539153590173,
+      "loss": 2.293,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990690891360229,
+      "loss": 2.1924,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906276553136924,
+      "loss": 2.3984,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905642072467345,
+      "loss": 2.3496,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905005471867738,
+      "loss": 2.3301,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904366751613204,
+      "loss": 2.3223,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990372591197976,
+      "loss": 2.3086,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990308295324434,
+      "loss": 2.2871,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902437875684794,
+      "loss": 2.3242,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901790679579883,
+      "loss": 2.3301,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901141365209292,
+      "loss": 2.3281,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900489932853612,
+      "loss": 2.3203,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899836382794357,
+      "loss": 2.332,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989918071531395,
+      "loss": 2.3379,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898522930695732,
+      "loss": 2.3008,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897863029223962,
+      "loss": 2.2812,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989720101118381,
+      "loss": 2.3066,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989653687686136,
+      "loss": 2.3105,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989587062654361,
+      "loss": 2.2754,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895202260518476,
+      "loss": 2.3203,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894531779074788,
+      "loss": 2.2871,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893859182502286,
+      "loss": 2.2891,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893184471091632,
+      "loss": 2.3301,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892507645134388,
+      "loss": 2.3301,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891828704923044,
+      "loss": 2.3047,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891147650750998,
+      "loss": 2.334,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989046448291256,
+      "loss": 2.293,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009889779201702955,
+      "loss": 2.2344,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009889091807418321,
+      "loss": 2.2598,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888402300355713,
+      "loss": 2.3066,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988771068081309,
+      "loss": 2.3574,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887016949089334,
+      "loss": 2.3359,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988632110548423,
+      "loss": 2.3047,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885623150298487,
+      "loss": 2.3027,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884923083833715,
+      "loss": 2.2129,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884220906392442,
+      "loss": 2.2812,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988351661827811,
+      "loss": 2.3574,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882810219795071,
+      "loss": 2.3535,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882101711248586,
+      "loss": 2.3652,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881391092944835,
+      "loss": 2.2695,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880678365190901,
+      "loss": 2.2812,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879963528294786,
+      "loss": 2.3555,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00098792465825654,
+      "loss": 2.3086,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878527528312563,
+      "loss": 2.2207,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877806365847011,
+      "loss": 2.3301,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877083095480386,
+      "loss": 2.2422,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876357717525245,
+      "loss": 2.3301,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875630232295048,
+      "loss": 2.2734,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874900640104178,
+      "loss": 2.2754,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874168941267918,
+      "loss": 2.2441,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873435136102467,
+      "loss": 2.293,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987269922492493,
+      "loss": 2.3242,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871961208053328,
+      "loss": 2.2188,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871221085806586,
+      "loss": 2.2441,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870478858504544,
+      "loss": 2.3867,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869734526467945,
+      "loss": 2.3672,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868988090018448,
+      "loss": 2.2422,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868239549478618,
+      "loss": 2.2676,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867488905171934,
+      "loss": 2.2363,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866736157422775,
+      "loss": 2.3379,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865981306556436,
+      "loss": 2.2812,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865224352899118,
+      "loss": 2.3438,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864465296777938,
+      "loss": 2.2598,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863704138520907,
+      "loss": 2.3184,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862940878456958,
+      "loss": 2.2637,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862175516915925,
+      "loss": 2.2715,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861408054228555,
+      "loss": 2.3262,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860638490726498,
+      "loss": 2.291,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859866826742313,
+      "loss": 2.2617,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859093062609468,
+      "loss": 2.3125,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858317198662343,
+      "loss": 2.2949,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857539235236215,
+      "loss": 2.3184,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856759172667277,
+      "loss": 2.291,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855977011292625,
+      "loss": 2.2812,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855192751450266,
+      "loss": 2.2148,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854406393479105,
+      "loss": 2.2383,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009853617937718967,
+      "loss": 2.2617,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000985282738451057,
+      "loss": 2.2559,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852034734195547,
+      "loss": 2.2305,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851239987116436,
+      "loss": 2.373,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000985044314361668,
+      "loss": 2.2734,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849644204040625,
+      "loss": 2.3047,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984884316873353,
+      "loss": 2.2539,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984804003804155,
+      "loss": 2.3418,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847234812311756,
+      "loss": 2.4023,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846427491892117,
+      "loss": 2.3867,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845618077131508,
+      "loss": 2.2559,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844806568379714,
+      "loss": 2.2637,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843992965987418,
+      "loss": 2.2734,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843177270306212,
+      "loss": 2.2793,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842359481688592,
+      "loss": 2.3066,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841539600487959,
+      "loss": 2.3555,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840717627058616,
+      "loss": 2.2832,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839893561755773,
+      "loss": 2.2979,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839067404935542,
+      "loss": 2.3535,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838239156954938,
+      "loss": 2.1562,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837408818171882,
+      "loss": 2.3086,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098365763889452,
+      "loss": 2.2197,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835741869634615,
+      "loss": 2.3125,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983490526060076,
+      "loss": 2.2422,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834066562205165,
+      "loss": 2.3359,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983322577481027,
+      "loss": 2.248,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832382898779413,
+      "loss": 2.2363,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831537934476835,
+      "loss": 2.3076,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830690882267677,
+      "loss": 2.2695,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982984174251799,
+      "loss": 2.2812,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982899051559472,
+      "loss": 2.2734,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828137201865715,
+      "loss": 2.2754,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827281801699731,
+      "loss": 2.2949,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982642431546642,
+      "loss": 2.2773,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825564743536335,
+      "loss": 2.2793,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824703086280938,
+      "loss": 2.3086,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982383934407258,
+      "loss": 2.2168,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822973517284524,
+      "loss": 2.2568,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822105606290927,
+      "loss": 2.2148,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982123561146685,
+      "loss": 2.334,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820363533188255,
+      "loss": 2.3047,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819489371832003,
+      "loss": 2.209,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981861312777585,
+      "loss": 2.2812,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817734801398463,
+      "loss": 2.2949,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816854393079402,
+      "loss": 2.3555,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009815971903199127,
+      "loss": 2.2852,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009815087332138996,
+      "loss": 2.3086,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814200680281272,
+      "loss": 2.252,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813311948009109,
+      "loss": 2.2715,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981242113570657,
+      "loss": 2.1299,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000981152824375861,
+      "loss": 2.2148,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810633272551085,
+      "loss": 2.3496,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009809736222470748,
+      "loss": 2.2871,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808837093905254,
+      "loss": 2.3086,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807935887243147,
+      "loss": 2.2559,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980703260287388,
+      "loss": 2.2461,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00098061272411878,
+      "loss": 2.2559,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980521980257615,
+      "loss": 2.2949,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980431028743107,
+      "loss": 2.2637,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803398696145603,
+      "loss": 2.2129,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802485029113681,
+      "loss": 2.3418,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980156928673014,
+      "loss": 2.2285,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800651469390705,
+      "loss": 2.291,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979973157749201,
+      "loss": 2.3086,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798809611431575,
+      "loss": 2.1855,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797885571607818,
+      "loss": 2.2227,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796959458420057,
+      "loss": 2.3242,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00097960312722685,
+      "loss": 2.2676,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009795101013554258,
+      "loss": 2.3027,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794168682679333,
+      "loss": 2.2676,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793234280046626,
+      "loss": 2.2188,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009792297806059927,
+      "loss": 2.2422,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791359261123926,
+      "loss": 2.334,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009790418645644209,
+      "loss": 2.2441,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978947596002725,
+      "loss": 2.2812,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788531204680429,
+      "loss": 2.2168,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978758438001201,
+      "loss": 2.2422,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786635486431152,
+      "loss": 2.2334,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785684524347916,
+      "loss": 2.1758,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009784731494173252,
+      "loss": 2.2969,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783776396318999,
+      "loss": 2.2246,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782819231197897,
+      "loss": 2.2891,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009781859999223577,
+      "loss": 2.3086,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780898700810562,
+      "loss": 2.3086,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779935336374267,
+      "loss": 2.2402,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009778969906331005,
+      "loss": 2.2617,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009778002411097975,
+      "loss": 2.2559,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777032851093273,
+      "loss": 2.2871,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009776061226735883,
+      "loss": 2.2217,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977508753844569,
+      "loss": 2.2969,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977411178664346,
+      "loss": 2.3379,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009773133971750856,
+      "loss": 2.4512,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009772154094190434,
+      "loss": 2.2822,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771172154385637,
+      "loss": 2.3047,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770188152760804,
+      "loss": 2.2461,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976920208974116,
+      "loss": 2.1621,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768213965752827,
+      "loss": 2.3457,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976722378122281,
+      "loss": 2.2871,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766231536579013,
+      "loss": 2.2617,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765237232250222,
+      "loss": 2.2559,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764240868666119,
+      "loss": 2.2871,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763242446257272,
+      "loss": 2.2109,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762241965455141,
+      "loss": 2.3438,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009761239426692076,
+      "loss": 2.2676,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760234830401316,
+      "loss": 2.2109,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009759228177016985,
+      "loss": 2.3457,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758219466974104,
+      "loss": 2.2461,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757208700708575,
+      "loss": 2.2891,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009756195878657192,
+      "loss": 2.2441,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755181001257637,
+      "loss": 2.293,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754164068948481,
+      "loss": 2.25,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753145082169183,
+      "loss": 2.3086,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752124041360089,
+      "loss": 2.2227,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009751100946962432,
+      "loss": 2.1973,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750075799418332,
+      "loss": 2.2461,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097490485991708,
+      "loss": 2.1953,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009748019346663731,
+      "loss": 2.1914,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746988042341906,
+      "loss": 2.2656,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745954686650997,
+      "loss": 2.2617,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744919280037555,
+      "loss": 2.2109,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009743881822949026,
+      "loss": 2.2812,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009742842315833735,
+      "loss": 2.2646,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741800759140898,
+      "loss": 2.25,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740757153320614,
+      "loss": 2.2559,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739711498823868,
+      "loss": 2.2266,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738663796102529,
+      "loss": 2.2158,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737614045609355,
+      "loss": 2.1758,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736562247797984,
+      "loss": 2.2793,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009735508403122944,
+      "loss": 2.2402,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734452512039642,
+      "loss": 2.2031,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009733394575004373,
+      "loss": 2.2441,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732334592474314,
+      "loss": 2.2637,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731272564907532,
+      "loss": 2.2383,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730208492762966,
+      "loss": 2.293,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009729142376500449,
+      "loss": 2.25,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728074216580694,
+      "loss": 2.293,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727004013465296,
+      "loss": 2.1875,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725931767616733,
+      "loss": 2.291,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972485747949837,
+      "loss": 2.1426,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723781149574449,
+      "loss": 2.3496,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722702778310095,
+      "loss": 2.2207,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009721622366171318,
+      "loss": 2.3145,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720539913625008,
+      "loss": 2.2734,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009719455421138938,
+      "loss": 2.2441,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718368889181764,
+      "loss": 2.2598,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717280318223018,
+      "loss": 2.291,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716189708733116,
+      "loss": 2.2031,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715097061183357,
+      "loss": 2.2598,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000971400237604592,
+      "loss": 2.2344,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712905653793862,
+      "loss": 2.2949,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009711806894901123,
+      "loss": 2.1875,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710706099842519,
+      "loss": 2.2285,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709603269093752,
+      "loss": 2.2285,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00097084984031314,
+      "loss": 2.2266,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707391502432921,
+      "loss": 2.2559,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706282567476651,
+      "loss": 2.2695,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705171598741808,
+      "loss": 2.1973,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704058596708488,
+      "loss": 2.2383,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702943561857661,
+      "loss": 2.2236,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701826494671184,
+      "loss": 2.2344,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700707395631787,
+      "loss": 2.1572,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699586265223077,
+      "loss": 2.2129,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009698463103929542,
+      "loss": 2.2402,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697337912236546,
+      "loss": 2.2461,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969621069063033,
+      "loss": 2.2637,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009695081439598014,
+      "loss": 2.2227,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009693950159627594,
+      "loss": 2.2559,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692816851207941,
+      "loss": 2.3496,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009691681514828806,
+      "loss": 2.2051,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690544150980814,
+      "loss": 2.2695,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689404760155465,
+      "loss": 2.2773,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009688263342845139,
+      "loss": 2.2285,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968711989954309,
+      "loss": 2.1621,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009685974430743445,
+      "loss": 2.3047,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684826936941209,
+      "loss": 2.2119,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009683677418632262,
+      "loss": 2.1865,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682525876313357,
+      "loss": 2.1309,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681372310482123,
+      "loss": 2.2227,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009680216721637065,
+      "loss": 2.2344,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967905911027756,
+      "loss": 2.1914,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009677899476903856,
+      "loss": 2.2734,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676737822017083,
+      "loss": 2.252,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675574146119239,
+      "loss": 2.3047,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674408449713193,
+      "loss": 2.25,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673240733302695,
+      "loss": 2.2344,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009672070997392361,
+      "loss": 2.2412,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009670899242487682,
+      "loss": 2.2275,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669725469095021,
+      "loss": 2.2051,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668549677721616,
+      "loss": 2.2598,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667371868875574,
+      "loss": 2.2793,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009666192043065873,
+      "loss": 2.1738,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665010200802367,
+      "loss": 2.2734,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663826342595776,
+      "loss": 2.2305,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662640468957696,
+      "loss": 2.2568,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661452580400591,
+      "loss": 2.3086,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009660262677437798,
+      "loss": 2.2109,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000965907076058352,
+      "loss": 2.2598,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009657876830352837,
+      "loss": 2.2598,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656680887261692,
+      "loss": 2.1895,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655482931826905,
+      "loss": 2.2812,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009654282964566158,
+      "loss": 2.2539,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653080985998009,
+      "loss": 2.2432,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000965187699664188,
+      "loss": 2.1641,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650670997018067,
+      "loss": 2.2188,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009649462987647732,
+      "loss": 2.2598,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648252969052904,
+      "loss": 2.1973,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009647040941756482,
+      "loss": 2.1973,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009645826906282233,
+      "loss": 2.1875,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009644610863154795,
+      "loss": 2.252,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009643392812899665,
+      "loss": 2.2188,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642172756043216,
+      "loss": 2.2051,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640950693112684,
+      "loss": 2.2051,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009639726624636174,
+      "loss": 2.2617,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009638500551142653,
+      "loss": 2.2188,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637272473161962,
+      "loss": 2.2832,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009636042391224802,
+      "loss": 2.2383,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963481030586274,
+      "loss": 2.1934,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633576217608215,
+      "loss": 2.2402,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009632340126994526,
+      "loss": 2.1523,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631102034555839,
+      "loss": 2.252,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629861940827182,
+      "loss": 2.209,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628619846344453,
+      "loss": 2.2188,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627375751644411,
+      "loss": 2.1816,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009626129657264681,
+      "loss": 2.2285,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009624881563743754,
+      "loss": 2.1992,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000962363147162098,
+      "loss": 2.1836,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622379381436575,
+      "loss": 2.2285,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621125293731619,
+      "loss": 2.2471,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619869209048058,
+      "loss": 2.2188,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009618611127928694,
+      "loss": 2.1914,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617351050917195,
+      "loss": 2.1895,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009616088978558099,
+      "loss": 2.21,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009614824911396792,
+      "loss": 2.2031,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613558849979534,
+      "loss": 2.2031,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009612290794853438,
+      "loss": 2.1602,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611020746566487,
+      "loss": 2.2656,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000960974870566752,
+      "loss": 2.2559,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009608474672706239,
+      "loss": 2.2559,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607198648233204,
+      "loss": 2.2148,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605920632799838,
+      "loss": 2.252,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009604640626958427,
+      "loss": 2.2012,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603358631262111,
+      "loss": 2.2812,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602074646264893,
+      "loss": 2.209,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000960078867252164,
+      "loss": 2.1963,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000959950071058807,
+      "loss": 2.2539,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009598210761020765,
+      "loss": 2.1914,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009596918824377168,
+      "loss": 2.1973,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009595624901215573,
+      "loss": 2.1689,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009594328992095144,
+      "loss": 2.2051,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593031097575893,
+      "loss": 2.2188,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591731218218693,
+      "loss": 2.2461,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009590429354585277,
+      "loss": 2.2324,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009589125507238233,
+      "loss": 2.1416,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009587819676741008,
+      "loss": 2.248,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009586511863657906,
+      "loss": 2.2207,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009585202068554083,
+      "loss": 2.2637,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583890291995561,
+      "loss": 2.1816,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009582576534549208,
+      "loss": 2.166,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009581260796782754,
+      "loss": 2.209,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009579943079264784,
+      "loss": 2.3008,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578623382564739,
+      "loss": 2.2227,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009577301707252913,
+      "loss": 2.2041,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009575978053900456,
+      "loss": 2.2227,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009574652423079375,
+      "loss": 2.2207,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573324815362528,
+      "loss": 2.2344,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009571995231323629,
+      "loss": 2.2363,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009570663671537248,
+      "loss": 2.1504,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569330136578803,
+      "loss": 2.1621,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567994627024572,
+      "loss": 2.1895,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009566657143451685,
+      "loss": 2.2734,
+      "step": 786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565317686438122,
+      "loss": 2.2773,
+      "step": 787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009563976256562717,
+      "loss": 2.248,
+      "step": 788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009562632854405158,
+      "loss": 2.1582,
+      "step": 789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009561287480545984,
+      "loss": 2.2344,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009559940135566587,
+      "loss": 2.2012,
+      "step": 791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000955859082004921,
+      "loss": 2.25,
+      "step": 792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009557239534576948,
+      "loss": 2.1953,
+      "step": 793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009555886279733746,
+      "loss": 2.2383,
+      "step": 794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009554531056104403,
+      "loss": 2.2656,
+      "step": 795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009553173864274567,
+      "loss": 2.2832,
+      "step": 796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009551814704830733,
+      "loss": 2.2441,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009550453578360254,
+      "loss": 2.25,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009549090485451327,
+      "loss": 2.1836,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009547725426693001,
+      "loss": 2.1367,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546358402675173,
+      "loss": 2.2246,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009544989413988592,
+      "loss": 2.2109,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009543618461224854,
+      "loss": 2.2871,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009542245544976402,
+      "loss": 2.3027,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009540870665836534,
+      "loss": 2.1797,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009539493824399388,
+      "loss": 2.2188,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009538115021259956,
+      "loss": 2.2773,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009536734257014074,
+      "loss": 2.1523,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000953535153225843,
+      "loss": 2.2461,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533966847590551,
+      "loss": 2.2188,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009532580203608823,
+      "loss": 2.251,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009531191600912468,
+      "loss": 2.1895,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009529801040101558,
+      "loss": 2.1592,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009528408521777012,
+      "loss": 2.252,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009527014046540597,
+      "loss": 2.1885,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009525617614994921,
+      "loss": 2.2637,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000952421922774344,
+      "loss": 2.1494,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009522818885390455,
+      "loss": 2.1963,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009521416588541111,
+      "loss": 2.1738,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00095200123378014,
+      "loss": 2.1377,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009518606133778153,
+      "loss": 2.2285,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009517197977079052,
+      "loss": 2.25,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515787868312618,
+      "loss": 2.1855,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009514375808088218,
+      "loss": 2.1514,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009512961797016062,
+      "loss": 2.1758,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009511545835707199,
+      "loss": 2.2148,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009510127924773528,
+      "loss": 2.291,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508708064827785,
+      "loss": 2.2441,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950728625648355,
+      "loss": 2.2422,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009505862500355246,
+      "loss": 2.1992,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009504436797058136,
+      "loss": 2.25,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009503009147208323,
+      "loss": 2.2188,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009501579551422759,
+      "loss": 2.1797,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009500148010319225,
+      "loss": 2.1875,
+      "step": 834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009498714524516352,
+      "loss": 2.127,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009497279094633607,
+      "loss": 2.1738,
+      "step": 836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009495841721291299,
+      "loss": 2.165,
+      "step": 837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009494402405110579,
+      "loss": 2.2012,
+      "step": 838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009492961146713428,
+      "loss": 2.1318,
+      "step": 839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009491517946722681,
+      "loss": 2.2188,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490072805761998,
+      "loss": 2.2578,
+      "step": 841
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009488625724455887,
+      "loss": 2.2285,
+      "step": 842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009487176703429691,
+      "loss": 2.2402,
+      "step": 843
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009485725743309593,
+      "loss": 2.1641,
+      "step": 844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009484272844722609,
+      "loss": 2.2383,
+      "step": 845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009482818008296599,
+      "loss": 2.2188,
+      "step": 846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009481361234660257,
+      "loss": 2.1855,
+      "step": 847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009479902524443116,
+      "loss": 2.2148,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009478441878275543,
+      "loss": 2.1953,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009476979296788746,
+      "loss": 2.2148,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009475514780614764,
+      "loss": 2.209,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009474048330386475,
+      "loss": 2.1836,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009472579946737594,
+      "loss": 2.1982,
+      "step": 853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947110963030267,
+      "loss": 2.2266,
+      "step": 854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009469637381717085,
+      "loss": 2.1816,
+      "step": 855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009468163201617061,
+      "loss": 2.1855,
+      "step": 856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009466687090639652,
+      "loss": 2.1699,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009465209049422744,
+      "loss": 2.2246,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009463729078605061,
+      "loss": 2.1914,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462247178826158,
+      "loss": 2.2285,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009460763350726428,
+      "loss": 2.1738,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009459277594947091,
+      "loss": 2.1738,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009457789912130206,
+      "loss": 2.1689,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009456300302918658,
+      "loss": 2.125,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009454808767956173,
+      "loss": 2.3477,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009453315307887303,
+      "loss": 2.2207,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009451819923357434,
+      "loss": 2.2012,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009450322615012782,
+      "loss": 2.1836,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448823383500396,
+      "loss": 2.25,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009447322229468156,
+      "loss": 2.2334,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009445819153564774,
+      "loss": 2.2578,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009444314156439787,
+      "loss": 2.1797,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000944280723874357,
+      "loss": 2.2568,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009441298401127322,
+      "loss": 2.1836,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009439787644243078,
+      "loss": 2.1211,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438274968743692,
+      "loss": 2.1543,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009436760375282858,
+      "loss": 2.207,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009435243864515093,
+      "loss": 2.291,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433725437095743,
+      "loss": 2.2168,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009432205093680983,
+      "loss": 2.1973,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430682834927817,
+      "loss": 2.252,
+      "step": 881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009429158661494077,
+      "loss": 2.2354,
+      "step": 882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009427632574038418,
+      "loss": 2.2871,
+      "step": 883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009426104573220326,
+      "loss": 2.3105,
+      "step": 884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009424574659700116,
+      "loss": 2.1621,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009423042834138924,
+      "loss": 2.207,
+      "step": 886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421509097198715,
+      "loss": 2.1641,
+      "step": 887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009419973449542279,
+      "loss": 2.2432,
+      "step": 888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418435891833236,
+      "loss": 2.1973,
+      "step": 889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416896424736025,
+      "loss": 2.1777,
+      "step": 890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009415355048915915,
+      "loss": 2.1133,
+      "step": 891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009413811765038995,
+      "loss": 2.1709,
+      "step": 892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009412266573772186,
+      "loss": 2.1777,
+      "step": 893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009410719475783225,
+      "loss": 2.2539,
+      "step": 894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940917047174068,
+      "loss": 2.1777,
+      "step": 895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009407619562313936,
+      "loss": 2.3145,
+      "step": 896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009406066748173208,
+      "loss": 2.2891,
+      "step": 897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009404512029989527,
+      "loss": 2.1816,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009402955408434755,
+      "loss": 2.1484,
+      "step": 899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940139688418157,
+      "loss": 2.1543,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009399836457903477,
+      "loss": 2.1865,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009398274130274798,
+      "loss": 2.2188,
+      "step": 902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009396709901970678,
+      "loss": 2.2539,
+      "step": 903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009395143773667088,
+      "loss": 2.1553,
+      "step": 904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009393575746040813,
+      "loss": 2.2285,
+      "step": 905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009392005819769465,
+      "loss": 2.2891,
+      "step": 906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009390433995531474,
+      "loss": 2.1309,
+      "step": 907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009388860274006087,
+      "loss": 2.2754,
+      "step": 908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009387284655873376,
+      "loss": 2.2168,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009385707141814231,
+      "loss": 2.2051,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009384127732510361,
+      "loss": 2.2578,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009382546428644291,
+      "loss": 2.1953,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009380963230899371,
+      "loss": 2.1865,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009379378139959764,
+      "loss": 2.291,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009377791156510454,
+      "loss": 2.2109,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009376202281237241,
+      "loss": 2.1504,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009374611514826745,
+      "loss": 2.25,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00093730188579664,
+      "loss": 2.2422,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009371424311344458,
+      "loss": 2.1279,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009369827875649992,
+      "loss": 2.1914,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009368229551572884,
+      "loss": 2.2246,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009366629339803836,
+      "loss": 2.1406,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009365027241034367,
+      "loss": 2.1914,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009363423255956809,
+      "loss": 2.2051,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009361817385264311,
+      "loss": 2.2891,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009360209629650834,
+      "loss": 2.2119,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009358599989811158,
+      "loss": 2.1367,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009356988466440872,
+      "loss": 2.2256,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009355375060236383,
+      "loss": 2.1836,
+      "step": 929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000935375977189491,
+      "loss": 2.2148,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009352142602114486,
+      "loss": 2.1934,
+      "step": 931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009350523551593957,
+      "loss": 2.2422,
+      "step": 932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000934890262103298,
+      "loss": 2.168,
+      "step": 933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009347279811132027,
+      "loss": 2.082,
+      "step": 934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000934565512259238,
+      "loss": 2.2246,
+      "step": 935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009344028556116135,
+      "loss": 2.2012,
+      "step": 936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00093424001124062,
+      "loss": 2.2148,
+      "step": 937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009340769792166288,
+      "loss": 2.2217,
+      "step": 938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000933913759610093,
+      "loss": 2.1045,
+      "step": 939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009337503524915467,
+      "loss": 2.2031,
+      "step": 940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009335867579316047,
+      "loss": 2.1631,
+      "step": 941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009334229760009628,
+      "loss": 2.21,
+      "step": 942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009332590067703981,
+      "loss": 2.1494,
+      "step": 943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009330948503107684,
+      "loss": 2.2129,
+      "step": 944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009329305066930125,
+      "loss": 2.1816,
+      "step": 945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009327659759881499,
+      "loss": 2.2422,
+      "step": 946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009326012582672814,
+      "loss": 2.1846,
+      "step": 947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009324363536015878,
+      "loss": 2.2051,
+      "step": 948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322712620623318,
+      "loss": 2.1348,
+      "step": 949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932105983720856,
+      "loss": 2.1582,
+      "step": 950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009319405186485838,
+      "loss": 2.1992,
+      "step": 951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317748669170198,
+      "loss": 2.1504,
+      "step": 952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009316090285977486,
+      "loss": 2.1436,
+      "step": 953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009314430037624362,
+      "loss": 2.1865,
+      "step": 954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009312767924828283,
+      "loss": 2.168,
+      "step": 955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009311103948307519,
+      "loss": 2.1309,
+      "step": 956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009309438108781142,
+      "loss": 2.1992,
+      "step": 957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000930777040696903,
+      "loss": 2.2051,
+      "step": 958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009306100843591867,
+      "loss": 2.1426,
+      "step": 959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009304429419371139,
+      "loss": 2.168,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009302756135029137,
+      "loss": 2.1934,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301080991288957,
+      "loss": 2.2539,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009299403988874495,
+      "loss": 2.2539,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297725128510456,
+      "loss": 2.0898,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009296044410922344,
+      "loss": 2.1396,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009294361836836465,
+      "loss": 2.2129,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000929267740697993,
+      "loss": 2.1621,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009290991122080649,
+      "loss": 2.1719,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009289302982867335,
+      "loss": 2.2559,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009287612990069508,
+      "loss": 2.2246,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009285921144417475,
+      "loss": 2.2207,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009284227446642357,
+      "loss": 2.1514,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009282531897476071,
+      "loss": 2.2324,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009280834497651332,
+      "loss": 2.2383,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009279135247901658,
+      "loss": 2.1953,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009277434148961363,
+      "loss": 2.1934,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009275731201565563,
+      "loss": 2.25,
+      "step": 977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000927402640645017,
+      "loss": 2.1758,
+      "step": 978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009272319764351899,
+      "loss": 2.1758,
+      "step": 979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009270611276008258,
+      "loss": 2.2168,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009268900942157556,
+      "loss": 2.1963,
+      "step": 981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00092671887635389,
+      "loss": 2.2139,
+      "step": 982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009265474740892192,
+      "loss": 2.1855,
+      "step": 983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009263758874958131,
+      "loss": 2.2598,
+      "step": 984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009262041166478213,
+      "loss": 2.1045,
+      "step": 985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009260321616194733,
+      "loss": 2.1855,
+      "step": 986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009258600224850777,
+      "loss": 2.2285,
+      "step": 987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009256876993190231,
+      "loss": 2.1133,
+      "step": 988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009255151921957772,
+      "loss": 2.1309,
+      "step": 989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009253425011898878,
+      "loss": 2.1729,
+      "step": 990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009251696263759815,
+      "loss": 2.1289,
+      "step": 991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009249965678287647,
+      "loss": 2.1982,
+      "step": 992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000924823325623023,
+      "loss": 2.2012,
+      "step": 993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009246498998336218,
+      "loss": 2.2754,
+      "step": 994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009244762905355053,
+      "loss": 2.1396,
+      "step": 995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009243024978036974,
+      "loss": 2.1357,
+      "step": 996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009241285217133009,
+      "loss": 2.2344,
+      "step": 997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009239543623394982,
+      "loss": 2.1309,
+      "step": 998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009237800197575505,
+      "loss": 2.252,
+      "step": 999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009236054940427987,
+      "loss": 2.2266,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009234307852706624,
+      "loss": 2.252,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009232558935166407,
+      "loss": 2.293,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009230808188563109,
+      "loss": 2.2139,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009229055613653308,
+      "loss": 2.168,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009227301211194356,
+      "loss": 2.126,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009225544981944409,
+      "loss": 2.2656,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009223786926662403,
+      "loss": 2.207,
+      "step": 1007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009222027046108066,
+      "loss": 2.2266,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009220265341041916,
+      "loss": 2.2754,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009218501812225257,
+      "loss": 2.1475,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009216736460420183,
+      "loss": 2.2188,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009214969286389577,
+      "loss": 2.1699,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009213200290897103,
+      "loss": 2.1328,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009211429474707221,
+      "loss": 2.1699,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009209656838585174,
+      "loss": 2.1494,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009207882383296988,
+      "loss": 2.1289,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000920610610960948,
+      "loss": 2.2363,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009204328018290252,
+      "loss": 2.209,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009202548110107688,
+      "loss": 2.207,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009200766385830962,
+      "loss": 2.1328,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009198982846230028,
+      "loss": 2.1846,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009197197492075631,
+      "loss": 2.2207,
+      "step": 1022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919541032413929,
+      "loss": 2.1426,
+      "step": 1023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919362134319332,
+      "loss": 2.127,
+      "step": 1024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009191830550010811,
+      "loss": 2.1318,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009190037945365637,
+      "loss": 2.1836,
+      "step": 1026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009188243530032457,
+      "loss": 2.127,
+      "step": 1027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009186447304786714,
+      "loss": 2.2217,
+      "step": 1028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009184649270404628,
+      "loss": 2.1367,
+      "step": 1029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009182849427663204,
+      "loss": 2.1348,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009181047777340232,
+      "loss": 2.2539,
+      "step": 1031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009179244320214275,
+      "loss": 2.2129,
+      "step": 1032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009177439057064682,
+      "loss": 2.209,
+      "step": 1033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009175631988671583,
+      "loss": 2.2383,
+      "step": 1034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009173823115815886,
+      "loss": 2.1582,
+      "step": 1035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000917201243927928,
+      "loss": 2.2207,
+      "step": 1036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009170199959844231,
+      "loss": 2.1885,
+      "step": 1037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009168385678293986,
+      "loss": 2.1562,
+      "step": 1038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009166569595412575,
+      "loss": 2.1729,
+      "step": 1039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009164751711984798,
+      "loss": 2.0889,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009162932028796239,
+      "loss": 2.1074,
+      "step": 1041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009161110546633258,
+      "loss": 2.2363,
+      "step": 1042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009159287266282992,
+      "loss": 2.1846,
+      "step": 1043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009157462188533356,
+      "loss": 2.2168,
+      "step": 1044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009155635314173039,
+      "loss": 2.1318,
+      "step": 1045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009153806643991514,
+      "loss": 2.1289,
+      "step": 1046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009151976178779019,
+      "loss": 2.1719,
+      "step": 1047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009150143919326577,
+      "loss": 2.1641,
+      "step": 1048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009148309866425981,
+      "loss": 2.2051,
+      "step": 1049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009146474020869801,
+      "loss": 2.167,
+      "step": 1050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009144636383451384,
+      "loss": 2.1445,
+      "step": 1051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009142796954964845,
+      "loss": 2.0801,
+      "step": 1052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009140955736205078,
+      "loss": 2.1484,
+      "step": 1053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009139112727967751,
+      "loss": 2.1602,
+      "step": 1054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009137267931049304,
+      "loss": 2.1631,
+      "step": 1055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009135421346246948,
+      "loss": 2.166,
+      "step": 1056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009133572974358668,
+      "loss": 2.1621,
+      "step": 1057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009131722816183224,
+      "loss": 2.1328,
+      "step": 1058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009129870872520143,
+      "loss": 2.2051,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009128017144169727,
+      "loss": 2.1289,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000912616163193305,
+      "loss": 2.2285,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009124304336611953,
+      "loss": 2.1562,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009122445259009052,
+      "loss": 2.1445,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009120584399927727,
+      "loss": 2.2148,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009118721760172137,
+      "loss": 2.1367,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009116857340547202,
+      "loss": 2.1816,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009114991141858617,
+      "loss": 2.166,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009113123164912842,
+      "loss": 2.2207,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009111253410517106,
+      "loss": 2.1602,
+      "step": 1069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009109381879479407,
+      "loss": 2.2109,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009107508572608512,
+      "loss": 2.2695,
+      "step": 1071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009105633490713952,
+      "loss": 2.168,
+      "step": 1072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009103756634606029,
+      "loss": 2.1709,
+      "step": 1073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009101878005095812,
+      "loss": 2.207,
+      "step": 1074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009099997602995128,
+      "loss": 2.1494,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009098115429116582,
+      "loss": 2.1934,
+      "step": 1076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009096231484273533,
+      "loss": 2.1074,
+      "step": 1077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009094345769280119,
+      "loss": 2.1846,
+      "step": 1078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009092458284951229,
+      "loss": 2.1885,
+      "step": 1079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009090569032102525,
+      "loss": 2.1738,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009088678011550431,
+      "loss": 2.2227,
+      "step": 1081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009086785224112132,
+      "loss": 2.1924,
+      "step": 1082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009084890670605586,
+      "loss": 2.1699,
+      "step": 1083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00090829943518495,
+      "loss": 2.1953,
+      "step": 1084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009081096268663358,
+      "loss": 2.2031,
+      "step": 1085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009079196421867395,
+      "loss": 2.2197,
+      "step": 1086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009077294812282615,
+      "loss": 2.1641,
+      "step": 1087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009075391440730783,
+      "loss": 2.165,
+      "step": 1088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009073486308034422,
+      "loss": 2.21,
+      "step": 1089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000907157941501682,
+      "loss": 2.168,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009069670762502023,
+      "loss": 2.1162,
+      "step": 1091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009067760351314837,
+      "loss": 2.25,
+      "step": 1092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009065848182280834,
+      "loss": 2.1426,
+      "step": 1093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009063934256226337,
+      "loss": 2.1377,
+      "step": 1094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009062018573978434,
+      "loss": 2.2217,
+      "step": 1095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000906010113636497,
+      "loss": 2.1826,
+      "step": 1096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009058181944214547,
+      "loss": 2.2461,
+      "step": 1097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000905626099835653,
+      "loss": 2.1875,
+      "step": 1098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009054338299621037,
+      "loss": 2.2598,
+      "step": 1099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009052413848838945,
+      "loss": 2.1641,
+      "step": 1100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000905048764684189,
+      "loss": 2.124,
+      "step": 1101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009048559694462262,
+      "loss": 2.0977,
+      "step": 1102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009046629992533208,
+      "loss": 2.1211,
+      "step": 1103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009044698541888634,
+      "loss": 2.1387,
+      "step": 1104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009042765343363197,
+      "loss": 2.1904,
+      "step": 1105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009040830397792313,
+      "loss": 2.1621,
+      "step": 1106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009038893706012152,
+      "loss": 2.1719,
+      "step": 1107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009036955268859638,
+      "loss": 2.1582,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009035015087172449,
+      "loss": 2.0938,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009033073161789019,
+      "loss": 2.0957,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009031129493548529,
+      "loss": 2.1377,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009029184083290926,
+      "loss": 2.2402,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009027236931856894,
+      "loss": 2.2217,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009025288040087885,
+      "loss": 2.1719,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009023337408826089,
+      "loss": 2.2148,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009021385038914457,
+      "loss": 2.1934,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009019430931196689,
+      "loss": 2.1855,
+      "step": 1117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009017475086517233,
+      "loss": 2.2402,
+      "step": 1118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009015517505721293,
+      "loss": 2.1426,
+      "step": 1119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009013558189654818,
+      "loss": 2.1943,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009011597139164511,
+      "loss": 2.1611,
+      "step": 1121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009009634355097822,
+      "loss": 2.1621,
+      "step": 1122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000900766983830295,
+      "loss": 2.1074,
+      "step": 1123
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009005703589628844,
+      "loss": 2.2559,
+      "step": 1124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009003735609925201,
+      "loss": 2.1104,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009001765900042468,
+      "loss": 2.2402,
+      "step": 1126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008999794460831835,
+      "loss": 2.0957,
+      "step": 1127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008997821293145243,
+      "loss": 2.1191,
+      "step": 1128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008995846397835381,
+      "loss": 2.1875,
+      "step": 1129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899386977575568,
+      "loss": 2.2031,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008991891427760321,
+      "loss": 2.1602,
+      "step": 1131
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008989911354704229,
+      "loss": 2.0957,
+      "step": 1132
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008987929557443075,
+      "loss": 2.1689,
+      "step": 1133
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008985946036833278,
+      "loss": 2.1768,
+      "step": 1134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008983960793731995,
+      "loss": 2.0723,
+      "step": 1135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008981973828997134,
+      "loss": 2.1279,
+      "step": 1136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008979985143487344,
+      "loss": 2.0947,
+      "step": 1137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008977994738062016,
+      "loss": 2.1426,
+      "step": 1138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008976002613581288,
+      "loss": 2.1104,
+      "step": 1139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008974008770906039,
+      "loss": 2.2246,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897201321089789,
+      "loss": 2.1729,
+      "step": 1141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008970015934419204,
+      "loss": 2.207,
+      "step": 1142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000896801694233309,
+      "loss": 2.2168,
+      "step": 1143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008966016235503389,
+      "loss": 2.082,
+      "step": 1144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008964013814794693,
+      "loss": 2.2129,
+      "step": 1145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008962009681072332,
+      "loss": 2.1328,
+      "step": 1146
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008960003835202369,
+      "loss": 2.1426,
+      "step": 1147
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008957996278051617,
+      "loss": 2.1074,
+      "step": 1148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008955987010487623,
+      "loss": 2.1729,
+      "step": 1149
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008953976033378674,
+      "loss": 2.1592,
+      "step": 1150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008951963347593796,
+      "loss": 2.2344,
+      "step": 1151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008949948954002756,
+      "loss": 2.168,
+      "step": 1152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008947932853476051,
+      "loss": 2.1543,
+      "step": 1153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008945915046884926,
+      "loss": 2.1836,
+      "step": 1154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008943895535101356,
+      "loss": 2.1963,
+      "step": 1155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008941874318998057,
+      "loss": 2.1328,
+      "step": 1156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008939851399448477,
+      "loss": 2.1855,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008937826777326807,
+      "loss": 2.1143,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008935800453507965,
+      "loss": 2.1875,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933772428867613,
+      "loss": 2.1621,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008931742704282139,
+      "loss": 2.1875,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008929711280628676,
+      "loss": 2.2793,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008927678158785085,
+      "loss": 2.2109,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008925643339629957,
+      "loss": 2.1748,
+      "step": 1164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008923606824042627,
+      "loss": 2.0762,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008921568612903153,
+      "loss": 2.168,
+      "step": 1166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008919528707092332,
+      "loss": 2.124,
+      "step": 1167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008917487107491693,
+      "loss": 2.2021,
+      "step": 1168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000891544381498349,
+      "loss": 2.209,
+      "step": 1169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913398830450721,
+      "loss": 2.1816,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008911352154777101,
+      "loss": 2.1162,
+      "step": 1171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008909303788847087,
+      "loss": 2.209,
+      "step": 1172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008907253733545864,
+      "loss": 2.1768,
+      "step": 1173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000890520198975934,
+      "loss": 2.1387,
+      "step": 1174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008903148558374162,
+      "loss": 2.1455,
+      "step": 1175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008901093440277701,
+      "loss": 2.1836,
+      "step": 1176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008899036636358058,
+      "loss": 2.2324,
+      "step": 1177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008896978147504063,
+      "loss": 2.1836,
+      "step": 1178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008894917974605275,
+      "loss": 2.2402,
+      "step": 1179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008892856118551978,
+      "loss": 2.1152,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008890792580235183,
+      "loss": 2.1367,
+      "step": 1181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008888727360546635,
+      "loss": 2.041,
+      "step": 1182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008886660460378798,
+      "loss": 2.1523,
+      "step": 1183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008884591880624861,
+      "loss": 2.1709,
+      "step": 1184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000888252162217875,
+      "loss": 2.1924,
+      "step": 1185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008880449685935102,
+      "loss": 2.1895,
+      "step": 1186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008878376072789291,
+      "loss": 2.1699,
+      "step": 1187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008876300783637408,
+      "loss": 2.1748,
+      "step": 1188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008874223819376272,
+      "loss": 2.1797,
+      "step": 1189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008872145180903423,
+      "loss": 2.1641,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000887006486911713,
+      "loss": 2.2324,
+      "step": 1191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008867982884916376,
+      "loss": 2.2539,
+      "step": 1192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008865899229200878,
+      "loss": 2.0557,
+      "step": 1193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008863813902871067,
+      "loss": 2.1943,
+      "step": 1194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008861726906828096,
+      "loss": 2.2246,
+      "step": 1195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008859638241973845,
+      "loss": 2.1816,
+      "step": 1196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008857547909210912,
+      "loss": 2.2002,
+      "step": 1197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008855455909442615,
+      "loss": 2.0742,
+      "step": 1198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008853362243572994,
+      "loss": 2.2871,
+      "step": 1199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008851266912506806,
+      "loss": 2.1484,
+      "step": 1200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008849169917149531,
+      "loss": 2.1758,
+      "step": 1201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000884707125840737,
+      "loss": 2.1484,
+      "step": 1202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008844970937187236,
+      "loss": 2.1328,
+      "step": 1203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008842868954396766,
+      "loss": 2.1035,
+      "step": 1204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008840765310944315,
+      "loss": 2.0947,
+      "step": 1205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000883866000773895,
+      "loss": 2.1201,
+      "step": 1206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008836553045690462,
+      "loss": 2.2061,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008834444425709356,
+      "loss": 2.1641,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008832334148706853,
+      "loss": 2.1201,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000883022221559489,
+      "loss": 2.2051,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008828108627286122,
+      "loss": 2.165,
+      "step": 1211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008825993384693917,
+      "loss": 2.1582,
+      "step": 1212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008823876488732357,
+      "loss": 2.1953,
+      "step": 1213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008821757940316242,
+      "loss": 2.1133,
+      "step": 1214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008819637740361081,
+      "loss": 2.1816,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008817515889783105,
+      "loss": 2.1387,
+      "step": 1216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008815392389499248,
+      "loss": 2.3008,
+      "step": 1217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008813267240427164,
+      "loss": 2.2305,
+      "step": 1218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008811140443485217,
+      "loss": 2.1758,
+      "step": 1219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008809011999592484,
+      "loss": 2.1562,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008806881909668754,
+      "loss": 2.165,
+      "step": 1221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008804750174634526,
+      "loss": 2.0996,
+      "step": 1222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008802616795411009,
+      "loss": 2.1123,
+      "step": 1223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008800481772920126,
+      "loss": 2.123,
+      "step": 1224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008798345108084507,
+      "loss": 2.1426,
+      "step": 1225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008796206801827493,
+      "loss": 2.1211,
+      "step": 1226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008794066855073135,
+      "loss": 2.1484,
+      "step": 1227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008791925268746192,
+      "loss": 2.1973,
+      "step": 1228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008789782043772132,
+      "loss": 2.1445,
+      "step": 1229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008787637181077129,
+      "loss": 2.1885,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008785490681588069,
+      "loss": 2.1982,
+      "step": 1231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000878334254623254,
+      "loss": 2.2021,
+      "step": 1232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008781192775938843,
+      "loss": 2.1572,
+      "step": 1233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008779041371635981,
+      "loss": 2.1016,
+      "step": 1234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008776888334253662,
+      "loss": 2.2422,
+      "step": 1235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008774733664722305,
+      "loss": 2.207,
+      "step": 1236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008772577363973033,
+      "loss": 2.2148,
+      "step": 1237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008770419432937667,
+      "loss": 2.2227,
+      "step": 1238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008768259872548742,
+      "loss": 2.1787,
+      "step": 1239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008766098683739494,
+      "loss": 2.1357,
+      "step": 1240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008763935867443858,
+      "loss": 2.251,
+      "step": 1241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008761771424596476,
+      "loss": 2.168,
+      "step": 1242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008759605356132698,
+      "loss": 2.2148,
+      "step": 1243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008757437662988566,
+      "loss": 2.3203,
+      "step": 1244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008755268346100834,
+      "loss": 2.126,
+      "step": 1245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008753097406406948,
+      "loss": 2.1035,
+      "step": 1246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008750924844845064,
+      "loss": 2.126,
+      "step": 1247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008748750662354034,
+      "loss": 2.1631,
+      "step": 1248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008746574859873416,
+      "loss": 2.1133,
+      "step": 1249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008744397438343456,
+      "loss": 2.1738,
+      "step": 1250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008742218398705114,
+      "loss": 2.1738,
+      "step": 1251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000874003774190004,
+      "loss": 2.1045,
+      "step": 1252
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008737855468870587,
+      "loss": 2.1621,
+      "step": 1253
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008735671580559803,
+      "loss": 2.1182,
+      "step": 1254
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008733486077911439,
+      "loss": 2.1523,
+      "step": 1255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008731298961869938,
+      "loss": 2.1348,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008729110233380445,
+      "loss": 2.1084,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008726919893388798,
+      "loss": 2.1895,
+      "step": 1258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008724727942841535,
+      "loss": 2.1768,
+      "step": 1259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008722534382685888,
+      "loss": 2.1094,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008720339213869781,
+      "loss": 2.1084,
+      "step": 1261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008718142437341843,
+      "loss": 2.1826,
+      "step": 1262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008715944054051386,
+      "loss": 2.1699,
+      "step": 1263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008713744064948425,
+      "loss": 2.1504,
+      "step": 1264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008711542470983667,
+      "loss": 2.1396,
+      "step": 1265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008709339273108507,
+      "loss": 2.1709,
+      "step": 1266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008707134472275041,
+      "loss": 2.0762,
+      "step": 1267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008704928069436054,
+      "loss": 2.1533,
+      "step": 1268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008702720065545023,
+      "loss": 2.2275,
+      "step": 1269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008700510461556117,
+      "loss": 2.0713,
+      "step": 1270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008698299258424198,
+      "loss": 2.1641,
+      "step": 1271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000869608645710482,
+      "loss": 2.1396,
+      "step": 1272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000869387205855422,
+      "loss": 2.1562,
+      "step": 1273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008691656063729334,
+      "loss": 2.2041,
+      "step": 1274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008689438473587785,
+      "loss": 2.1533,
+      "step": 1275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008687219289087885,
+      "loss": 2.1348,
+      "step": 1276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008684998511188633,
+      "loss": 2.1895,
+      "step": 1277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008682776140849722,
+      "loss": 2.2139,
+      "step": 1278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008680552179031527,
+      "loss": 2.1562,
+      "step": 1279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008678326626695114,
+      "loss": 2.1914,
+      "step": 1280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008676099484802238,
+      "loss": 2.0693,
+      "step": 1281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008673870754315336,
+      "loss": 2.1348,
+      "step": 1282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008671640436197537,
+      "loss": 2.1377,
+      "step": 1283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008669408531412651,
+      "loss": 2.2422,
+      "step": 1284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008667175040925174,
+      "loss": 2.1484,
+      "step": 1285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008664939965700296,
+      "loss": 2.1826,
+      "step": 1286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008662703306703878,
+      "loss": 2.125,
+      "step": 1287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008660465064902477,
+      "loss": 2.1084,
+      "step": 1288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008658225241263329,
+      "loss": 2.1885,
+      "step": 1289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000865598383675435,
+      "loss": 2.0977,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008653740852344149,
+      "loss": 2.1807,
+      "step": 1291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000865149628900201,
+      "loss": 2.248,
+      "step": 1292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008649250147697898,
+      "loss": 2.1592,
+      "step": 1293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008647002429402467,
+      "loss": 2.1162,
+      "step": 1294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008644753135087047,
+      "loss": 2.1621,
+      "step": 1295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008642502265723653,
+      "loss": 2.1377,
+      "step": 1296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008640249822284975,
+      "loss": 2.1729,
+      "step": 1297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008637995805744387,
+      "loss": 2.0908,
+      "step": 1298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008635740217075946,
+      "loss": 2.1494,
+      "step": 1299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008633483057254381,
+      "loss": 2.2344,
+      "step": 1300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008631224327255105,
+      "loss": 2.0938,
+      "step": 1301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008628964028054208,
+      "loss": 2.1123,
+      "step": 1302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862670216062846,
+      "loss": 2.1162,
+      "step": 1303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008624438725955305,
+      "loss": 2.0459,
+      "step": 1304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008622173725012867,
+      "loss": 2.124,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008619907158779947,
+      "loss": 2.0742,
+      "step": 1306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008617639028236021,
+      "loss": 2.2012,
+      "step": 1307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008615369334361241,
+      "loss": 2.1504,
+      "step": 1308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008613098078136437,
+      "loss": 2.1992,
+      "step": 1309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008610825260543109,
+      "loss": 2.084,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860855088256344,
+      "loss": 2.0938,
+      "step": 1311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008606274945180279,
+      "loss": 2.2041,
+      "step": 1312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008603997449377152,
+      "loss": 2.0566,
+      "step": 1313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008601718396138262,
+      "loss": 2.2012,
+      "step": 1314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008599437786448479,
+      "loss": 2.1299,
+      "step": 1315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008597155621293349,
+      "loss": 2.1777,
+      "step": 1316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008594871901659089,
+      "loss": 2.2051,
+      "step": 1317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008592586628532588,
+      "loss": 2.0771,
+      "step": 1318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008590299802901411,
+      "loss": 2.1953,
+      "step": 1319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008588011425753783,
+      "loss": 2.0117,
+      "step": 1320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008585721498078611,
+      "loss": 2.1562,
+      "step": 1321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008583430020865464,
+      "loss": 2.0488,
+      "step": 1322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008581136995104585,
+      "loss": 2.1348,
+      "step": 1323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008578842421786887,
+      "loss": 2.1426,
+      "step": 1324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008576546301903947,
+      "loss": 2.084,
+      "step": 1325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008574248636448013,
+      "loss": 2.1709,
+      "step": 1326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008571949426412002,
+      "loss": 2.123,
+      "step": 1327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008569648672789496,
+      "loss": 2.2002,
+      "step": 1328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008567346376574746,
+      "loss": 2.0967,
+      "step": 1329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008565042538762669,
+      "loss": 2.1035,
+      "step": 1330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000856273716034885,
+      "loss": 2.1855,
+      "step": 1331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008560430242329536,
+      "loss": 2.0771,
+      "step": 1332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008558121785701643,
+      "loss": 2.1201,
+      "step": 1333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008555811791462746,
+      "loss": 2.1895,
+      "step": 1334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008553500260611094,
+      "loss": 2.1816,
+      "step": 1335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000855118719414559,
+      "loss": 2.0947,
+      "step": 1336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008548872593065811,
+      "loss": 2.1855,
+      "step": 1337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008546556458371987,
+      "loss": 2.1094,
+      "step": 1338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008544238791065016,
+      "loss": 2.1738,
+      "step": 1339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008541919592146458,
+      "loss": 2.123,
+      "step": 1340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008539598862618533,
+      "loss": 2.1582,
+      "step": 1341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008537276603484128,
+      "loss": 2.1104,
+      "step": 1342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008534952815746782,
+      "loss": 2.0547,
+      "step": 1343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008532627500410702,
+      "loss": 2.1348,
+      "step": 1344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008530300658480752,
+      "loss": 2.1084,
+      "step": 1345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008527972290962455,
+      "loss": 2.083,
+      "step": 1346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008525642398861997,
+      "loss": 2.1377,
+      "step": 1347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008523310983186219,
+      "loss": 2.1064,
+      "step": 1348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008520978044942622,
+      "loss": 2.0986,
+      "step": 1349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008518643585139364,
+      "loss": 2.0986,
+      "step": 1350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008516307604785261,
+      "loss": 2.1543,
+      "step": 1351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000851397010488979,
+      "loss": 2.1475,
+      "step": 1352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008511631086463074,
+      "loss": 2.1992,
+      "step": 1353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008509290550515907,
+      "loss": 2.1611,
+      "step": 1354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008506948498059726,
+      "loss": 2.0312,
+      "step": 1355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008504604930106628,
+      "loss": 2.1719,
+      "step": 1356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008502259847669371,
+      "loss": 2.0986,
+      "step": 1357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008499913251761353,
+      "loss": 2.0605,
+      "step": 1358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008497565143396643,
+      "loss": 2.125,
+      "step": 1359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849521552358995,
+      "loss": 2.1201,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008492864393356644,
+      "loss": 2.1426,
+      "step": 1361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008490511753712744,
+      "loss": 2.1387,
+      "step": 1362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008488157605674924,
+      "loss": 2.2305,
+      "step": 1363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008485801950260507,
+      "loss": 2.1162,
+      "step": 1364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008483444788487469,
+      "loss": 2.1416,
+      "step": 1365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008481086121374437,
+      "loss": 2.1182,
+      "step": 1366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008478725949940686,
+      "loss": 2.168,
+      "step": 1367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008476364275206149,
+      "loss": 2.1162,
+      "step": 1368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008474001098191398,
+      "loss": 2.1025,
+      "step": 1369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008471636419917659,
+      "loss": 2.1602,
+      "step": 1370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000846927024140681,
+      "loss": 2.1953,
+      "step": 1371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008466902563681371,
+      "loss": 2.125,
+      "step": 1372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008464533387764517,
+      "loss": 2.0547,
+      "step": 1373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008462162714680062,
+      "loss": 2.1064,
+      "step": 1374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008459790545452475,
+      "loss": 2.1094,
+      "step": 1375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008457416881106866,
+      "loss": 2.123,
+      "step": 1376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008455041722668994,
+      "loss": 2.2158,
+      "step": 1377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008452665071165262,
+      "loss": 2.1094,
+      "step": 1378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008450286927622721,
+      "loss": 2.1348,
+      "step": 1379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008447907293069063,
+      "loss": 1.9922,
+      "step": 1380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008445526168532626,
+      "loss": 2.1025,
+      "step": 1381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008443143555042393,
+      "loss": 2.1768,
+      "step": 1382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008440759453627989,
+      "loss": 2.1621,
+      "step": 1383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008438373865319683,
+      "loss": 2.0693,
+      "step": 1384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008435986791148385,
+      "loss": 2.2158,
+      "step": 1385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008433598232145647,
+      "loss": 2.0928,
+      "step": 1386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008431208189343669,
+      "loss": 2.1621,
+      "step": 1387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008428816663775281,
+      "loss": 2.1211,
+      "step": 1388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000842642365647396,
+      "loss": 2.1152,
+      "step": 1389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008424029168473828,
+      "loss": 2.1582,
+      "step": 1390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008421633200809637,
+      "loss": 2.2158,
+      "step": 1391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008419235754516788,
+      "loss": 2.0908,
+      "step": 1392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008416836830631311,
+      "loss": 2.1152,
+      "step": 1393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008414436430189882,
+      "loss": 2.1045,
+      "step": 1394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008412034554229814,
+      "loss": 2.1582,
+      "step": 1395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008409631203789056,
+      "loss": 2.1357,
+      "step": 1396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008407226379906197,
+      "loss": 2.1855,
+      "step": 1397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008404820083620455,
+      "loss": 2.1035,
+      "step": 1398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008402412315971692,
+      "loss": 2.1777,
+      "step": 1399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008400003078000407,
+      "loss": 2.1738,
+      "step": 1400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008397592370747726,
+      "loss": 2.0889,
+      "step": 1401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008395180195255419,
+      "loss": 2.127,
+      "step": 1402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008392766552565885,
+      "loss": 2.1221,
+      "step": 1403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008390351443722157,
+      "loss": 2.1699,
+      "step": 1404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008387934869767904,
+      "loss": 2.0342,
+      "step": 1405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008385516831747429,
+      "loss": 2.0869,
+      "step": 1406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008383097330705662,
+      "loss": 2.2363,
+      "step": 1407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008380676367688171,
+      "loss": 2.1465,
+      "step": 1408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008378253943741157,
+      "loss": 2.1035,
+      "step": 1409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008375830059911442,
+      "loss": 2.127,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008373404717246494,
+      "loss": 2.1865,
+      "step": 1411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008370977916794397,
+      "loss": 2.2129,
+      "step": 1412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008368549659603876,
+      "loss": 2.0654,
+      "step": 1413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008366119946724279,
+      "loss": 2.1357,
+      "step": 1414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008363688779205585,
+      "loss": 2.0469,
+      "step": 1415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008361256158098401,
+      "loss": 2.1543,
+      "step": 1416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008358822084453964,
+      "loss": 2.1631,
+      "step": 1417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008356386559324137,
+      "loss": 2.123,
+      "step": 1418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008353949583761411,
+      "loss": 2.1572,
+      "step": 1419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008351511158818904,
+      "loss": 2.168,
+      "step": 1420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008349071285550358,
+      "loss": 2.1367,
+      "step": 1421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008346629965010146,
+      "loss": 2.1621,
+      "step": 1422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008344187198253259,
+      "loss": 2.1719,
+      "step": 1423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008341742986335321,
+      "loss": 2.084,
+      "step": 1424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008339297330312574,
+      "loss": 2.167,
+      "step": 1425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008336850231241888,
+      "loss": 2.1396,
+      "step": 1426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008334401690180755,
+      "loss": 2.2148,
+      "step": 1427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008331951708187291,
+      "loss": 2.2051,
+      "step": 1428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008329500286320233,
+      "loss": 2.0791,
+      "step": 1429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008327047425638942,
+      "loss": 2.1777,
+      "step": 1430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00083245931272034,
+      "loss": 2.1836,
+      "step": 1431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008322137392074211,
+      "loss": 2.0967,
+      "step": 1432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00083196802213126,
+      "loss": 2.0088,
+      "step": 1433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000831722161598041,
+      "loss": 2.1094,
+      "step": 1434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008314761577140105,
+      "loss": 2.1172,
+      "step": 1435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008312300105854771,
+      "loss": 2.0596,
+      "step": 1436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008309837203188111,
+      "loss": 2.125,
+      "step": 1437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008307372870204446,
+      "loss": 2.1777,
+      "step": 1438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008304907107968716,
+      "loss": 2.1582,
+      "step": 1439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008302439917546478,
+      "loss": 2.1025,
+      "step": 1440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008299971300003908,
+      "loss": 2.1211,
+      "step": 1441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008297501256407795,
+      "loss": 2.1709,
+      "step": 1442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000829502978782555,
+      "loss": 2.0312,
+      "step": 1443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008292556895325194,
+      "loss": 2.0732,
+      "step": 1444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008290082579975364,
+      "loss": 2.1152,
+      "step": 1445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008287606842845319,
+      "loss": 2.126,
+      "step": 1446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008285129685004922,
+      "loss": 2.0977,
+      "step": 1447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008282651107524657,
+      "loss": 2.124,
+      "step": 1448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008280171111475619,
+      "loss": 2.0928,
+      "step": 1449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008277689697929516,
+      "loss": 2.1611,
+      "step": 1450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008275206867958671,
+      "loss": 2.1689,
+      "step": 1451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008272722622636015,
+      "loss": 2.1924,
+      "step": 1452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008270236963035092,
+      "loss": 2.1934,
+      "step": 1453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000826774989023006,
+      "loss": 2.2363,
+      "step": 1454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008265261405295685,
+      "loss": 2.1387,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008262771509307345,
+      "loss": 2.1211,
+      "step": 1456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008260280203341026,
+      "loss": 2.1064,
+      "step": 1457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008257787488473321,
+      "loss": 2.1387,
+      "step": 1458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000825529336578144,
+      "loss": 2.125,
+      "step": 1459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000825279783634319,
+      "loss": 2.0811,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008250300901236998,
+      "loss": 2.1016,
+      "step": 1461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008247802561541889,
+      "loss": 2.0791,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00082453028183375,
+      "loss": 2.1484,
+      "step": 1463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008242801672704072,
+      "loss": 2.0723,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008240299125722454,
+      "loss": 2.0811,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008237795178474098,
+      "loss": 2.125,
+      "step": 1466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008235289832041065,
+      "loss": 2.1982,
+      "step": 1467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008232783087506017,
+      "loss": 2.124,
+      "step": 1468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000823027494595222,
+      "loss": 2.1465,
+      "step": 1469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000822776540846355,
+      "loss": 2.1309,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008225254476124477,
+      "loss": 2.1035,
+      "step": 1471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008222742150020082,
+      "loss": 2.082,
+      "step": 1472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008220228431236041,
+      "loss": 2.1025,
+      "step": 1473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008217713320858637,
+      "loss": 2.1553,
+      "step": 1474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008215196819974754,
+      "loss": 2.1318,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008212678929671873,
+      "loss": 2.1348,
+      "step": 1476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008210159651038081,
+      "loss": 2.0732,
+      "step": 1477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008207638985162061,
+      "loss": 2.1084,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008205116933133094,
+      "loss": 2.1113,
+      "step": 1479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008202593496041067,
+      "loss": 2.1191,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008200068674976458,
+      "loss": 2.1289,
+      "step": 1481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008197542471030348,
+      "loss": 2.0566,
+      "step": 1482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008195014885294414,
+      "loss": 2.0684,
+      "step": 1483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008192485918860927,
+      "loss": 2.2266,
+      "step": 1484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008189955572822761,
+      "loss": 2.168,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008187423848273379,
+      "loss": 2.0635,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008184890746306848,
+      "loss": 2.1367,
+      "step": 1487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008182356268017822,
+      "loss": 2.0635,
+      "step": 1488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008179820414501556,
+      "loss": 2.0996,
+      "step": 1489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008177283186853895,
+      "loss": 2.0957,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008174744586171281,
+      "loss": 2.0537,
+      "step": 1491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008172204613550747,
+      "loss": 2.166,
+      "step": 1492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000816966327008992,
+      "loss": 2.1406,
+      "step": 1493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008167120556887018,
+      "loss": 2.1367,
+      "step": 1494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008164576475040854,
+      "loss": 2.0898,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008162031025650831,
+      "loss": 2.1514,
+      "step": 1496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008159484209816941,
+      "loss": 2.0439,
+      "step": 1497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008156936028639768,
+      "loss": 2.0625,
+      "step": 1498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008154386483220486,
+      "loss": 2.0918,
+      "step": 1499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008151835574660862,
+      "loss": 2.1318,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008149283304063243,
+      "loss": 2.0635,
+      "step": 1501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008146729672530574,
+      "loss": 2.165,
+      "step": 1502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008144174681166383,
+      "loss": 2.1133,
+      "step": 1503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008141618331074788,
+      "loss": 2.0732,
+      "step": 1504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008139060623360494,
+      "loss": 2.0723,
+      "step": 1505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008136501559128788,
+      "loss": 2.0342,
+      "step": 1506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008133941139485551,
+      "loss": 2.0957,
+      "step": 1507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008131379365537245,
+      "loss": 2.1494,
+      "step": 1508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008128816238390916,
+      "loss": 2.1201,
+      "step": 1509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008126251759154199,
+      "loss": 2.0811,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008123685928935312,
+      "loss": 2.0498,
+      "step": 1511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008121118748843052,
+      "loss": 2.1074,
+      "step": 1512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008118550219986807,
+      "loss": 2.0488,
+      "step": 1513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008115980343476542,
+      "loss": 2.0947,
+      "step": 1514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008113409120422808,
+      "loss": 2.1367,
+      "step": 1515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008110836551936735,
+      "loss": 2.1436,
+      "step": 1516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008108262639130037,
+      "loss": 2.1221,
+      "step": 1517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008105687383115008,
+      "loss": 2.1787,
+      "step": 1518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000810311078500452,
+      "loss": 2.0508,
+      "step": 1519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008100532845912028,
+      "loss": 2.123,
+      "step": 1520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008097953566951568,
+      "loss": 2.0,
+      "step": 1521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008095372949237751,
+      "loss": 2.0605,
+      "step": 1522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008092790993885768,
+      "loss": 2.1484,
+      "step": 1523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000809020770201139,
+      "loss": 2.083,
+      "step": 1524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000808762307473096,
+      "loss": 2.0918,
+      "step": 1525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008085037113161406,
+      "loss": 2.2012,
+      "step": 1526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008082449818420226,
+      "loss": 2.1006,
+      "step": 1527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008079861191625497,
+      "loss": 2.0674,
+      "step": 1528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008077271233895872,
+      "loss": 2.166,
+      "step": 1529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008074679946350578,
+      "loss": 2.1387,
+      "step": 1530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008072087330109418,
+      "loss": 2.082,
+      "step": 1531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008069493386292768,
+      "loss": 2.1064,
+      "step": 1532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008066898116021576,
+      "loss": 2.1895,
+      "step": 1533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008064301520417367,
+      "loss": 2.1348,
+      "step": 1534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008061703600602238,
+      "loss": 2.0879,
+      "step": 1535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008059104357698854,
+      "loss": 2.1113,
+      "step": 1536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008056503792830458,
+      "loss": 2.085,
+      "step": 1537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000805390190712086,
+      "loss": 2.1113,
+      "step": 1538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008051298701694441,
+      "loss": 2.1045,
+      "step": 1539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008048694177676156,
+      "loss": 2.0674,
+      "step": 1540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008046088336191524,
+      "loss": 2.0947,
+      "step": 1541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008043481178366638,
+      "loss": 2.1592,
+      "step": 1542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000804087270532816,
+      "loss": 2.1299,
+      "step": 1543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008038262918203314,
+      "loss": 2.0566,
+      "step": 1544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008035651818119902,
+      "loss": 2.1191,
+      "step": 1545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008033039406206282,
+      "loss": 2.1084,
+      "step": 1546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008030425683591391,
+      "loss": 2.1631,
+      "step": 1547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008027810651404721,
+      "loss": 2.1279,
+      "step": 1548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008025194310776338,
+      "loss": 2.0713,
+      "step": 1549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008022576662836871,
+      "loss": 2.1289,
+      "step": 1550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008019957708717512,
+      "loss": 2.1133,
+      "step": 1551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000801733744955002,
+      "loss": 2.0938,
+      "step": 1552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008014715886466716,
+      "loss": 2.0752,
+      "step": 1553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008012093020600486,
+      "loss": 2.0908,
+      "step": 1554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008009468853084775,
+      "loss": 2.1309,
+      "step": 1555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008006843385053601,
+      "loss": 2.0596,
+      "step": 1556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008004216617641532,
+      "loss": 2.1475,
+      "step": 1557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008001588551983703,
+      "loss": 2.0781,
+      "step": 1558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007998959189215811,
+      "loss": 2.1504,
+      "step": 1559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007996328530474108,
+      "loss": 2.1309,
+      "step": 1560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007993696576895415,
+      "loss": 2.0029,
+      "step": 1561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007991063329617106,
+      "loss": 2.1348,
+      "step": 1562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007988428789777112,
+      "loss": 2.1797,
+      "step": 1563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007985792958513931,
+      "loss": 2.0928,
+      "step": 1564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007983155836966613,
+      "loss": 2.1494,
+      "step": 1565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007980517426274765,
+      "loss": 2.1094,
+      "step": 1566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007977877727578555,
+      "loss": 2.1523,
+      "step": 1567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007975236742018704,
+      "loss": 2.1787,
+      "step": 1568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007972594470736492,
+      "loss": 2.3047,
+      "step": 1569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007969950914873753,
+      "loss": 2.1016,
+      "step": 1570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007967306075572875,
+      "loss": 2.0703,
+      "step": 1571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007964659953976805,
+      "loss": 2.0742,
+      "step": 1572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007962012551229038,
+      "loss": 2.1689,
+      "step": 1573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007959363868473625,
+      "loss": 2.0645,
+      "step": 1574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007956713906855178,
+      "loss": 2.0664,
+      "step": 1575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007954062667518845,
+      "loss": 2.1709,
+      "step": 1576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007951410151610343,
+      "loss": 2.1484,
+      "step": 1577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007948756360275929,
+      "loss": 2.0615,
+      "step": 1578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007946101294662418,
+      "loss": 2.0762,
+      "step": 1579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007943444955917173,
+      "loss": 2.1797,
+      "step": 1580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007940787345188106,
+      "loss": 2.0518,
+      "step": 1581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007938128463623682,
+      "loss": 2.123,
+      "step": 1582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007935468312372911,
+      "loss": 2.0674,
+      "step": 1583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007932806892585356,
+      "loss": 2.2012,
+      "step": 1584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007930144205411125,
+      "loss": 2.1104,
+      "step": 1585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007927480252000875,
+      "loss": 2.1816,
+      "step": 1586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000792481503350581,
+      "loss": 2.1816,
+      "step": 1587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007922148551077682,
+      "loss": 2.0469,
+      "step": 1588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007919480805868784,
+      "loss": 2.0869,
+      "step": 1589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007916811799031963,
+      "loss": 2.0576,
+      "step": 1590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007914141531720607,
+      "loss": 2.1309,
+      "step": 1591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007911470005088643,
+      "loss": 2.0908,
+      "step": 1592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007908797220290553,
+      "loss": 2.1191,
+      "step": 1593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007906123178481357,
+      "loss": 2.1621,
+      "step": 1594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007903447880816617,
+      "loss": 2.1445,
+      "step": 1595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007900771328452438,
+      "loss": 2.1006,
+      "step": 1596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007898093522545471,
+      "loss": 2.0469,
+      "step": 1597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007895414464252906,
+      "loss": 2.1309,
+      "step": 1598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007892734154732474,
+      "loss": 2.0938,
+      "step": 1599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007890052595142446,
+      "loss": 2.1348,
+      "step": 1600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007887369786641637,
+      "loss": 2.1113,
+      "step": 1601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007884685730389397,
+      "loss": 2.0664,
+      "step": 1602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007882000427545617,
+      "loss": 2.1396,
+      "step": 1603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007879313879270729,
+      "loss": 2.1104,
+      "step": 1604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00078766260867257,
+      "loss": 2.1328,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007873937051072036,
+      "loss": 2.0332,
+      "step": 1606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007871246773471779,
+      "loss": 2.0508,
+      "step": 1607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007868555255087509,
+      "loss": 2.0996,
+      "step": 1608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007865862497082344,
+      "loss": 2.2383,
+      "step": 1609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007863168500619934,
+      "loss": 2.0479,
+      "step": 1610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007860473266864467,
+      "loss": 2.0674,
+      "step": 1611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007857776796980663,
+      "loss": 2.1143,
+      "step": 1612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007855079092133778,
+      "loss": 2.1396,
+      "step": 1613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007852380153489602,
+      "loss": 2.1182,
+      "step": 1614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007849679982214458,
+      "loss": 2.2148,
+      "step": 1615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007846978579475201,
+      "loss": 2.0547,
+      "step": 1616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007844275946439216,
+      "loss": 2.1914,
+      "step": 1617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007841572084274427,
+      "loss": 2.085,
+      "step": 1618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000783886699414928,
+      "loss": 2.168,
+      "step": 1619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007836160677232758,
+      "loss": 2.1445,
+      "step": 1620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000783345313469437,
+      "loss": 2.1338,
+      "step": 1621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007830744367704159,
+      "loss": 1.9678,
+      "step": 1622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007828034377432694,
+      "loss": 2.0273,
+      "step": 1623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007825323165051072,
+      "loss": 2.0459,
+      "step": 1624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007822610731730922,
+      "loss": 2.1182,
+      "step": 1625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007819897078644396,
+      "loss": 2.167,
+      "step": 1626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007817182206964177,
+      "loss": 2.1445,
+      "step": 1627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007814466117863472,
+      "loss": 2.0576,
+      "step": 1628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007811748812516012,
+      "loss": 2.124,
+      "step": 1629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007809030292096064,
+      "loss": 2.1338,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007806310557778406,
+      "loss": 2.1562,
+      "step": 1631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007803589610738351,
+      "loss": 2.0693,
+      "step": 1632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000780086745215173,
+      "loss": 2.0889,
+      "step": 1633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007798144083194903,
+      "loss": 2.1318,
+      "step": 1634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007795419505044745,
+      "loss": 2.123,
+      "step": 1635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007792693718878662,
+      "loss": 2.1162,
+      "step": 1636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007789966725874577,
+      "loss": 2.1279,
+      "step": 1637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007787238527210937,
+      "loss": 2.1006,
+      "step": 1638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000778450912406671,
+      "loss": 2.0889,
+      "step": 1639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007781778517621379,
+      "loss": 2.082,
+      "step": 1640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007779046709054958,
+      "loss": 2.1289,
+      "step": 1641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000777631369954797,
+      "loss": 2.1113,
+      "step": 1642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007773579490281459,
+      "loss": 2.1426,
+      "step": 1643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007770844082436996,
+      "loss": 2.1094,
+      "step": 1644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007768107477196658,
+      "loss": 2.1133,
+      "step": 1645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007765369675743046,
+      "loss": 2.0654,
+      "step": 1646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007762630679259279,
+      "loss": 2.0791,
+      "step": 1647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007759890488928987,
+      "loss": 2.1025,
+      "step": 1648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007757149105936321,
+      "loss": 2.0957,
+      "step": 1649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007754406531465945,
+      "loss": 2.1465,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007751662766703038,
+      "loss": 2.0859,
+      "step": 1651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007748917812833296,
+      "loss": 2.041,
+      "step": 1652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007746171671042924,
+      "loss": 2.0693,
+      "step": 1653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007743424342518643,
+      "loss": 2.0762,
+      "step": 1654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000774067582844769,
+      "loss": 2.0693,
+      "step": 1655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007737926130017808,
+      "loss": 2.0957,
+      "step": 1656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007735175248417255,
+      "loss": 2.084,
+      "step": 1657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007732423184834803,
+      "loss": 1.9834,
+      "step": 1658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007729669940459729,
+      "loss": 2.1357,
+      "step": 1659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007726915516481824,
+      "loss": 2.1328,
+      "step": 1660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007724159914091388,
+      "loss": 2.082,
+      "step": 1661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007721403134479234,
+      "loss": 2.0693,
+      "step": 1662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007718645178836675,
+      "loss": 2.0986,
+      "step": 1663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000771588604835554,
+      "loss": 2.0898,
+      "step": 1664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000771312574422816,
+      "loss": 2.04,
+      "step": 1665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000771036426764738,
+      "loss": 2.0801,
+      "step": 1666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007707601619806547,
+      "loss": 2.0488,
+      "step": 1667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007704837801899512,
+      "loss": 2.0205,
+      "step": 1668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007702072815120637,
+      "loss": 2.1338,
+      "step": 1669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007699306660664787,
+      "loss": 2.0557,
+      "step": 1670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007696539339727332,
+      "loss": 2.1152,
+      "step": 1671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007693770853504143,
+      "loss": 2.1191,
+      "step": 1672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007691001203191597,
+      "loss": 2.0693,
+      "step": 1673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007688230389986575,
+      "loss": 2.0723,
+      "step": 1674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007685458415086459,
+      "loss": 2.1514,
+      "step": 1675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682685279689134,
+      "loss": 2.1104,
+      "step": 1676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007679910984992988,
+      "loss": 2.0977,
+      "step": 1677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007677135532196904,
+      "loss": 1.9971,
+      "step": 1678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007674358922500273,
+      "loss": 2.1104,
+      "step": 1679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000767158115710298,
+      "loss": 2.1299,
+      "step": 1680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007668802237205412,
+      "loss": 2.0566,
+      "step": 1681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007666022164008457,
+      "loss": 2.1494,
+      "step": 1682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007663240938713497,
+      "loss": 2.0352,
+      "step": 1683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007660458562522414,
+      "loss": 2.1094,
+      "step": 1684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000765767503663759,
+      "loss": 2.0635,
+      "step": 1685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00076548903622619,
+      "loss": 2.0957,
+      "step": 1686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007652104540598712,
+      "loss": 2.0254,
+      "step": 1687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007649317572851901,
+      "loss": 1.9375,
+      "step": 1688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007646529460225827,
+      "loss": 2.0107,
+      "step": 1689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007643740203925352,
+      "loss": 2.1836,
+      "step": 1690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000764094980515582,
+      "loss": 2.1152,
+      "step": 1691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007638158265123085,
+      "loss": 2.0732,
+      "step": 1692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007635365585033487,
+      "loss": 2.1387,
+      "step": 1693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007632571766093854,
+      "loss": 2.123,
+      "step": 1694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000762977680951151,
+      "loss": 2.0664,
+      "step": 1695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007626980716494274,
+      "loss": 2.0947,
+      "step": 1696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007624183488250451,
+      "loss": 2.0918,
+      "step": 1697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007621385125988839,
+      "loss": 2.1113,
+      "step": 1698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007618585630918726,
+      "loss": 2.1494,
+      "step": 1699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007615785004249888,
+      "loss": 2.1543,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007612983247192593,
+      "loss": 2.1113,
+      "step": 1701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007610180360957594,
+      "loss": 2.0693,
+      "step": 1702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007607376346756137,
+      "loss": 2.0254,
+      "step": 1703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007604571205799946,
+      "loss": 2.0752,
+      "step": 1704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007601764939301241,
+      "loss": 2.0566,
+      "step": 1705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007598957548472728,
+      "loss": 2.0977,
+      "step": 1706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007596149034527593,
+      "loss": 2.1064,
+      "step": 1707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007593339398679511,
+      "loss": 2.0449,
+      "step": 1708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007590528642142642,
+      "loss": 2.0518,
+      "step": 1709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000758771676613163,
+      "loss": 2.0654,
+      "step": 1710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00075849037718616,
+      "loss": 2.1904,
+      "step": 1711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007582089660548166,
+      "loss": 2.002,
+      "step": 1712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007579274433407422,
+      "loss": 2.2383,
+      "step": 1713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000757645809165594,
+      "loss": 2.0811,
+      "step": 1714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007573640636510776,
+      "loss": 2.0898,
+      "step": 1715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007570822069189475,
+      "loss": 2.1406,
+      "step": 1716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007568002390910054,
+      "loss": 2.0625,
+      "step": 1717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007565181602891009,
+      "loss": 2.1104,
+      "step": 1718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007562359706351318,
+      "loss": 2.0557,
+      "step": 1719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007559536702510443,
+      "loss": 2.2188,
+      "step": 1720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007556712592588318,
+      "loss": 2.0957,
+      "step": 1721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007553887377805358,
+      "loss": 2.0781,
+      "step": 1722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007551061059382454,
+      "loss": 2.0752,
+      "step": 1723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007548233638540975,
+      "loss": 2.0527,
+      "step": 1724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007545405116502767,
+      "loss": 1.96,
+      "step": 1725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007542575494490148,
+      "loss": 2.0166,
+      "step": 1726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007539744773725916,
+      "loss": 2.0566,
+      "step": 1727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007536912955433344,
+      "loss": 2.0723,
+      "step": 1728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007534080040836173,
+      "loss": 2.1016,
+      "step": 1729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007531246031158627,
+      "loss": 2.0791,
+      "step": 1730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007528410927625396,
+      "loss": 2.0488,
+      "step": 1731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007525574731461642,
+      "loss": 2.0889,
+      "step": 1732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007522737443893007,
+      "loss": 2.0547,
+      "step": 1733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007519899066145597,
+      "loss": 2.0879,
+      "step": 1734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007517059599445995,
+      "loss": 2.0928,
+      "step": 1735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007514219045021246,
+      "loss": 2.1699,
+      "step": 1736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007511377404098876,
+      "loss": 2.1309,
+      "step": 1737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007508534677906874,
+      "loss": 2.0469,
+      "step": 1738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007505690867673697,
+      "loss": 2.1045,
+      "step": 1739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007502845974628274,
+      "loss": 2.1006,
+      "step": 1740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00075,
+      "loss": 2.0566,
+      "step": 1741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007497152945018739,
+      "loss": 2.0264,
+      "step": 1742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007494304810914819,
+      "loss": 2.0664,
+      "step": 1743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007491455598919037,
+      "loss": 2.1064,
+      "step": 1744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007488605310262655,
+      "loss": 2.1621,
+      "step": 1745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007485753946177401,
+      "loss": 2.1299,
+      "step": 1746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007482901507895463,
+      "loss": 2.0098,
+      "step": 1747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007480047996649502,
+      "loss": 2.0449,
+      "step": 1748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007477193413672637,
+      "loss": 2.0518,
+      "step": 1749
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007474337760198447,
+      "loss": 2.0479,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007471481037460981,
+      "loss": 2.1396,
+      "step": 1751
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007468623246694746,
+      "loss": 2.0381,
+      "step": 1752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007465764389134711,
+      "loss": 2.0518,
+      "step": 1753
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007462904466016305,
+      "loss": 2.0586,
+      "step": 1754
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000746004347857542,
+      "loss": 2.0938,
+      "step": 1755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007457181428048405,
+      "loss": 2.0615,
+      "step": 1756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007454318315672073,
+      "loss": 2.1436,
+      "step": 1757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007451454142683689,
+      "loss": 2.1289,
+      "step": 1758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007448588910320983,
+      "loss": 2.0459,
+      "step": 1759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007445722619822137,
+      "loss": 2.124,
+      "step": 1760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007442855272425797,
+      "loss": 2.0381,
+      "step": 1761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000743998686937106,
+      "loss": 2.1562,
+      "step": 1762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007437117411897481,
+      "loss": 2.1152,
+      "step": 1763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007434246901245069,
+      "loss": 2.1475,
+      "step": 1764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007431375338654295,
+      "loss": 2.123,
+      "step": 1765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007428502725366074,
+      "loss": 2.1025,
+      "step": 1766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007425629062621784,
+      "loss": 2.1201,
+      "step": 1767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007422754351663251,
+      "loss": 2.0996,
+      "step": 1768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007419878593732757,
+      "loss": 2.042,
+      "step": 1769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007417001790073035,
+      "loss": 2.1074,
+      "step": 1770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007414123941927272,
+      "loss": 2.1172,
+      "step": 1771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007411245050539103,
+      "loss": 2.0293,
+      "step": 1772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007408365117152617,
+      "loss": 2.127,
+      "step": 1773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000740548414301235,
+      "loss": 2.085,
+      "step": 1774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007402602129363288,
+      "loss": 2.1309,
+      "step": 1775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007399719077450875,
+      "loss": 2.0889,
+      "step": 1776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007396834988520988,
+      "loss": 2.0449,
+      "step": 1777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007393949863819966,
+      "loss": 2.1748,
+      "step": 1778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007391063704594589,
+      "loss": 2.0957,
+      "step": 1779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007388176512092084,
+      "loss": 2.084,
+      "step": 1780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007385288287560128,
+      "loss": 2.0576,
+      "step": 1781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007382399032246841,
+      "loss": 2.1055,
+      "step": 1782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007379508747400787,
+      "loss": 2.0488,
+      "step": 1783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007376617434270981,
+      "loss": 2.0781,
+      "step": 1784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007373725094106875,
+      "loss": 2.1387,
+      "step": 1785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007370831728158371,
+      "loss": 2.0586,
+      "step": 1786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000736793733767581,
+      "loss": 2.0557,
+      "step": 1787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007365041923909978,
+      "loss": 2.0342,
+      "step": 1788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007362145488112102,
+      "loss": 2.0898,
+      "step": 1789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007359248031533852,
+      "loss": 2.1289,
+      "step": 1790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007356349555427338,
+      "loss": 2.0635,
+      "step": 1791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000735345006104511,
+      "loss": 2.167,
+      "step": 1792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000735054954964016,
+      "loss": 2.0732,
+      "step": 1793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007347648022465919,
+      "loss": 2.1084,
+      "step": 1794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007344745480776256,
+      "loss": 2.1504,
+      "step": 1795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007341841925825478,
+      "loss": 2.1904,
+      "step": 1796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007338937358868332,
+      "loss": 1.9873,
+      "step": 1797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000733603178116,
+      "loss": 2.0586,
+      "step": 1798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007333125193956101,
+      "loss": 2.0654,
+      "step": 1799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007330217598512695,
+      "loss": 2.1152,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000732730899608627,
+      "loss": 2.0352,
+      "step": 1801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007324399387933753,
+      "loss": 2.001,
+      "step": 1802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007321488775312506,
+      "loss": 2.1543,
+      "step": 1803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007318577159480327,
+      "loss": 2.1182,
+      "step": 1804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007315664541695441,
+      "loss": 2.0508,
+      "step": 1805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007312750923216514,
+      "loss": 2.0693,
+      "step": 1806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007309836305302637,
+      "loss": 2.0947,
+      "step": 1807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007306920689213343,
+      "loss": 2.1123,
+      "step": 1808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000730400407620858,
+      "loss": 2.0684,
+      "step": 1809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007301086467548744,
+      "loss": 2.0381,
+      "step": 1810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007298167864494652,
+      "loss": 2.1025,
+      "step": 1811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007295248268307553,
+      "loss": 2.0518,
+      "step": 1812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007292327680249122,
+      "loss": 2.085,
+      "step": 1813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000728940610158147,
+      "loss": 2.0,
+      "step": 1814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007286483533567127,
+      "loss": 1.9971,
+      "step": 1815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007283559977469059,
+      "loss": 2.1104,
+      "step": 1816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007280635434550652,
+      "loss": 2.0518,
+      "step": 1817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007277709906075722,
+      "loss": 2.0225,
+      "step": 1818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007274783393308513,
+      "loss": 2.0684,
+      "step": 1819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007271855897513687,
+      "loss": 1.9697,
+      "step": 1820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000726892741995634,
+      "loss": 2.0303,
+      "step": 1821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007265997961901987,
+      "loss": 2.0547,
+      "step": 1822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007263067524616565,
+      "loss": 2.0576,
+      "step": 1823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007260136109366439,
+      "loss": 2.0068,
+      "step": 1824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007257203717418393,
+      "loss": 2.0332,
+      "step": 1825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007254270350039633,
+      "loss": 2.0518,
+      "step": 1826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007251336008497793,
+      "loss": 2.0615,
+      "step": 1827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007248400694060917,
+      "loss": 2.0742,
+      "step": 1828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000724546440799748,
+      "loss": 2.1357,
+      "step": 1829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007242527151576368,
+      "loss": 2.0615,
+      "step": 1830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007239588926066893,
+      "loss": 2.0439,
+      "step": 1831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007236649732738784,
+      "loss": 2.166,
+      "step": 1832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007233709572862187,
+      "loss": 2.0605,
+      "step": 1833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007230768447707663,
+      "loss": 2.1035,
+      "step": 1834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007227826358546199,
+      "loss": 2.1084,
+      "step": 1835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007224883306649188,
+      "loss": 2.0713,
+      "step": 1836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007221939293288444,
+      "loss": 2.1494,
+      "step": 1837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007218994319736201,
+      "loss": 1.9648,
+      "step": 1838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007216048387265099,
+      "loss": 1.9951,
+      "step": 1839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00072131014971482,
+      "loss": 2.0791,
+      "step": 1840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007210153650658974,
+      "loss": 2.0635,
+      "step": 1841
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007207204849071308,
+      "loss": 2.0254,
+      "step": 1842
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007204255093659501,
+      "loss": 2.0264,
+      "step": 1843
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007201304385698263,
+      "loss": 2.0703,
+      "step": 1844
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007198352726462717,
+      "loss": 2.0283,
+      "step": 1845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007195400117228397,
+      "loss": 2.0342,
+      "step": 1846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007192446559271246,
+      "loss": 2.1357,
+      "step": 1847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007189492053867618,
+      "loss": 1.9854,
+      "step": 1848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007186536602294277,
+      "loss": 2.083,
+      "step": 1849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007183580205828396,
+      "loss": 2.001,
+      "step": 1850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007180622865747555,
+      "loss": 2.0576,
+      "step": 1851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007177664583329742,
+      "loss": 2.1025,
+      "step": 1852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007174705359853355,
+      "loss": 2.1592,
+      "step": 1853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007171745196597193,
+      "loss": 2.0293,
+      "step": 1854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007168784094840467,
+      "loss": 2.0938,
+      "step": 1855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007165822055862791,
+      "loss": 2.1504,
+      "step": 1856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007162859080944182,
+      "loss": 2.0195,
+      "step": 1857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007159895171365065,
+      "loss": 2.0332,
+      "step": 1858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007156930328406268,
+      "loss": 2.1592,
+      "step": 1859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000715396455334902,
+      "loss": 2.0986,
+      "step": 1860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007150997847474956,
+      "loss": 2.0615,
+      "step": 1861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007148030212066112,
+      "loss": 2.1289,
+      "step": 1862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007145061648404923,
+      "loss": 2.1758,
+      "step": 1863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007142092157774232,
+      "loss": 2.0459,
+      "step": 1864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007139121741457275,
+      "loss": 2.0635,
+      "step": 1865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007136150400737691,
+      "loss": 2.0273,
+      "step": 1866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007133178136899522,
+      "loss": 2.084,
+      "step": 1867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00071302049512272,
+      "loss": 2.1279,
+      "step": 1868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007127230845005568,
+      "loss": 1.9531,
+      "step": 1869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007124255819519854,
+      "loss": 2.0361,
+      "step": 1870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007121279876055693,
+      "loss": 2.041,
+      "step": 1871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007118303015899109,
+      "loss": 2.0449,
+      "step": 1872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000711532524033653,
+      "loss": 2.0762,
+      "step": 1873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007112346550654774,
+      "loss": 1.9893,
+      "step": 1874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007109366948141052,
+      "loss": 2.0596,
+      "step": 1875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007106386434082979,
+      "loss": 2.0498,
+      "step": 1876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007103405009768553,
+      "loss": 2.1025,
+      "step": 1877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007100422676486174,
+      "loss": 2.0742,
+      "step": 1878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007097439435524627,
+      "loss": 2.0654,
+      "step": 1879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007094455288173097,
+      "loss": 2.1318,
+      "step": 1880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007091470235721155,
+      "loss": 2.0928,
+      "step": 1881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007088484279458766,
+      "loss": 2.0352,
+      "step": 1882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007085497420676285,
+      "loss": 2.0928,
+      "step": 1883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007082509660664456,
+      "loss": 2.1504,
+      "step": 1884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007079521000714412,
+      "loss": 2.0947,
+      "step": 1885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007076531442117679,
+      "loss": 2.0488,
+      "step": 1886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007073540986166166,
+      "loss": 2.1133,
+      "step": 1887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007070549634152172,
+      "loss": 2.0869,
+      "step": 1888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007067557387368386,
+      "loss": 2.0947,
+      "step": 1889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000706456424710788,
+      "loss": 2.042,
+      "step": 1890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007061570214664112,
+      "loss": 2.0234,
+      "step": 1891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007058575291330928,
+      "loss": 2.0615,
+      "step": 1892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007055579478402556,
+      "loss": 2.1006,
+      "step": 1893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007052582777173613,
+      "loss": 2.0518,
+      "step": 1894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007049585188939095,
+      "loss": 2.0801,
+      "step": 1895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007046586714994383,
+      "loss": 2.0889,
+      "step": 1896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007043587356635242,
+      "loss": 2.0762,
+      "step": 1897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000704058711515782,
+      "loss": 2.1025,
+      "step": 1898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007037585991858642,
+      "loss": 2.0547,
+      "step": 1899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007034583988034618,
+      "loss": 2.1924,
+      "step": 1900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007031581104983038,
+      "loss": 2.0889,
+      "step": 1901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007028577344001573,
+      "loss": 2.0977,
+      "step": 1902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007025572706388268,
+      "loss": 2.1211,
+      "step": 1903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007022567193441553,
+      "loss": 2.041,
+      "step": 1904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007019560806460235,
+      "loss": 2.0967,
+      "step": 1905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007016553546743495,
+      "loss": 2.0605,
+      "step": 1906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007013545415590896,
+      "loss": 2.1602,
+      "step": 1907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007010536414302375,
+      "loss": 1.9336,
+      "step": 1908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007007526544178243,
+      "loss": 1.9941,
+      "step": 1909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007004515806519193,
+      "loss": 2.0254,
+      "step": 1910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007001504202626285,
+      "loss": 2.0586,
+      "step": 1911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000699849173380096,
+      "loss": 2.0889,
+      "step": 1912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006995478401345028,
+      "loss": 2.082,
+      "step": 1913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006992464206560675,
+      "loss": 2.125,
+      "step": 1914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000698944915075046,
+      "loss": 2.0879,
+      "step": 1915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000698643323521731,
+      "loss": 2.0439,
+      "step": 1916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006983416461264527,
+      "loss": 2.0264,
+      "step": 1917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006980398830195785,
+      "loss": 2.1162,
+      "step": 1918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006977380343315125,
+      "loss": 2.0264,
+      "step": 1919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000697436100192696,
+      "loss": 2.0156,
+      "step": 1920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006971340807336072,
+      "loss": 2.0205,
+      "step": 1921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006968319760847613,
+      "loss": 2.0381,
+      "step": 1922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006965297863767097,
+      "loss": 1.9639,
+      "step": 1923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006962275117400415,
+      "loss": 2.0625,
+      "step": 1924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006959251523053819,
+      "loss": 2.127,
+      "step": 1925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006956227082033929,
+      "loss": 2.0625,
+      "step": 1926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006953201795647728,
+      "loss": 2.0996,
+      "step": 1927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006950175665202568,
+      "loss": 2.0781,
+      "step": 1928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006947148692006169,
+      "loss": 2.0713,
+      "step": 1929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006944120877366604,
+      "loss": 2.0605,
+      "step": 1930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006941092222592319,
+      "loss": 1.9629,
+      "step": 1931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006938062728992123,
+      "loss": 2.0664,
+      "step": 1932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006935032397875182,
+      "loss": 1.9453,
+      "step": 1933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006932001230551028,
+      "loss": 2.0332,
+      "step": 1934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006928969228329554,
+      "loss": 2.1221,
+      "step": 1935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000692593639252101,
+      "loss": 2.1445,
+      "step": 1936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006922902724436016,
+      "loss": 2.0703,
+      "step": 1937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006919868225385536,
+      "loss": 2.1719,
+      "step": 1938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006916832896680908,
+      "loss": 2.0361,
+      "step": 1939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006913796739633822,
+      "loss": 2.0605,
+      "step": 1940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006910759755556325,
+      "loss": 2.0449,
+      "step": 1941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006907721945760823,
+      "loss": 1.9736,
+      "step": 1942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006904683311560081,
+      "loss": 2.0391,
+      "step": 1943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006901643854267214,
+      "loss": 2.0176,
+      "step": 1944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006898603575195701,
+      "loss": 2.0771,
+      "step": 1945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000689556247565937,
+      "loss": 2.0186,
+      "step": 1946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006892520556972403,
+      "loss": 2.0186,
+      "step": 1947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006889477820449342,
+      "loss": 2.1201,
+      "step": 1948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006886434267405078,
+      "loss": 2.0928,
+      "step": 1949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006883389899154856,
+      "loss": 2.1406,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006880344717014271,
+      "loss": 1.9775,
+      "step": 1951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000687729872229927,
+      "loss": 2.0918,
+      "step": 1952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000687425191632616,
+      "loss": 2.0576,
+      "step": 1953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006871204300411584,
+      "loss": 2.1602,
+      "step": 1954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006868155875872546,
+      "loss": 2.2031,
+      "step": 1955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006865106644026396,
+      "loss": 2.1123,
+      "step": 1956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006862056606190829,
+      "loss": 2.0195,
+      "step": 1957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006859005763683899,
+      "loss": 2.0977,
+      "step": 1958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006855954117823993,
+      "loss": 2.0977,
+      "step": 1959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006852901669929856,
+      "loss": 2.0986,
+      "step": 1960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000684984842132058,
+      "loss": 2.1445,
+      "step": 1961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006846794373315593,
+      "loss": 2.0752,
+      "step": 1962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000684373952723468,
+      "loss": 2.0508,
+      "step": 1963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006840683884397965,
+      "loss": 2.0791,
+      "step": 1964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006837627446125916,
+      "loss": 2.0693,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006834570213739346,
+      "loss": 2.0264,
+      "step": 1966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000683151218855941,
+      "loss": 2.041,
+      "step": 1967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000682845337190761,
+      "loss": 2.1426,
+      "step": 1968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006825393765105782,
+      "loss": 2.001,
+      "step": 1969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006822333369476113,
+      "loss": 2.0664,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006819272186341123,
+      "loss": 2.1562,
+      "step": 1971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006816210217023675,
+      "loss": 2.1113,
+      "step": 1972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006813147462846974,
+      "loss": 2.124,
+      "step": 1973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006810083925134561,
+      "loss": 1.9893,
+      "step": 1974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006807019605210319,
+      "loss": 1.9541,
+      "step": 1975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006803954504398462,
+      "loss": 1.9932,
+      "step": 1976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006800888624023553,
+      "loss": 2.0771,
+      "step": 1977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000679782196541048,
+      "loss": 2.1074,
+      "step": 1978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006794754529884473,
+      "loss": 2.0146,
+      "step": 1979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00067916863187711,
+      "loss": 2.0332,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000678861733339626,
+      "loss": 2.0576,
+      "step": 1981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006785547575086187,
+      "loss": 2.0312,
+      "step": 1982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006782477045167451,
+      "loss": 2.0137,
+      "step": 1983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006779405744966955,
+      "loss": 2.083,
+      "step": 1984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006776333675811934,
+      "loss": 2.085,
+      "step": 1985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006773260839029955,
+      "loss": 2.0908,
+      "step": 1986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006770187235948916,
+      "loss": 2.0312,
+      "step": 1987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006767112867897051,
+      "loss": 2.082,
+      "step": 1988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000676403773620292,
+      "loss": 2.0664,
+      "step": 1989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006760961842195413,
+      "loss": 2.0312,
+      "step": 1990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006757885187203748,
+      "loss": 2.0859,
+      "step": 1991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006754807772557479,
+      "loss": 2.04,
+      "step": 1992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006751729599586483,
+      "loss": 2.0674,
+      "step": 1993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006748650669620964,
+      "loss": 2.0254,
+      "step": 1994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006745570983991454,
+      "loss": 2.0488,
+      "step": 1995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006742490544028814,
+      "loss": 2.1641,
+      "step": 1996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006739409351064226,
+      "loss": 1.9443,
+      "step": 1997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006736327406429202,
+      "loss": 2.0371,
+      "step": 1998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000673324471145558,
+      "loss": 2.0449,
+      "step": 1999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006730161267475515,
+      "loss": 2.1338,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006727077075821493,
+      "loss": 1.9824,
+      "step": 2001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006723992137826319,
+      "loss": 2.042,
+      "step": 2002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006720906454823121,
+      "loss": 2.1094,
+      "step": 2003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006717820028145352,
+      "loss": 2.0693,
+      "step": 2004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006714732859126782,
+      "loss": 2.0605,
+      "step": 2005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006711644949101505,
+      "loss": 2.0225,
+      "step": 2006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006708556299403935,
+      "loss": 2.0635,
+      "step": 2007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006705466911368803,
+      "loss": 2.042,
+      "step": 2008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006702376786331164,
+      "loss": 2.043,
+      "step": 2009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006699285925626383,
+      "loss": 2.0576,
+      "step": 2010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006696194330590151,
+      "loss": 2.0928,
+      "step": 2011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006693102002558475,
+      "loss": 2.0303,
+      "step": 2012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006690008942867677,
+      "loss": 2.0332,
+      "step": 2013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006686915152854393,
+      "loss": 2.0342,
+      "step": 2014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000668382063385558,
+      "loss": 2.0859,
+      "step": 2015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006680725387208505,
+      "loss": 2.0312,
+      "step": 2016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006677629414250753,
+      "loss": 1.9346,
+      "step": 2017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006674532716320219,
+      "loss": 2.1807,
+      "step": 2018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006671435294755118,
+      "loss": 2.0371,
+      "step": 2019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000666833715089397,
+      "loss": 2.0703,
+      "step": 2020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006665238286075612,
+      "loss": 2.1387,
+      "step": 2021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006662138701639191,
+      "loss": 2.04,
+      "step": 2022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006659038398924165,
+      "loss": 2.1367,
+      "step": 2023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006655937379270303,
+      "loss": 2.0293,
+      "step": 2024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006652835644017682,
+      "loss": 2.1064,
+      "step": 2025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000664973319450669,
+      "loss": 2.1016,
+      "step": 2026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006646630032078024,
+      "loss": 2.0469,
+      "step": 2027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006643526158072688,
+      "loss": 2.0596,
+      "step": 2028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006640421573831994,
+      "loss": 2.1494,
+      "step": 2029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000663731628069756,
+      "loss": 2.0723,
+      "step": 2030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006634210280011312,
+      "loss": 2.0898,
+      "step": 2031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000663110357311548,
+      "loss": 2.0918,
+      "step": 2032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00066279961613526,
+      "loss": 2.0254,
+      "step": 2033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006624888046065516,
+      "loss": 2.0693,
+      "step": 2034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006621779228597368,
+      "loss": 2.0244,
+      "step": 2035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006618669710291606,
+      "loss": 2.0576,
+      "step": 2036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000661555949249198,
+      "loss": 2.1299,
+      "step": 2037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006612448576542544,
+      "loss": 2.1006,
+      "step": 2038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006609336963787655,
+      "loss": 2.043,
+      "step": 2039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006606224655571966,
+      "loss": 2.0625,
+      "step": 2040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006603111653240433,
+      "loss": 2.0566,
+      "step": 2041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006599997958138315,
+      "loss": 2.1348,
+      "step": 2042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006596883571611168,
+      "loss": 2.0859,
+      "step": 2043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006593768495004848,
+      "loss": 2.0352,
+      "step": 2044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006590652729665503,
+      "loss": 2.127,
+      "step": 2045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006587536276939585,
+      "loss": 2.0312,
+      "step": 2046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006584419138173845,
+      "loss": 1.9766,
+      "step": 2047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006581301314715325,
+      "loss": 2.0508,
+      "step": 2048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006578182807911363,
+      "loss": 2.0371,
+      "step": 2049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006575063619109599,
+      "loss": 2.0537,
+      "step": 2050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006571943749657959,
+      "loss": 2.0889,
+      "step": 2051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006568823200904667,
+      "loss": 2.0166,
+      "step": 2052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006565701974198244,
+      "loss": 2.0801,
+      "step": 2053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006562580070887498,
+      "loss": 2.0078,
+      "step": 2054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006559457492321533,
+      "loss": 2.0781,
+      "step": 2055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006556334239849743,
+      "loss": 1.9902,
+      "step": 2056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006553210314821814,
+      "loss": 1.9727,
+      "step": 2057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006550085718587724,
+      "loss": 2.084,
+      "step": 2058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006546960452497738,
+      "loss": 1.9941,
+      "step": 2059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006543834517902415,
+      "loss": 1.9961,
+      "step": 2060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006540707916152598,
+      "loss": 2.0566,
+      "step": 2061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006537580648599419,
+      "loss": 2.0879,
+      "step": 2062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006534452716594301,
+      "loss": 2.0918,
+      "step": 2063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006531324121488954,
+      "loss": 2.0381,
+      "step": 2064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000652819486463537,
+      "loss": 2.0586,
+      "step": 2065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006525064947385832,
+      "loss": 2.0303,
+      "step": 2066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006521934371092901,
+      "loss": 2.0947,
+      "step": 2067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006518803137109437,
+      "loss": 2.0801,
+      "step": 2068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006515671246788567,
+      "loss": 2.0625,
+      "step": 2069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006512538701483712,
+      "loss": 1.9912,
+      "step": 2070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006509405502548578,
+      "loss": 2.0078,
+      "step": 2071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006506271651337144,
+      "loss": 2.0127,
+      "step": 2072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006503137149203679,
+      "loss": 2.0215,
+      "step": 2073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000650000199750273,
+      "loss": 2.1064,
+      "step": 2074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006496866197589125,
+      "loss": 2.0947,
+      "step": 2075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006493729750817974,
+      "loss": 2.0674,
+      "step": 2076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000649059265854466,
+      "loss": 2.1055,
+      "step": 2077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006487454922124854,
+      "loss": 2.0371,
+      "step": 2078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006484316542914502,
+      "loss": 1.9561,
+      "step": 2079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006481177522269824,
+      "loss": 2.0049,
+      "step": 2080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006478037861547321,
+      "loss": 2.1172,
+      "step": 2081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006474897562103771,
+      "loss": 2.0713,
+      "step": 2082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006471756625296225,
+      "loss": 2.0449,
+      "step": 2083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006468615052482011,
+      "loss": 2.0479,
+      "step": 2084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006465472845018735,
+      "loss": 2.0879,
+      "step": 2085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006462330004264272,
+      "loss": 2.0762,
+      "step": 2086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006459186531576771,
+      "loss": 2.167,
+      "step": 2087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006456042428314661,
+      "loss": 2.0244,
+      "step": 2088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006452897695836635,
+      "loss": 1.9688,
+      "step": 2089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006449752335501662,
+      "loss": 2.0557,
+      "step": 2090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006446606348668982,
+      "loss": 2.0137,
+      "step": 2091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006443459736698105,
+      "loss": 2.1084,
+      "step": 2092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006440312500948814,
+      "loss": 2.0898,
+      "step": 2093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006437164642781155,
+      "loss": 2.0107,
+      "step": 2094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006434016163555452,
+      "loss": 2.0244,
+      "step": 2095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006430867064632289,
+      "loss": 2.0938,
+      "step": 2096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006427717347372523,
+      "loss": 1.9639,
+      "step": 2097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006424567013137276,
+      "loss": 2.1016,
+      "step": 2098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006421416063287937,
+      "loss": 1.9219,
+      "step": 2099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006418264499186162,
+      "loss": 2.0166,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641511232219387,
+      "loss": 2.1084,
+      "step": 2101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641195953367325,
+      "loss": 1.9727,
+      "step": 2102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006408806134986749,
+      "loss": 2.0195,
+      "step": 2103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000640565212749708,
+      "loss": 2.0039,
+      "step": 2104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000640249751256722,
+      "loss": 2.0352,
+      "step": 2105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006399342291560411,
+      "loss": 2.0576,
+      "step": 2106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006396186465840152,
+      "loss": 2.041,
+      "step": 2107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006393030036770203,
+      "loss": 1.9531,
+      "step": 2108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000638987300571459,
+      "loss": 2.0312,
+      "step": 2109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006386715374037595,
+      "loss": 2.0303,
+      "step": 2110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006383557143103762,
+      "loss": 2.1758,
+      "step": 2111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006380398314277889,
+      "loss": 2.0615,
+      "step": 2112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000637723888892504,
+      "loss": 2.0049,
+      "step": 2113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006374078868410534,
+      "loss": 2.0635,
+      "step": 2114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006370918254099939,
+      "loss": 2.0605,
+      "step": 2115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006367757047359093,
+      "loss": 2.0273,
+      "step": 2116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006364595249554084,
+      "loss": 2.0703,
+      "step": 2117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006361432862051251,
+      "loss": 2.1406,
+      "step": 2118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006358269886217194,
+      "loss": 2.0664,
+      "step": 2119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006355106323418766,
+      "loss": 1.9521,
+      "step": 2120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000635194217502307,
+      "loss": 2.0654,
+      "step": 2121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006348777442397467,
+      "loss": 2.043,
+      "step": 2122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000634561212690957,
+      "loss": 2.0283,
+      "step": 2123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006342446229927239,
+      "loss": 2.0928,
+      "step": 2124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006339279752818591,
+      "loss": 2.0771,
+      "step": 2125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006336112696951988,
+      "loss": 2.0312,
+      "step": 2126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006332945063696048,
+      "loss": 2.0264,
+      "step": 2127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006329776854419636,
+      "loss": 2.0322,
+      "step": 2128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006326608070491864,
+      "loss": 2.0967,
+      "step": 2129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006323438713282095,
+      "loss": 2.1104,
+      "step": 2130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006320268784159939,
+      "loss": 2.0752,
+      "step": 2131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006317098284495251,
+      "loss": 2.0693,
+      "step": 2132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006313927215658138,
+      "loss": 2.0908,
+      "step": 2133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006310755579018948,
+      "loss": 2.0508,
+      "step": 2134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006307583375948273,
+      "loss": 2.0117,
+      "step": 2135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006304410607816957,
+      "loss": 2.1191,
+      "step": 2136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006301237275996082,
+      "loss": 2.0615,
+      "step": 2137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006298063381856977,
+      "loss": 2.0254,
+      "step": 2138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006294888926771208,
+      "loss": 2.0605,
+      "step": 2139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006291713912110591,
+      "loss": 2.0713,
+      "step": 2140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006288538339247182,
+      "loss": 2.0488,
+      "step": 2141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006285362209553273,
+      "loss": 1.959,
+      "step": 2142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006282185524401404,
+      "loss": 2.0566,
+      "step": 2143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000627900828516435,
+      "loss": 2.0479,
+      "step": 2144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006275830493215127,
+      "loss": 2.0459,
+      "step": 2145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006272652149926989,
+      "loss": 2.0781,
+      "step": 2146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000626947325667343,
+      "loss": 1.9443,
+      "step": 2147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006266293814828181,
+      "loss": 2.04,
+      "step": 2148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000626311382576521,
+      "loss": 1.9629,
+      "step": 2149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006259933290858719,
+      "loss": 2.0479,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006256752211483151,
+      "loss": 2.0137,
+      "step": 2151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006253570589013182,
+      "loss": 2.0137,
+      "step": 2152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006250388424823719,
+      "loss": 2.0527,
+      "step": 2153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006247205720289907,
+      "loss": 2.0859,
+      "step": 2154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006244022476787124,
+      "loss": 2.0469,
+      "step": 2155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006240838695690983,
+      "loss": 1.9951,
+      "step": 2156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006237654378377324,
+      "loss": 2.0527,
+      "step": 2157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006234469526222223,
+      "loss": 2.0322,
+      "step": 2158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006231284140601985,
+      "loss": 2.0732,
+      "step": 2159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006228098222893148,
+      "loss": 2.0059,
+      "step": 2160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006224911774472476,
+      "loss": 1.9941,
+      "step": 2161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006221724796716966,
+      "loss": 2.0146,
+      "step": 2162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006218537291003842,
+      "loss": 2.1162,
+      "step": 2163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006215349258710558,
+      "loss": 2.0537,
+      "step": 2164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006212160701214793,
+      "loss": 1.9971,
+      "step": 2165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006208971619894452,
+      "loss": 1.9971,
+      "step": 2166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006205782016127672,
+      "loss": 2.0068,
+      "step": 2167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006202591891292809,
+      "loss": 2.0586,
+      "step": 2168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006199401246768449,
+      "loss": 2.0732,
+      "step": 2169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006196210083933403,
+      "loss": 2.0762,
+      "step": 2170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006193018404166698,
+      "loss": 2.0547,
+      "step": 2171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006189826208847596,
+      "loss": 2.0596,
+      "step": 2172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006186633499355575,
+      "loss": 2.0479,
+      "step": 2173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006183440277070334,
+      "loss": 1.9805,
+      "step": 2174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006180246543371797,
+      "loss": 1.9707,
+      "step": 2175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006177052299640109,
+      "loss": 2.0566,
+      "step": 2176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006173857547255633,
+      "loss": 2.0293,
+      "step": 2177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006170662287598955,
+      "loss": 1.9697,
+      "step": 2178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006167466522050875,
+      "loss": 2.082,
+      "step": 2179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006164270251992418,
+      "loss": 1.9033,
+      "step": 2180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006161073478804822,
+      "loss": 2.0801,
+      "step": 2181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006157876203869546,
+      "loss": 2.0635,
+      "step": 2182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006154678428568262,
+      "loss": 2.0439,
+      "step": 2183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006151480154282865,
+      "loss": 2.041,
+      "step": 2184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006148281382395455,
+      "loss": 2.0156,
+      "step": 2185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006145082114288357,
+      "loss": 1.9707,
+      "step": 2186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006141882351344105,
+      "loss": 2.0918,
+      "step": 2187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000613868209494545,
+      "loss": 1.9482,
+      "step": 2188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006135481346475352,
+      "loss": 1.9824,
+      "step": 2189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000613228010731699,
+      "loss": 2.0508,
+      "step": 2190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006129078378853748,
+      "loss": 2.0449,
+      "step": 2191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006125876162469226,
+      "loss": 2.1182,
+      "step": 2192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006122673459547233,
+      "loss": 1.9883,
+      "step": 2193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000611947027147179,
+      "loss": 2.0264,
+      "step": 2194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006116266599627126,
+      "loss": 2.0527,
+      "step": 2195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006113062445397679,
+      "loss": 2.1123,
+      "step": 2196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006109857810168098,
+      "loss": 2.0205,
+      "step": 2197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006106652695323236,
+      "loss": 2.0039,
+      "step": 2198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006103447102248153,
+      "loss": 2.0029,
+      "step": 2199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006100241032328124,
+      "loss": 1.9961,
+      "step": 2200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006097034486948618,
+      "loss": 2.0449,
+      "step": 2201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006093827467495319,
+      "loss": 2.0781,
+      "step": 2202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000609061997535411,
+      "loss": 2.0381,
+      "step": 2203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006087412011911083,
+      "loss": 1.9395,
+      "step": 2204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006084203578552528,
+      "loss": 2.0615,
+      "step": 2205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006080994676664943,
+      "loss": 2.0254,
+      "step": 2206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006077785307635026,
+      "loss": 2.04,
+      "step": 2207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000607457547284968,
+      "loss": 2.0674,
+      "step": 2208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006071365173696003,
+      "loss": 2.0215,
+      "step": 2209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00060681544115613,
+      "loss": 2.043,
+      "step": 2210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006064943187833074,
+      "loss": 2.0605,
+      "step": 2211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006061731503899024,
+      "loss": 1.9619,
+      "step": 2212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006058519361147054,
+      "loss": 2.0576,
+      "step": 2213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006055306760965263,
+      "loss": 2.0137,
+      "step": 2214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006052093704741945,
+      "loss": 2.0449,
+      "step": 2215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006048880193865598,
+      "loss": 1.9971,
+      "step": 2216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006045666229724908,
+      "loss": 1.9434,
+      "step": 2217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006042451813708766,
+      "loss": 2.1191,
+      "step": 2218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006039236947206252,
+      "loss": 1.9434,
+      "step": 2219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000603602163160664,
+      "loss": 2.0059,
+      "step": 2220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006032805868299402,
+      "loss": 1.9248,
+      "step": 2221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006029589658674202,
+      "loss": 1.9951,
+      "step": 2222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006026373004120896,
+      "loss": 2.1191,
+      "step": 2223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006023155906029534,
+      "loss": 2.0615,
+      "step": 2224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006019938365790356,
+      "loss": 2.0498,
+      "step": 2225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006016720384793793,
+      "loss": 1.9854,
+      "step": 2226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006013501964430468,
+      "loss": 2.0664,
+      "step": 2227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006010283106091194,
+      "loss": 1.9609,
+      "step": 2228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006007063811166969,
+      "loss": 2.0684,
+      "step": 2229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006003844081048985,
+      "loss": 1.9336,
+      "step": 2230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006000623917128622,
+      "loss": 2.082,
+      "step": 2231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005997403320797443,
+      "loss": 2.0615,
+      "step": 2232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00059941822934472,
+      "loss": 1.9434,
+      "step": 2233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005990960836469832,
+      "loss": 2.0029,
+      "step": 2234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005987738951257467,
+      "loss": 2.0674,
+      "step": 2235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005984516639202408,
+      "loss": 2.0361,
+      "step": 2236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005981293901697152,
+      "loss": 1.9824,
+      "step": 2237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005978070740134379,
+      "loss": 1.9854,
+      "step": 2238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005974847155906944,
+      "loss": 2.0674,
+      "step": 2239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005971623150407896,
+      "loss": 2.0186,
+      "step": 2240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005968398725030459,
+      "loss": 2.0557,
+      "step": 2241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005965173881168037,
+      "loss": 2.0488,
+      "step": 2242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005961948620214223,
+      "loss": 1.9951,
+      "step": 2243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005958722943562782,
+      "loss": 1.9658,
+      "step": 2244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005955496852607661,
+      "loss": 2.0596,
+      "step": 2245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005952270348742991,
+      "loss": 2.0645,
+      "step": 2246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005949043433363072,
+      "loss": 2.0205,
+      "step": 2247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000594581610786239,
+      "loss": 1.9775,
+      "step": 2248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005942588373635606,
+      "loss": 2.0215,
+      "step": 2249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005939360232077554,
+      "loss": 2.1055,
+      "step": 2250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005936131684583249,
+      "loss": 1.9531,
+      "step": 2251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005932902732547879,
+      "loss": 1.998,
+      "step": 2252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005929673377366808,
+      "loss": 2.1279,
+      "step": 2253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005926443620435572,
+      "loss": 2.0664,
+      "step": 2254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005923213463149882,
+      "loss": 2.0117,
+      "step": 2255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005919982906905625,
+      "loss": 2.0645,
+      "step": 2256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005916751953098853,
+      "loss": 2.0361,
+      "step": 2257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005913520603125794,
+      "loss": 1.959,
+      "step": 2258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005910288858382855,
+      "loss": 2.0166,
+      "step": 2259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005907056720266598,
+      "loss": 2.0625,
+      "step": 2260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005903824190173765,
+      "loss": 2.0166,
+      "step": 2261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005900591269501268,
+      "loss": 1.9932,
+      "step": 2262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005897357959646183,
+      "loss": 2.0762,
+      "step": 2263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005894124262005758,
+      "loss": 2.1182,
+      "step": 2264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005890890177977403,
+      "loss": 2.043,
+      "step": 2265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005887655708958704,
+      "loss": 1.9424,
+      "step": 2266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005884420856347405,
+      "loss": 1.9473,
+      "step": 2267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005881185621541419,
+      "loss": 2.0215,
+      "step": 2268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005877950005938827,
+      "loss": 2.0918,
+      "step": 2269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005874714010937871,
+      "loss": 2.0146,
+      "step": 2270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005871477637936955,
+      "loss": 2.0195,
+      "step": 2271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005868240888334653,
+      "loss": 1.9873,
+      "step": 2272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005865003763529694,
+      "loss": 2.083,
+      "step": 2273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005861766264920975,
+      "loss": 1.9863,
+      "step": 2274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005858528393907552,
+      "loss": 2.0059,
+      "step": 2275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005855290151888644,
+      "loss": 2.0312,
+      "step": 2276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005852051540263627,
+      "loss": 2.0469,
+      "step": 2277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000584881256043204,
+      "loss": 1.9844,
+      "step": 2278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005845573213793577,
+      "loss": 2.1035,
+      "step": 2279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005842333501748096,
+      "loss": 2.0625,
+      "step": 2280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005839093425695609,
+      "loss": 1.9668,
+      "step": 2281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005835852987036285,
+      "loss": 2.1484,
+      "step": 2282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005832612187170453,
+      "loss": 2.0791,
+      "step": 2283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005829371027498596,
+      "loss": 1.9785,
+      "step": 2284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005826129509421351,
+      "loss": 1.9893,
+      "step": 2285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005822887634339512,
+      "loss": 2.0889,
+      "step": 2286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005819645403654027,
+      "loss": 1.9629,
+      "step": 2287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005816402818766,
+      "loss": 2.0312,
+      "step": 2288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005813159881076681,
+      "loss": 1.9268,
+      "step": 2289
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005809916591987479,
+      "loss": 2.0195,
+      "step": 2290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005806672952899954,
+      "loss": 1.9971,
+      "step": 2291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005803428965215815,
+      "loss": 2.0342,
+      "step": 2292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005800184630336922,
+      "loss": 1.9941,
+      "step": 2293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005796939949665287,
+      "loss": 2.1299,
+      "step": 2294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005793694924603071,
+      "loss": 2.0332,
+      "step": 2295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000579044955655258,
+      "loss": 1.9902,
+      "step": 2296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005787203846916272,
+      "loss": 2.0869,
+      "step": 2297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005783957797096755,
+      "loss": 2.0029,
+      "step": 2298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005780711408496777,
+      "loss": 1.9688,
+      "step": 2299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005777464682519239,
+      "loss": 1.9463,
+      "step": 2300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005774217620567183,
+      "loss": 2.0088,
+      "step": 2301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005770970224043802,
+      "loss": 2.0107,
+      "step": 2302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005767722494352422,
+      "loss": 2.0342,
+      "step": 2303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005764474432896528,
+      "loss": 1.915,
+      "step": 2304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000576122604107974,
+      "loss": 2.0039,
+      "step": 2305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005757977320305821,
+      "loss": 2.043,
+      "step": 2306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005754728271978675,
+      "loss": 2.0918,
+      "step": 2307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005751478897502352,
+      "loss": 1.9834,
+      "step": 2308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005748229198281041,
+      "loss": 1.9404,
+      "step": 2309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005744979175719069,
+      "loss": 1.9775,
+      "step": 2310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005741728831220907,
+      "loss": 2.0029,
+      "step": 2311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000573847816619116,
+      "loss": 1.9746,
+      "step": 2312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005735227182034577,
+      "loss": 2.0576,
+      "step": 2313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000573197588015604,
+      "loss": 1.999,
+      "step": 2314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005728724261960572,
+      "loss": 2.0215,
+      "step": 2315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005725472328853329,
+      "loss": 2.0391,
+      "step": 2316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005722220082239607,
+      "loss": 1.9932,
+      "step": 2317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005718967523524837,
+      "loss": 2.0234,
+      "step": 2318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005715714654114581,
+      "loss": 2.0068,
+      "step": 2319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000571246147541454,
+      "loss": 1.9463,
+      "step": 2320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005709207988830545,
+      "loss": 2.0498,
+      "step": 2321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005705954195768561,
+      "loss": 2.0332,
+      "step": 2322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000570270009763469,
+      "loss": 2.0674,
+      "step": 2323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005699445695835155,
+      "loss": 2.0869,
+      "step": 2324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005696190991776323,
+      "loss": 2.0635,
+      "step": 2325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005692935986864685,
+      "loss": 2.0732,
+      "step": 2326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000568968068250686,
+      "loss": 1.9609,
+      "step": 2327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00056864250801096,
+      "loss": 1.9893,
+      "step": 2328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005683169181079787,
+      "loss": 2.0312,
+      "step": 2329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005679912986824427,
+      "loss": 2.0273,
+      "step": 2330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005676656498750656,
+      "loss": 1.9863,
+      "step": 2331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005673399718265737,
+      "loss": 2.0166,
+      "step": 2332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005670142646777059,
+      "loss": 1.9766,
+      "step": 2333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005666885285692137,
+      "loss": 2.0186,
+      "step": 2334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005663627636418611,
+      "loss": 1.9658,
+      "step": 2335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005660369700364242,
+      "loss": 1.9658,
+      "step": 2336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005657111478936926,
+      "loss": 2.0078,
+      "step": 2337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005653852973544666,
+      "loss": 2.0986,
+      "step": 2338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005650594185595604,
+      "loss": 2.0,
+      "step": 2339
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005647335116497992,
+      "loss": 2.0469,
+      "step": 2340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005644075767660209,
+      "loss": 2.1289,
+      "step": 2341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005640816140490753,
+      "loss": 1.9541,
+      "step": 2342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005637556236398246,
+      "loss": 2.0449,
+      "step": 2343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005634296056791423,
+      "loss": 2.0518,
+      "step": 2344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005631035603079146,
+      "loss": 1.9492,
+      "step": 2345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005627774876670388,
+      "loss": 2.0254,
+      "step": 2346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005624513878974244,
+      "loss": 2.0908,
+      "step": 2347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005621252611399927,
+      "loss": 1.9893,
+      "step": 2348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005617991075356763,
+      "loss": 2.0322,
+      "step": 2349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005614729272254199,
+      "loss": 2.0557,
+      "step": 2350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005611467203501791,
+      "loss": 2.0469,
+      "step": 2351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005608204870509214,
+      "loss": 1.998,
+      "step": 2352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000560494227468626,
+      "loss": 2.125,
+      "step": 2353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005601679417442827,
+      "loss": 2.1152,
+      "step": 2354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005598416300188931,
+      "loss": 2.0352,
+      "step": 2355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00055951529243347,
+      "loss": 2.0117,
+      "step": 2356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005591889291290373,
+      "loss": 2.0566,
+      "step": 2357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005588625402466302,
+      "loss": 1.9902,
+      "step": 2358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005585361259272943,
+      "loss": 2.0801,
+      "step": 2359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005582096863120872,
+      "loss": 2.0049,
+      "step": 2360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005578832215420767,
+      "loss": 1.9766,
+      "step": 2361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005575567317583414,
+      "loss": 1.9658,
+      "step": 2362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005572302171019715,
+      "loss": 1.9775,
+      "step": 2363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005569036777140672,
+      "loss": 2.1045,
+      "step": 2364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005565771137357395,
+      "loss": 2.002,
+      "step": 2365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005562505253081103,
+      "loss": 2.0957,
+      "step": 2366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005559239125723119,
+      "loss": 2.0088,
+      "step": 2367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005555972756694869,
+      "loss": 2.0107,
+      "step": 2368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005552706147407888,
+      "loss": 2.0605,
+      "step": 2369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005549439299273814,
+      "loss": 2.0781,
+      "step": 2370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005546172213704383,
+      "loss": 1.9805,
+      "step": 2371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000554290489211144,
+      "loss": 1.9834,
+      "step": 2372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005539637335906926,
+      "loss": 1.9521,
+      "step": 2373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005536369546502891,
+      "loss": 2.1377,
+      "step": 2374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005533101525311477,
+      "loss": 1.9531,
+      "step": 2375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005529833273744933,
+      "loss": 1.9678,
+      "step": 2376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005526564793215606,
+      "loss": 2.0039,
+      "step": 2377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000552329608513594,
+      "loss": 2.0166,
+      "step": 2378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005520027150918477,
+      "loss": 2.0547,
+      "step": 2379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000551675799197586,
+      "loss": 2.0742,
+      "step": 2380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005513488609720827,
+      "loss": 2.0811,
+      "step": 2381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005510219005566214,
+      "loss": 2.0918,
+      "step": 2382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005506949180924947,
+      "loss": 1.9629,
+      "step": 2383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005503679137210057,
+      "loss": 2.0547,
+      "step": 2384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005500408875834664,
+      "loss": 2.043,
+      "step": 2385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005497138398211979,
+      "loss": 1.9678,
+      "step": 2386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005493867705755312,
+      "loss": 2.0615,
+      "step": 2387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005490596799878067,
+      "loss": 2.1045,
+      "step": 2388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005487325681993732,
+      "loss": 1.9668,
+      "step": 2389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005484054353515896,
+      "loss": 2.0293,
+      "step": 2390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005480782815858233,
+      "loss": 2.0381,
+      "step": 2391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005477511070434509,
+      "loss": 1.9727,
+      "step": 2392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005474239118658579,
+      "loss": 2.0225,
+      "step": 2393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005470966961944392,
+      "loss": 2.0537,
+      "step": 2394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005467694601705977,
+      "loss": 2.0557,
+      "step": 2395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000546442203935746,
+      "loss": 2.043,
+      "step": 2396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005461149276313046,
+      "loss": 2.0156,
+      "step": 2397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005457876313987033,
+      "loss": 1.9912,
+      "step": 2398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005454603153793804,
+      "loss": 1.9873,
+      "step": 2399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005451329797147822,
+      "loss": 2.0547,
+      "step": 2400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005448056245463644,
+      "loss": 2.0107,
+      "step": 2401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005444782500155904,
+      "loss": 2.0225,
+      "step": 2402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005441508562639322,
+      "loss": 1.9775,
+      "step": 2403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005438234434328704,
+      "loss": 1.9922,
+      "step": 2404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005434960116638932,
+      "loss": 2.0664,
+      "step": 2405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005431685610984976,
+      "loss": 1.9971,
+      "step": 2406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005428410918781884,
+      "loss": 1.9717,
+      "step": 2407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005425136041444786,
+      "loss": 1.9717,
+      "step": 2408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005421860980388892,
+      "loss": 2.0234,
+      "step": 2409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005418585737029489,
+      "loss": 1.9727,
+      "step": 2410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005415310312781943,
+      "loss": 1.9863,
+      "step": 2411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005412034709061705,
+      "loss": 1.9912,
+      "step": 2412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005408758927284294,
+      "loss": 1.9316,
+      "step": 2413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005405482968865312,
+      "loss": 2.0098,
+      "step": 2414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005402206835220435,
+      "loss": 2.0059,
+      "step": 2415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005398930527765415,
+      "loss": 1.9893,
+      "step": 2416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005395654047916082,
+      "loss": 1.9473,
+      "step": 2417
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005392377397088334,
+      "loss": 2.041,
+      "step": 2418
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005389100576698149,
+      "loss": 1.9648,
+      "step": 2419
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005385823588161577,
+      "loss": 1.9619,
+      "step": 2420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005382546432894737,
+      "loss": 2.1045,
+      "step": 2421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005379269112313823,
+      "loss": 2.0215,
+      "step": 2422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005375991627835103,
+      "loss": 1.9697,
+      "step": 2423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005372713980874913,
+      "loss": 2.0342,
+      "step": 2424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005369436172849657,
+      "loss": 1.9727,
+      "step": 2425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000536615820517581,
+      "loss": 2.0576,
+      "step": 2426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005362880079269921,
+      "loss": 1.9473,
+      "step": 2427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00053596017965486,
+      "loss": 2.041,
+      "step": 2428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005356323358428528,
+      "loss": 1.9424,
+      "step": 2429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005353044766326455,
+      "loss": 2.0342,
+      "step": 2430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005349766021659195,
+      "loss": 1.9502,
+      "step": 2431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005346487125843627,
+      "loss": 1.9775,
+      "step": 2432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00053432080802967,
+      "loss": 1.9199,
+      "step": 2433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005339928886435423,
+      "loss": 2.0977,
+      "step": 2434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005336649545676869,
+      "loss": 2.082,
+      "step": 2435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005333370059438179,
+      "loss": 1.9824,
+      "step": 2436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005330090429136552,
+      "loss": 2.0381,
+      "step": 2437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005326810656189254,
+      "loss": 2.0186,
+      "step": 2438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005323530742013608,
+      "loss": 1.9434,
+      "step": 2439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005320250688027,
+      "loss": 2.0215,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005316970495646878,
+      "loss": 2.0859,
+      "step": 2441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005313690166290746,
+      "loss": 1.9941,
+      "step": 2442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000531040970137617,
+      "loss": 2.0029,
+      "step": 2443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005307129102320777,
+      "loss": 2.0098,
+      "step": 2444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005303848370542246,
+      "loss": 2.0508,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005300567507458315,
+      "loss": 2.041,
+      "step": 2446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005297286514486785,
+      "loss": 1.9814,
+      "step": 2447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005294005393045502,
+      "loss": 2.0312,
+      "step": 2448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005290724144552379,
+      "loss": 1.9873,
+      "step": 2449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005287442770425377,
+      "loss": 2.0361,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005284161272082513,
+      "loss": 1.8965,
+      "step": 2451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005280879650941857,
+      "loss": 2.0898,
+      "step": 2452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005277597908421531,
+      "loss": 1.9492,
+      "step": 2453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005274316045939714,
+      "loss": 1.9316,
+      "step": 2454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005271034064914634,
+      "loss": 2.0137,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005267751966764569,
+      "loss": 1.9707,
+      "step": 2456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005264469752907848,
+      "loss": 2.0205,
+      "step": 2457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005261187424762854,
+      "loss": 1.9834,
+      "step": 2458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005257904983748013,
+      "loss": 1.999,
+      "step": 2459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005254622431281804,
+      "loss": 2.0439,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005251339768782753,
+      "loss": 1.916,
+      "step": 2461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005248056997669434,
+      "loss": 1.9844,
+      "step": 2462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005244774119360468,
+      "loss": 1.918,
+      "step": 2463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005241491135274521,
+      "loss": 2.0,
+      "step": 2464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005238208046830307,
+      "loss": 2.0088,
+      "step": 2465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005234924855446584,
+      "loss": 2.043,
+      "step": 2466
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000523164156254215,
+      "loss": 1.9482,
+      "step": 2467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005228358169535856,
+      "loss": 2.002,
+      "step": 2468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000522507467784659,
+      "loss": 2.0674,
+      "step": 2469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005221791088893282,
+      "loss": 1.9971,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000521850740409491,
+      "loss": 2.0059,
+      "step": 2471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005215223624870487,
+      "loss": 2.0332,
+      "step": 2472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005211939752639067,
+      "loss": 2.0264,
+      "step": 2473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005208655788819751,
+      "loss": 1.9824,
+      "step": 2474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005205371734831675,
+      "loss": 2.0234,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000520208759209401,
+      "loss": 1.9795,
+      "step": 2476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000519880336202597,
+      "loss": 1.9854,
+      "step": 2477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005195519046046808,
+      "loss": 2.0635,
+      "step": 2478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005192234645575814,
+      "loss": 2.0215,
+      "step": 2479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005188950162032309,
+      "loss": 1.9307,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005185665596835655,
+      "loss": 2.0176,
+      "step": 2481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005182380951405249,
+      "loss": 2.0312,
+      "step": 2482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000517909622716052,
+      "loss": 2.0381,
+      "step": 2483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005175811425520933,
+      "loss": 2.0781,
+      "step": 2484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005172526547905987,
+      "loss": 1.9307,
+      "step": 2485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005169241595735213,
+      "loss": 1.9912,
+      "step": 2486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005165956570428172,
+      "loss": 1.9766,
+      "step": 2487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005162671473404461,
+      "loss": 2.0254,
+      "step": 2488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005159386306083705,
+      "loss": 1.9795,
+      "step": 2489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005156101069885563,
+      "loss": 1.9834,
+      "step": 2490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005152815766229714,
+      "loss": 2.1152,
+      "step": 2491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005149530396535879,
+      "loss": 1.9971,
+      "step": 2492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005146244962223799,
+      "loss": 2.0586,
+      "step": 2493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005142959464713247,
+      "loss": 2.0264,
+      "step": 2494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005139673905424021,
+      "loss": 2.0186,
+      "step": 2495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005136388285775946,
+      "loss": 1.9893,
+      "step": 2496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005133102607188875,
+      "loss": 2.1211,
+      "step": 2497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005129816871082683,
+      "loss": 1.9775,
+      "step": 2498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005126531078877274,
+      "loss": 1.9717,
+      "step": 2499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005123245231992573,
+      "loss": 1.9199,
+      "step": 2500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005119959331848531,
+      "loss": 1.9912,
+      "step": 2501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005116673379865119,
+      "loss": 1.9531,
+      "step": 2502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005113387377462334,
+      "loss": 1.9492,
+      "step": 2503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005110101326060192,
+      "loss": 2.1455,
+      "step": 2504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005106815227078729,
+      "loss": 2.1533,
+      "step": 2505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000510352908193801,
+      "loss": 1.9785,
+      "step": 2506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005100242892058108,
+      "loss": 1.9795,
+      "step": 2507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005096956658859122,
+      "loss": 2.0674,
+      "step": 2508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005093670383761171,
+      "loss": 1.9873,
+      "step": 2509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005090384068184387,
+      "loss": 2.0791,
+      "step": 2510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005087097713548926,
+      "loss": 2.0088,
+      "step": 2511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005083811321274953,
+      "loss": 2.0479,
+      "step": 2512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005080524892782655,
+      "loss": 2.1641,
+      "step": 2513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005077238429492236,
+      "loss": 1.8809,
+      "step": 2514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005073951932823907,
+      "loss": 2.0283,
+      "step": 2515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005070665404197903,
+      "loss": 1.9551,
+      "step": 2516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005067378845034467,
+      "loss": 2.0352,
+      "step": 2517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005064092256753856,
+      "loss": 2.084,
+      "step": 2518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000506080564077634,
+      "loss": 1.9961,
+      "step": 2519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005057518998522202,
+      "loss": 1.9199,
+      "step": 2520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005054232331411735,
+      "loss": 2.0205,
+      "step": 2521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005050945640865244,
+      "loss": 1.8857,
+      "step": 2522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005047658928303043,
+      "loss": 1.9902,
+      "step": 2523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005044372195145455,
+      "loss": 2.0029,
+      "step": 2524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005041085442812815,
+      "loss": 2.0615,
+      "step": 2525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005037798672725461,
+      "loss": 1.9492,
+      "step": 2526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005034511886303746,
+      "loss": 1.9941,
+      "step": 2527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000503122508496802,
+      "loss": 2.0498,
+      "step": 2528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000502793827013865,
+      "loss": 1.9785,
+      "step": 2529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005024651443236003,
+      "loss": 2.0537,
+      "step": 2530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000502136460568045,
+      "loss": 1.9639,
+      "step": 2531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005018077758892372,
+      "loss": 2.0283,
+      "step": 2532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005014790904292149,
+      "loss": 2.0635,
+      "step": 2533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005011504043300167,
+      "loss": 1.9355,
+      "step": 2534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005008217177336817,
+      "loss": 1.9795,
+      "step": 2535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005004930307822481,
+      "loss": 2.1133,
+      "step": 2536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000500164343617756,
+      "loss": 2.043,
+      "step": 2537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000499835656382244,
+      "loss": 1.9775,
+      "step": 2538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004995069692177519,
+      "loss": 1.9316,
+      "step": 2539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004991782822663186,
+      "loss": 2.0068,
+      "step": 2540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004988495956699832,
+      "loss": 1.9658,
+      "step": 2541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004985209095707851,
+      "loss": 1.9951,
+      "step": 2542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004981922241107629,
+      "loss": 1.8896,
+      "step": 2543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004978635394319549,
+      "loss": 2.0547,
+      "step": 2544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004975348556763999,
+      "loss": 1.9326,
+      "step": 2545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004972061729861351,
+      "loss": 2.0342,
+      "step": 2546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004968774915031981,
+      "loss": 1.9502,
+      "step": 2547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004965488113696257,
+      "loss": 2.0498,
+      "step": 2548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000496220132727454,
+      "loss": 1.9971,
+      "step": 2549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004958914557187187,
+      "loss": 2.0625,
+      "step": 2550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004955627804854544,
+      "loss": 1.9648,
+      "step": 2551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004952341071696959,
+      "loss": 2.0283,
+      "step": 2552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004949054359134757,
+      "loss": 2.0557,
+      "step": 2553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004945767668588264,
+      "loss": 1.9893,
+      "step": 2554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004942481001477799,
+      "loss": 2.0244,
+      "step": 2555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004939194359223662,
+      "loss": 1.9932,
+      "step": 2556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004935907743246145,
+      "loss": 1.9189,
+      "step": 2557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004932621154965534,
+      "loss": 1.9561,
+      "step": 2558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004929334595802098,
+      "loss": 2.0254,
+      "step": 2559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004926048067176093,
+      "loss": 1.8916,
+      "step": 2560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004922761570507765,
+      "loss": 2.002,
+      "step": 2561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004919475107217346,
+      "loss": 2.002,
+      "step": 2562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004916188678725049,
+      "loss": 2.0635,
+      "step": 2563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004912902286451074,
+      "loss": 2.0469,
+      "step": 2564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004909615931815613,
+      "loss": 2.0732,
+      "step": 2565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004906329616238831,
+      "loss": 1.9092,
+      "step": 2566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004903043341140879,
+      "loss": 1.9873,
+      "step": 2567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004899757107941894,
+      "loss": 2.0225,
+      "step": 2568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004896470918061992,
+      "loss": 2.0361,
+      "step": 2569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000489318477292127,
+      "loss": 2.0059,
+      "step": 2570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000488989867393981,
+      "loss": 2.0479,
+      "step": 2571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004886612622537668,
+      "loss": 1.9541,
+      "step": 2572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004883326620134881,
+      "loss": 2.0127,
+      "step": 2573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004880040668151471,
+      "loss": 1.9717,
+      "step": 2574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004876754768007428,
+      "loss": 2.0371,
+      "step": 2575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004873468921122726,
+      "loss": 2.0195,
+      "step": 2576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004870183128917318,
+      "loss": 2.082,
+      "step": 2577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004866897392811127,
+      "loss": 1.9688,
+      "step": 2578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004863611714224054,
+      "loss": 1.9697,
+      "step": 2579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000486032609457598,
+      "loss": 2.1094,
+      "step": 2580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048570405352867536,
+      "loss": 2.0508,
+      "step": 2581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004853755037776202,
+      "loss": 2.0381,
+      "step": 2582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048504696034641214,
+      "loss": 1.9492,
+      "step": 2583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004847184233770288,
+      "loss": 1.9492,
+      "step": 2584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000484389893011444,
+      "loss": 1.9639,
+      "step": 2585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004840613693916294,
+      "loss": 1.9004,
+      "step": 2586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004837328526595539,
+      "loss": 2.001,
+      "step": 2587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048340434295718283,
+      "loss": 2.0293,
+      "step": 2588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048307584042647875,
+      "loss": 1.9902,
+      "step": 2589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004827473452094013,
+      "loss": 1.8955,
+      "step": 2590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004824188574479067,
+      "loss": 1.9873,
+      "step": 2591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004820903772839481,
+      "loss": 2.0176,
+      "step": 2592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004817619048594752,
+      "loss": 1.9395,
+      "step": 2593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004814334403164346,
+      "loss": 1.9844,
+      "step": 2594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048110498379676914,
+      "loss": 2.085,
+      "step": 2595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004807765354424186,
+      "loss": 2.0156,
+      "step": 2596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004804480953953192,
+      "loss": 2.0742,
+      "step": 2597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048011966379740305,
+      "loss": 1.9951,
+      "step": 2598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047979124079059913,
+      "loss": 1.9736,
+      "step": 2599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047946282651683273,
+      "loss": 2.1328,
+      "step": 2600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000479134421118025,
+      "loss": 2.0225,
+      "step": 2601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047880602473609324,
+      "loss": 1.9219,
+      "step": 2602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047847763751295144,
+      "loss": 1.9033,
+      "step": 2603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047814925959050917,
+      "loss": 2.1104,
+      "step": 2604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047782089111067175,
+      "loss": 1.9268,
+      "step": 2605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047749253221534107,
+      "loss": 1.9814,
+      "step": 2606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004771641830464146,
+      "loss": 2.0107,
+      "step": 2607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047683584374578504,
+      "loss": 2.0156,
+      "step": 2608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047650751445534175,
+      "loss": 2.04,
+      "step": 2609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004761791953169694,
+      "loss": 2.0244,
+      "step": 2610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004758508864725478,
+      "loss": 1.9893,
+      "step": 2611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047552258806395325,
+      "loss": 2.0596,
+      "step": 2612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047519430023305664,
+      "loss": 1.9248,
+      "step": 2613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004748660231217248,
+      "loss": 2.1367,
+      "step": 2614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047453775687181964,
+      "loss": 1.9619,
+      "step": 2615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047420950162519875,
+      "loss": 1.9668,
+      "step": 2616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047388125752371475,
+      "loss": 2.0186,
+      "step": 2617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004735530247092151,
+      "loss": 2.0107,
+      "step": 2618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004732248033235432,
+      "loss": 1.9824,
+      "step": 2619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004728965935085367,
+      "loss": 1.9277,
+      "step": 2620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047256839540602854,
+      "loss": 2.0283,
+      "step": 2621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000472240209157847,
+      "loss": 1.9707,
+      "step": 2622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004719120349058145,
+      "loss": 2.041,
+      "step": 2623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047158387279174873,
+      "loss": 2.042,
+      "step": 2624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004712557229574624,
+      "loss": 1.9814,
+      "step": 2625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004709275855447621,
+      "loss": 1.9805,
+      "step": 2626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047059946069544966,
+      "loss": 1.9805,
+      "step": 2627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047027134855132167,
+      "loss": 2.0391,
+      "step": 2628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046994324925416864,
+      "loss": 2.0332,
+      "step": 2629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004696151629457756,
+      "loss": 1.998,
+      "step": 2630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004692870897679224,
+      "loss": 1.9893,
+      "step": 2631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046895902986238304,
+      "loss": 2.0615,
+      "step": 2632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046863098337092553,
+      "loss": 2.0254,
+      "step": 2633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004683029504353123,
+      "loss": 2.1318,
+      "step": 2634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004679749311973001,
+      "loss": 1.9883,
+      "step": 2635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004676469257986394,
+      "loss": 2.0078,
+      "step": 2636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004673189343810747,
+      "loss": 1.9746,
+      "step": 2637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046699095708634483,
+      "loss": 2.0498,
+      "step": 2638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004666629940561823,
+      "loss": 1.8848,
+      "step": 2639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004663350454323132,
+      "loss": 2.0146,
+      "step": 2640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004660071113564579,
+      "loss": 2.0254,
+      "step": 2641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004656791919703302,
+      "loss": 1.9619,
+      "step": 2642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046535128741563727,
+      "loss": 1.9424,
+      "step": 2643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004650233978340805,
+      "loss": 1.9629,
+      "step": 2644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004646955233673546,
+      "loss": 1.9902,
+      "step": 2645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004643676641571472,
+      "loss": 1.9961,
+      "step": 2646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046403982034514015,
+      "loss": 1.9668,
+      "step": 2647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046371199207300795,
+      "loss": 2.0918,
+      "step": 2648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046338417948241906,
+      "loss": 1.9385,
+      "step": 2649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004630563827150344,
+      "loss": 2.001,
+      "step": 2650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046272860191250875,
+      "loss": 1.9688,
+      "step": 2651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046240083721648973,
+      "loss": 1.9355,
+      "step": 2652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046207308876861764,
+      "loss": 1.917,
+      "step": 2653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004617453567105264,
+      "loss": 2.0459,
+      "step": 2654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046141764118384256,
+      "loss": 2.0615,
+      "step": 2655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004610899423301851,
+      "loss": 2.0566,
+      "step": 2656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046076226029116674,
+      "loss": 2.0381,
+      "step": 2657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000460434595208392,
+      "loss": 1.9893,
+      "step": 2658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004601069472234584,
+      "loss": 1.9834,
+      "step": 2659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004597793164779566,
+      "loss": 2.0273,
+      "step": 2660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004594517031134689,
+      "loss": 1.9824,
+      "step": 2661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004591241072715706,
+      "loss": 1.9775,
+      "step": 2662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004587965290938296,
+      "loss": 1.998,
+      "step": 2663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045846896872180575,
+      "loss": 1.9541,
+      "step": 2664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045814142629705133,
+      "loss": 2.041,
+      "step": 2665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004578139019611109,
+      "loss": 1.9551,
+      "step": 2666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045748639585552143,
+      "loss": 2.0791,
+      "step": 2667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004571589081218116,
+      "loss": 2.0918,
+      "step": 2668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045683143890150237,
+      "loss": 1.8779,
+      "step": 2669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045650398833610683,
+      "loss": 1.8643,
+      "step": 2670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004561765565671297,
+      "loss": 1.9961,
+      "step": 2671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004558491437360677,
+      "loss": 1.9209,
+      "step": 2672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004555217499844097,
+      "loss": 2.0791,
+      "step": 2673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004551943754536358,
+      "loss": 1.999,
+      "step": 2674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004548670202852178,
+      "loss": 2.0361,
+      "step": 2675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004545396846206198,
+      "loss": 2.0166,
+      "step": 2676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045421236860129685,
+      "loss": 1.9873,
+      "step": 2677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045388507236869546,
+      "loss": 1.9922,
+      "step": 2678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045355779606425406,
+      "loss": 1.9834,
+      "step": 2679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045323053982940237,
+      "loss": 2.002,
+      "step": 2680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004529033038055609,
+      "loss": 1.7871,
+      "step": 2681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004525760881341421,
+      "loss": 1.9414,
+      "step": 2682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045224889295654924,
+      "loss": 1.9717,
+      "step": 2683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045192171841417693,
+      "loss": 1.9561,
+      "step": 2684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004515945646484105,
+      "loss": 1.9551,
+      "step": 2685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004512674318006268,
+      "loss": 1.9346,
+      "step": 2686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045094032001219355,
+      "loss": 1.8955,
+      "step": 2687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004506132294244687,
+      "loss": 1.9209,
+      "step": 2688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004502861601788022,
+      "loss": 2.0293,
+      "step": 2689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044995911241653387,
+      "loss": 1.9404,
+      "step": 2690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044963208627899425,
+      "loss": 1.9336,
+      "step": 2691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044930508190750535,
+      "loss": 1.9111,
+      "step": 2692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004489780994433788,
+      "loss": 2.0479,
+      "step": 2693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004486511390279172,
+      "loss": 1.9287,
+      "step": 2694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044832420080241403,
+      "loss": 1.916,
+      "step": 2695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044799728490815235,
+      "loss": 1.9854,
+      "step": 2696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044767039148640596,
+      "loss": 2.0479,
+      "step": 2697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044734352067843944,
+      "loss": 1.96,
+      "step": 2698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004470166726255068,
+      "loss": 1.9463,
+      "step": 2699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004466898474688524,
+      "loss": 2.04,
+      "step": 2700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000446363045349711,
+      "loss": 1.9561,
+      "step": 2701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004460362664093075,
+      "loss": 2.001,
+      "step": 2702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044570951078885617,
+      "loss": 1.9629,
+      "step": 2703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004453827786295617,
+      "loss": 1.9336,
+      "step": 2704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044505607007261865,
+      "loss": 2.0088,
+      "step": 2705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044472938525921113,
+      "loss": 2.0068,
+      "step": 2706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000444402724330513,
+      "loss": 2.0498,
+      "step": 2707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044407608742768825,
+      "loss": 2.0195,
+      "step": 2708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044374947469188986,
+      "loss": 2.0,
+      "step": 2709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004434228862642605,
+      "loss": 2.0469,
+      "step": 2710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004430963222859329,
+      "loss": 2.0664,
+      "step": 2711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004427697828980286,
+      "loss": 1.9941,
+      "step": 2712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044244326824165853,
+      "loss": 1.9844,
+      "step": 2713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044211677845792344,
+      "loss": 2.0781,
+      "step": 2714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044179031368791295,
+      "loss": 1.9814,
+      "step": 2715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004414638740727058,
+      "loss": 1.9609,
+      "step": 2716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044113745975336995,
+      "loss": 2.0049,
+      "step": 2717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004408110708709628,
+      "loss": 1.9463,
+      "step": 2718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004404847075665302,
+      "loss": 2.0098,
+      "step": 2719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000440158369981107,
+      "loss": 2.0029,
+      "step": 2720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043983205825571744,
+      "loss": 1.9805,
+      "step": 2721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043950577253137424,
+      "loss": 1.917,
+      "step": 2722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004391795129490786,
+      "loss": 1.959,
+      "step": 2723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043885327964982105,
+      "loss": 1.9707,
+      "step": 2724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043852707277458033,
+      "loss": 2.0596,
+      "step": 2725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004382008924643237,
+      "loss": 2.0439,
+      "step": 2726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043787473886000745,
+      "loss": 2.0439,
+      "step": 2727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043754861210257566,
+      "loss": 1.9531,
+      "step": 2728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004372225123329613,
+      "loss": 1.9053,
+      "step": 2729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043689643969208557,
+      "loss": 1.8955,
+      "step": 2730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043657039432085776,
+      "loss": 2.0381,
+      "step": 2731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043624437636017547,
+      "loss": 2.0879,
+      "step": 2732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004359183859509247,
+      "loss": 1.9219,
+      "step": 2733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043559242323397917,
+      "loss": 2.0898,
+      "step": 2734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000435266488350201,
+      "loss": 1.9912,
+      "step": 2735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004349405814404396,
+      "loss": 1.9746,
+      "step": 2736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004346147026455334,
+      "loss": 1.9229,
+      "step": 2737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043428885210630757,
+      "loss": 2.0049,
+      "step": 2738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004339630299635756,
+      "loss": 1.8623,
+      "step": 2739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043363723635813907,
+      "loss": 2.0234,
+      "step": 2740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004333114714307864,
+      "loss": 2.041,
+      "step": 2741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004329857353222941,
+      "loss": 1.9121,
+      "step": 2742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043266002817342634,
+      "loss": 2.001,
+      "step": 2743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004323343501249346,
+      "loss": 1.8916,
+      "step": 2744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004320087013175574,
+      "loss": 1.9375,
+      "step": 2745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043168308189202136,
+      "loss": 2.0684,
+      "step": 2746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004313574919890401,
+      "loss": 2.0176,
+      "step": 2747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004310319317493141,
+      "loss": 1.9336,
+      "step": 2748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004307064013135316,
+      "loss": 1.8398,
+      "step": 2749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004303809008223678,
+      "loss": 1.999,
+      "step": 2750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004300554304164846,
+      "loss": 2.0371,
+      "step": 2751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004297299902365311,
+      "loss": 2.0215,
+      "step": 2752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042940458042314394,
+      "loss": 1.9893,
+      "step": 2753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042907920111694576,
+      "loss": 1.9775,
+      "step": 2754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004287538524585461,
+      "loss": 1.9121,
+      "step": 2755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004284285345885419,
+      "loss": 2.0527,
+      "step": 2756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042810324764751647,
+      "loss": 1.8789,
+      "step": 2757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004277779917760393,
+      "loss": 1.9512,
+      "step": 2758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042745276711466713,
+      "loss": 1.9766,
+      "step": 2759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000427127573803943,
+      "loss": 1.9678,
+      "step": 2760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004268024119843961,
+      "loss": 1.9346,
+      "step": 2761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042647728179654235,
+      "loss": 2.0225,
+      "step": 2762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042615218338088405,
+      "loss": 2.0156,
+      "step": 2763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004258271168779093,
+      "loss": 1.9775,
+      "step": 2764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004255020824280931,
+      "loss": 1.8975,
+      "step": 2765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000425177080171896,
+      "loss": 2.0322,
+      "step": 2766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042485211024976497,
+      "loss": 2.0352,
+      "step": 2767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042452717280213253,
+      "loss": 1.9814,
+      "step": 2768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000424202267969418,
+      "loss": 2.0215,
+      "step": 2769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004238773958920261,
+      "loss": 2.0156,
+      "step": 2770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042355255671034715,
+      "loss": 1.9297,
+      "step": 2771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004232277505647579,
+      "loss": 2.0635,
+      "step": 2772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042290297759562013,
+      "loss": 1.9502,
+      "step": 2773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004225782379432817,
+      "loss": 1.9131,
+      "step": 2774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004222535317480762,
+      "loss": 1.9404,
+      "step": 2775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042192885915032235,
+      "loss": 1.9258,
+      "step": 2776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004216042202903245,
+      "loss": 1.8545,
+      "step": 2777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042127961530837276,
+      "loss": 2.0303,
+      "step": 2778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004209550443447421,
+      "loss": 2.0293,
+      "step": 2779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042063050753969296,
+      "loss": 1.9043,
+      "step": 2780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004203060050334713,
+      "loss": 1.8926,
+      "step": 2781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041998153696630794,
+      "loss": 1.9824,
+      "step": 2782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004196571034784186,
+      "loss": 1.9121,
+      "step": 2783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004193327047100046,
+      "loss": 1.9795,
+      "step": 2784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041900834080125214,
+      "loss": 1.9521,
+      "step": 2785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041868401189233207,
+      "loss": 2.0361,
+      "step": 2786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041835971812340014,
+      "loss": 1.9424,
+      "step": 2787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041803545963459734,
+      "loss": 1.8418,
+      "step": 2788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041771123656604906,
+      "loss": 1.8945,
+      "step": 2789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041738704905786505,
+      "loss": 1.8672,
+      "step": 2790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041706289725014056,
+      "loss": 1.9248,
+      "step": 2791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004167387812829549,
+      "loss": 1.957,
+      "step": 2792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041641470129637155,
+      "loss": 1.9443,
+      "step": 2793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041609065743043917,
+      "loss": 2.0312,
+      "step": 2794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041576664982519054,
+      "loss": 1.9512,
+      "step": 2795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004154426786206423,
+      "loss": 2.0059,
+      "step": 2796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041511874395679603,
+      "loss": 2.0195,
+      "step": 2797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041479484597363735,
+      "loss": 1.9697,
+      "step": 2798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041447098481113553,
+      "loss": 1.9033,
+      "step": 2799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041414716060924483,
+      "loss": 1.9404,
+      "step": 2800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041382337350790257,
+      "loss": 1.9648,
+      "step": 2801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041349962364703083,
+      "loss": 1.9795,
+      "step": 2802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041317591116653486,
+      "loss": 2.0967,
+      "step": 2803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004128522362063045,
+      "loss": 1.9551,
+      "step": 2804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004125285989062131,
+      "loss": 2.0049,
+      "step": 2805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041220499940611727,
+      "loss": 1.9043,
+      "step": 2806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041188143784585816,
+      "loss": 2.0156,
+      "step": 2807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041155791436525967,
+      "loss": 2.001,
+      "step": 2808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004112344291041297,
+      "loss": 1.9727,
+      "step": 2809
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004109109822022598,
+      "loss": 2.0518,
+      "step": 2810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004105875737994244,
+      "loss": 1.9961,
+      "step": 2811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041026420403538166,
+      "loss": 1.9365,
+      "step": 2812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040994087304987325,
+      "loss": 1.9863,
+      "step": 2813
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004096175809826235,
+      "loss": 1.9189,
+      "step": 2814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004092943279733402,
+      "loss": 1.9824,
+      "step": 2815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040897111416171464,
+      "loss": 1.9092,
+      "step": 2816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040864793968742053,
+      "loss": 2.0059,
+      "step": 2817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004083248046901149,
+      "loss": 1.8975,
+      "step": 2818
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040800170930943764,
+      "loss": 1.9658,
+      "step": 2819
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004076786536850119,
+      "loss": 1.8955,
+      "step": 2820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004073556379564429,
+      "loss": 1.8838,
+      "step": 2821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004070326622633192,
+      "loss": 2.0234,
+      "step": 2822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040670972674521213,
+      "loss": 1.9531,
+      "step": 2823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040638683154167515,
+      "loss": 1.9766,
+      "step": 2824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040606397679224455,
+      "loss": 1.9834,
+      "step": 2825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004057411626364395,
+      "loss": 1.9883,
+      "step": 2826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004054183892137611,
+      "loss": 1.8818,
+      "step": 2827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004050956566636928,
+      "loss": 1.8994,
+      "step": 2828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040477296512570095,
+      "loss": 1.9111,
+      "step": 2829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004044503147392339,
+      "loss": 1.9688,
+      "step": 2830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004041277056437218,
+      "loss": 2.0293,
+      "step": 2831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040380513797857765,
+      "loss": 1.9307,
+      "step": 2832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004034826118831963,
+      "loss": 1.8916,
+      "step": 2833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004031601274969542,
+      "loss": 1.9941,
+      "step": 2834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040283768495921046,
+      "loss": 1.9355,
+      "step": 2835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004025152844093056,
+      "loss": 2.0264,
+      "step": 2836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040219292598656244,
+      "loss": 2.0029,
+      "step": 2837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040187060983028477,
+      "loss": 1.9736,
+      "step": 2838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004015483360797593,
+      "loss": 2.001,
+      "step": 2839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004012261048742536,
+      "loss": 2.0254,
+      "step": 2840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004009039163530167,
+      "loss": 1.9893,
+      "step": 2841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040058177065528,
+      "loss": 2.0234,
+      "step": 2842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040025966792025586,
+      "loss": 2.0361,
+      "step": 2843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039993760828713784,
+      "loss": 1.8809,
+      "step": 2844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039961559189510154,
+      "loss": 2.0312,
+      "step": 2845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039929361888330317,
+      "loss": 1.9844,
+      "step": 2846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039897168939088067,
+      "loss": 1.9336,
+      "step": 2847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039864980355695323,
+      "loss": 1.8994,
+      "step": 2848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003983279615206208,
+      "loss": 1.9521,
+      "step": 2849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003980061634209643,
+      "loss": 2.0059,
+      "step": 2850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039768440939704663,
+      "loss": 2.002,
+      "step": 2851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003973626995879105,
+      "loss": 1.9365,
+      "step": 2852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003970410341325799,
+      "loss": 1.9609,
+      "step": 2853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039671941317005976,
+      "loss": 2.0361,
+      "step": 2854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003963978368393361,
+      "loss": 1.9775,
+      "step": 2855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000396076305279375,
+      "loss": 1.9531,
+      "step": 2856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003957548186291233,
+      "loss": 1.9609,
+      "step": 2857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039543337702750914,
+      "loss": 1.9053,
+      "step": 2858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039511198061344035,
+      "loss": 2.0459,
+      "step": 2859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003947906295258054,
+      "loss": 1.9404,
+      "step": 2860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039446932390347386,
+      "loss": 1.9482,
+      "step": 2861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039414806388529477,
+      "loss": 1.9463,
+      "step": 2862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003938268496100976,
+      "loss": 1.8555,
+      "step": 2863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039350568121669275,
+      "loss": 1.9072,
+      "step": 2864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003931845588438701,
+      "loss": 1.9307,
+      "step": 2865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039286348263039975,
+      "loss": 2.0732,
+      "step": 2866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003925424527150321,
+      "loss": 2.0098,
+      "step": 2867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003922214692364974,
+      "loss": 1.9736,
+      "step": 2868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039190053233350577,
+      "loss": 2.0039,
+      "step": 2869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003915796421447472,
+      "loss": 1.9395,
+      "step": 2870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003912587988088918,
+      "loss": 1.9893,
+      "step": 2871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003909380024645891,
+      "loss": 1.9736,
+      "step": 2872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003906172532504681,
+      "loss": 2.0049,
+      "step": 2873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003902965513051382,
+      "loss": 2.0938,
+      "step": 2874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038997589676718783,
+      "loss": 1.9072,
+      "step": 2875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003896552897751846,
+      "loss": 2.0488,
+      "step": 2876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038933473046767655,
+      "loss": 1.9727,
+      "step": 2877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003890142189831904,
+      "loss": 1.9395,
+      "step": 2878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038869375546023204,
+      "loss": 1.8867,
+      "step": 2879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003883733400372875,
+      "loss": 1.9971,
+      "step": 2880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003880529728528211,
+      "loss": 1.9824,
+      "step": 2881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003877326540452767,
+      "loss": 2.0391,
+      "step": 2882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003874123837530775,
+      "loss": 1.9521,
+      "step": 2883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038709216211462534,
+      "loss": 1.9473,
+      "step": 2884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003867719892683013,
+      "loss": 1.9844,
+      "step": 2885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003864518653524648,
+      "loss": 1.9893,
+      "step": 2886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003861317905054551,
+      "loss": 1.8955,
+      "step": 2887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038581176486558966,
+      "loss": 1.9521,
+      "step": 2888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003854917885711644,
+      "loss": 2.0098,
+      "step": 2889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003851718617604547,
+      "loss": 2.043,
+      "step": 2890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003848519845717138,
+      "loss": 1.9307,
+      "step": 2891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003845321571431737,
+      "loss": 2.0244,
+      "step": 2892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003842123796130455,
+      "loss": 1.9473,
+      "step": 2893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003838926521195178,
+      "loss": 1.9531,
+      "step": 2894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003835729748007582,
+      "loss": 1.9707,
+      "step": 2895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038325334779491256,
+      "loss": 1.9746,
+      "step": 2896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003829337712401048,
+      "loss": 1.9844,
+      "step": 2897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003826142452744367,
+      "loss": 1.8311,
+      "step": 2898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003822947700359891,
+      "loss": 1.9658,
+      "step": 2899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003819753456628204,
+      "loss": 2.0146,
+      "step": 2900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003816559722929667,
+      "loss": 1.9619,
+      "step": 2901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038133665006444257,
+      "loss": 1.9072,
+      "step": 2902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003810173791152405,
+      "loss": 1.9561,
+      "step": 2903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003806981595833303,
+      "loss": 1.8965,
+      "step": 2904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038037899160665986,
+      "loss": 1.9277,
+      "step": 2905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038005987532315513,
+      "loss": 1.9297,
+      "step": 2906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037974081087071925,
+      "loss": 2.0049,
+      "step": 2907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037942179838723296,
+      "loss": 1.9756,
+      "step": 2908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037910283801055486,
+      "loss": 1.9609,
+      "step": 2909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037878392987852097,
+      "loss": 1.9951,
+      "step": 2910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003784650741289442,
+      "loss": 1.9971,
+      "step": 2911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003781462708996157,
+      "loss": 1.9277,
+      "step": 2912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037782752032830346,
+      "loss": 1.9404,
+      "step": 2913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037750882255275245,
+      "loss": 1.9482,
+      "step": 2914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037719017771068537,
+      "loss": 1.9512,
+      "step": 2915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003768715859398016,
+      "loss": 1.9072,
+      "step": 2916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003765530473777777,
+      "loss": 2.0596,
+      "step": 2917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003762345621622677,
+      "loss": 1.8887,
+      "step": 2918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037591613043090177,
+      "loss": 1.9492,
+      "step": 2919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037559775232128765,
+      "loss": 1.9688,
+      "step": 2920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003752794279710094,
+      "loss": 1.9482,
+      "step": 2921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003749611575176282,
+      "loss": 1.9922,
+      "step": 2922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000374642941098682,
+      "loss": 2.001,
+      "step": 2923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003743247788516848,
+      "loss": 1.9033,
+      "step": 2924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003740066709141281,
+      "loss": 2.0664,
+      "step": 2925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037368861742347916,
+      "loss": 1.9805,
+      "step": 2926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003733706185171819,
+      "loss": 1.9072,
+      "step": 2927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003730526743326571,
+      "loss": 1.9502,
+      "step": 2928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003727347850073012,
+      "loss": 1.916,
+      "step": 2929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037241695067848736,
+      "loss": 1.9717,
+      "step": 2930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003720991714835651,
+      "loss": 1.917,
+      "step": 2931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003717814475598597,
+      "loss": 2.0,
+      "step": 2932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003714637790446726,
+      "loss": 1.9033,
+      "step": 2933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003711461660752819,
+      "loss": 1.874,
+      "step": 2934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000370828608788941,
+      "loss": 1.96,
+      "step": 2935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037051110732287933,
+      "loss": 1.959,
+      "step": 2936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037019366181430247,
+      "loss": 1.8506,
+      "step": 2937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003698762724003919,
+      "loss": 1.9385,
+      "step": 2938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003695589392183043,
+      "loss": 1.8965,
+      "step": 2939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003692416624051727,
+      "loss": 1.8291,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003689244420981054,
+      "loss": 1.9014,
+      "step": 2941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003686072784341864,
+      "loss": 1.9102,
+      "step": 2942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036829017155047496,
+      "loss": 1.9756,
+      "step": 2943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003679731215840063,
+      "loss": 1.9297,
+      "step": 2944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003676561286717908,
+      "loss": 1.8652,
+      "step": 2945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036733919295081375,
+      "loss": 1.9697,
+      "step": 2946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036702231455803657,
+      "loss": 1.957,
+      "step": 2947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003667054936303954,
+      "loss": 1.9062,
+      "step": 2948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036638873030480127,
+      "loss": 2.04,
+      "step": 2949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036607202471814106,
+      "loss": 1.8906,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036575537700727625,
+      "loss": 1.9287,
+      "step": 2951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000365438787309043,
+      "loss": 1.9697,
+      "step": 2952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036512225576025326,
+      "loss": 2.0273,
+      "step": 2953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036480578249769305,
+      "loss": 1.916,
+      "step": 2954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003644893676581237,
+      "loss": 1.998,
+      "step": 2955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036417301137828067,
+      "loss": 2.0264,
+      "step": 2956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036385671379487495,
+      "loss": 1.9668,
+      "step": 2957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003635404750445918,
+      "loss": 1.9238,
+      "step": 2958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003632242952640906,
+      "loss": 2.0293,
+      "step": 2959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003629081745900062,
+      "loss": 1.8926,
+      "step": 2960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003625921131589469,
+      "loss": 1.9453,
+      "step": 2961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036227611110749595,
+      "loss": 1.9082,
+      "step": 2962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036196016857221113,
+      "loss": 1.8994,
+      "step": 2963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000361644285689624,
+      "loss": 1.96,
+      "step": 2964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003613284625962405,
+      "loss": 2.0498,
+      "step": 2965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003610126994285411,
+      "loss": 2.0303,
+      "step": 2966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036069699632297977,
+      "loss": 1.9336,
+      "step": 2967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036038135341598485,
+      "loss": 1.9756,
+      "step": 2968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003600657708439589,
+      "loss": 1.8848,
+      "step": 2969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035975024874327797,
+      "loss": 2.0605,
+      "step": 2970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003594347872502921,
+      "loss": 1.9443,
+      "step": 2971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003591193865013251,
+      "loss": 1.958,
+      "step": 2972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035880404663267515,
+      "loss": 1.96,
+      "step": 2973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000358488767780613,
+      "loss": 1.9795,
+      "step": 2974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035817355008138386,
+      "loss": 1.9512,
+      "step": 2975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035785839367120643,
+      "loss": 2.0078,
+      "step": 2976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035754329868627254,
+      "loss": 1.9883,
+      "step": 2977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003572282652627477,
+      "loss": 2.0137,
+      "step": 2978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035691329353677116,
+      "loss": 1.9551,
+      "step": 2979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035659838364445503,
+      "loss": 2.0137,
+      "step": 2980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003562835357218845,
+      "loss": 1.9004,
+      "step": 2981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035596874990511876,
+      "loss": 1.9355,
+      "step": 2982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003556540263301896,
+      "loss": 1.9521,
+      "step": 2983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035533936513310185,
+      "loss": 1.834,
+      "step": 2984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035502476644983384,
+      "loss": 1.9629,
+      "step": 2985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035471023041633667,
+      "loss": 2.0791,
+      "step": 2986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003543957571685341,
+      "loss": 1.9043,
+      "step": 2987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003540813468423229,
+      "loss": 1.96,
+      "step": 2988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035376699957357293,
+      "loss": 2.0117,
+      "step": 2989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003534527154981267,
+      "loss": 1.9971,
+      "step": 2990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003531384947517989,
+      "loss": 1.9678,
+      "step": 2991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035282433747037755,
+      "loss": 1.8291,
+      "step": 2992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003525102437896231,
+      "loss": 1.9473,
+      "step": 2993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003521962138452679,
+      "loss": 2.002,
+      "step": 2994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035188224777301763,
+      "loss": 1.9521,
+      "step": 2995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003515683457085499,
+      "loss": 1.9482,
+      "step": 2996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003512545077875145,
+      "loss": 1.917,
+      "step": 2997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003509407341455341,
+      "loss": 1.9102,
+      "step": 2998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003506270249182028,
+      "loss": 1.9844,
+      "step": 2999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035031338024108747,
+      "loss": 1.9014,
+      "step": 3000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000349999800249727,
+      "loss": 1.9805,
+      "step": 3001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034968628507963214,
+      "loss": 1.9336,
+      "step": 3002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003493728348662855,
+      "loss": 1.9121,
+      "step": 3003
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034905944974514226,
+      "loss": 1.9316,
+      "step": 3004
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003487461298516288,
+      "loss": 1.9375,
+      "step": 3005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003484328753211435,
+      "loss": 1.915,
+      "step": 3006
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034811968628905643,
+      "loss": 2.0566,
+      "step": 3007
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003478065628907099,
+      "loss": 2.0479,
+      "step": 3008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034749350526141706,
+      "loss": 2.0234,
+      "step": 3009
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000347180513536463,
+      "loss": 1.9297,
+      "step": 3010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003468675878511047,
+      "loss": 1.9336,
+      "step": 3011
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003465547283405699,
+      "loss": 1.9375,
+      "step": 3012
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034624193514005806,
+      "loss": 1.9102,
+      "step": 3013
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003459292083847404,
+      "loss": 1.8984,
+      "step": 3014
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003456165482097587,
+      "loss": 1.8281,
+      "step": 3015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034530395475022613,
+      "loss": 1.9619,
+      "step": 3016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034499142814122766,
+      "loss": 1.9365,
+      "step": 3017
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034467896851781863,
+      "loss": 1.9619,
+      "step": 3018
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034436657601502575,
+      "loss": 1.9922,
+      "step": 3019
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003440542507678467,
+      "loss": 1.9893,
+      "step": 3020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003437419929112503,
+      "loss": 1.9268,
+      "step": 3021
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034342980258017573,
+      "loss": 1.9854,
+      "step": 3022
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034311767990953323,
+      "loss": 1.9424,
+      "step": 3023
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003428056250342042,
+      "loss": 2.0234,
+      "step": 3024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003424936380890403,
+      "loss": 1.9336,
+      "step": 3025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034218171920886363,
+      "loss": 1.9619,
+      "step": 3026
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003418698685284676,
+      "loss": 1.874,
+      "step": 3027
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003415580861826156,
+      "loss": 2.0596,
+      "step": 3028
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003412463723060415,
+      "loss": 2.0312,
+      "step": 3029
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034093472703344987,
+      "loss": 1.8945,
+      "step": 3030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003406231504995155,
+      "loss": 2.0049,
+      "step": 3031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003403116428388832,
+      "loss": 1.9336,
+      "step": 3032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003400002041861685,
+      "loss": 1.9434,
+      "step": 3033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033968883467595676,
+      "loss": 1.8623,
+      "step": 3034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003393775344428035,
+      "loss": 2.0156,
+      "step": 3035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033906630362123466,
+      "loss": 1.9443,
+      "step": 3036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033875514234574553,
+      "loss": 2.001,
+      "step": 3037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003384440507508022,
+      "loss": 1.9639,
+      "step": 3038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003381330289708395,
+      "loss": 1.9609,
+      "step": 3039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033782207714026324,
+      "loss": 1.8701,
+      "step": 3040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003375111953934485,
+      "loss": 2.0059,
+      "step": 3041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003372003838647398,
+      "loss": 1.9541,
+      "step": 3042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033688964268845204,
+      "loss": 1.9082,
+      "step": 3043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003365789719988689,
+      "loss": 1.916,
+      "step": 3044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033626837193024395,
+      "loss": 1.8643,
+      "step": 3045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003359578426168007,
+      "loss": 1.9746,
+      "step": 3046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003356473841927313,
+      "loss": 1.9297,
+      "step": 3047
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003353369967921976,
+      "loss": 1.9795,
+      "step": 3048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033502668054933107,
+      "loss": 1.9932,
+      "step": 3049
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033471643559823206,
+      "loss": 1.9082,
+      "step": 3050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033440626207296986,
+      "loss": 1.7656,
+      "step": 3051
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033409616010758353,
+      "loss": 2.0488,
+      "step": 3052
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033378612983608105,
+      "loss": 1.9609,
+      "step": 3053
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003334761713924388,
+      "loss": 1.9746,
+      "step": 3054
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033316628491060295,
+      "loss": 1.8984,
+      "step": 3055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003328564705244883,
+      "loss": 1.9336,
+      "step": 3056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003325467283679781,
+      "loss": 1.9893,
+      "step": 3057
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033223705857492473,
+      "loss": 1.8486,
+      "step": 3058
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003319274612791496,
+      "loss": 1.9199,
+      "step": 3059
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033161793661444223,
+      "loss": 1.9834,
+      "step": 3060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033130848471456075,
+      "loss": 1.9639,
+      "step": 3061
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003309991057132324,
+      "loss": 1.8223,
+      "step": 3062
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003306897997441526,
+      "loss": 2.0068,
+      "step": 3063
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033038056694098485,
+      "loss": 1.9941,
+      "step": 3064
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033007140743736177,
+      "loss": 1.9297,
+      "step": 3065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003297623213668839,
+      "loss": 1.9297,
+      "step": 3066
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032945330886311966,
+      "loss": 1.9766,
+      "step": 3067
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032914437005960657,
+      "loss": 2.0518,
+      "step": 3068
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003288355050898495,
+      "loss": 1.9619,
+      "step": 3069
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032852671408732176,
+      "loss": 1.9209,
+      "step": 3070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032821799718546496,
+      "loss": 1.9941,
+      "step": 3071
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032790935451768797,
+      "loss": 1.8916,
+      "step": 3072
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032760078621736833,
+      "loss": 1.9941,
+      "step": 3073
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032729229241785086,
+      "loss": 1.9082,
+      "step": 3074
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003269838732524485,
+      "loss": 2.0098,
+      "step": 3075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003266755288544422,
+      "loss": 2.0322,
+      "step": 3076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003263672593570797,
+      "loss": 1.9268,
+      "step": 3077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003260590648935775,
+      "loss": 1.9199,
+      "step": 3078
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032575094559711876,
+      "loss": 1.8594,
+      "step": 3079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032544290160085453,
+      "loss": 2.04,
+      "step": 3080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003251349330379037,
+      "loss": 1.9619,
+      "step": 3081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032482704004135177,
+      "loss": 1.8682,
+      "step": 3082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003245192227442519,
+      "loss": 2.0078,
+      "step": 3083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003242114812796252,
+      "loss": 1.9697,
+      "step": 3084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003239038157804589,
+      "loss": 1.9043,
+      "step": 3085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032359622637970805,
+      "loss": 2.0049,
+      "step": 3086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032328871321029494,
+      "loss": 1.9941,
+      "step": 3087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032298127640510845,
+      "loss": 1.9053,
+      "step": 3088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032267391609700467,
+      "loss": 1.8271,
+      "step": 3089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003223666324188067,
+      "loss": 1.9736,
+      "step": 3090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003220594255033046,
+      "loss": 1.9824,
+      "step": 3091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032175229548325496,
+      "loss": 1.9805,
+      "step": 3092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003214452424913813,
+      "loss": 1.9551,
+      "step": 3093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003211382666603741,
+      "loss": 1.999,
+      "step": 3094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003208313681228902,
+      "loss": 1.9248,
+      "step": 3095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003205245470115527,
+      "loss": 2.043,
+      "step": 3096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003202178034589521,
+      "loss": 1.8545,
+      "step": 3097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003199111375976449,
+      "loss": 2.0273,
+      "step": 3098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003196045495601537,
+      "loss": 1.9453,
+      "step": 3099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003192980394789682,
+      "loss": 1.8779,
+      "step": 3100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000318991607486544,
+      "loss": 1.9912,
+      "step": 3101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003186852537153026,
+      "loss": 1.9541,
+      "step": 3102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031837897829763254,
+      "loss": 1.9824,
+      "step": 3103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003180727813658879,
+      "loss": 1.9551,
+      "step": 3104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003177666630523888,
+      "loss": 1.8848,
+      "step": 3105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031746062348942186,
+      "loss": 1.8906,
+      "step": 3106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031715466280923924,
+      "loss": 1.9424,
+      "step": 3107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003168487811440591,
+      "loss": 1.9658,
+      "step": 3108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003165429786260655,
+      "loss": 1.9346,
+      "step": 3109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003162372553874085,
+      "loss": 1.9248,
+      "step": 3110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003159316115602036,
+      "loss": 1.9004,
+      "step": 3111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031562604727653187,
+      "loss": 1.96,
+      "step": 3112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003153205626684407,
+      "loss": 1.9609,
+      "step": 3113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003150151578679422,
+      "loss": 1.9326,
+      "step": 3114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031470983300701426,
+      "loss": 2.0518,
+      "step": 3115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003144045882176009,
+      "loss": 1.7979,
+      "step": 3116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031409942363161034,
+      "loss": 1.9482,
+      "step": 3117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003137943393809169,
+      "loss": 1.96,
+      "step": 3118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003134893355973606,
+      "loss": 1.8691,
+      "step": 3119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031318441241274544,
+      "loss": 1.8799,
+      "step": 3120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003128795699588416,
+      "loss": 1.9668,
+      "step": 3121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003125748083673842,
+      "loss": 1.9355,
+      "step": 3122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000312270127770073,
+      "loss": 1.9932,
+      "step": 3123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031196552829857317,
+      "loss": 1.9189,
+      "step": 3124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003116610100845145,
+      "loss": 1.9854,
+      "step": 3125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003113565732594923,
+      "loss": 2.0586,
+      "step": 3126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003110522179550658,
+      "loss": 1.8857,
+      "step": 3127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031074794430275963,
+      "loss": 1.9287,
+      "step": 3128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031044375243406317,
+      "loss": 1.9912,
+      "step": 3129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003101396424804301,
+      "loss": 2.0156,
+      "step": 3130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030983561457327856,
+      "loss": 1.8564,
+      "step": 3131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000309531668843992,
+      "loss": 1.8896,
+      "step": 3132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003092278054239178,
+      "loss": 1.9521,
+      "step": 3133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030892402444436756,
+      "loss": 1.9082,
+      "step": 3134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030862032603661785,
+      "loss": 1.8584,
+      "step": 3135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003083167103319093,
+      "loss": 1.9912,
+      "step": 3136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003080131774614464,
+      "loss": 1.917,
+      "step": 3137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003077097275563985,
+      "loss": 1.8525,
+      "step": 3138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000307406360747899,
+      "loss": 1.9834,
+      "step": 3139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030710307716704476,
+      "loss": 1.9199,
+      "step": 3140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030679987694489725,
+      "loss": 1.8945,
+      "step": 3141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030649676021248185,
+      "loss": 2.04,
+      "step": 3142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030619372710078785,
+      "loss": 2.0303,
+      "step": 3143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030589077774076807,
+      "loss": 1.9375,
+      "step": 3144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003055879122633397,
+      "loss": 1.999,
+      "step": 3145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030528513079938334,
+      "loss": 1.9473,
+      "step": 3146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030498243347974307,
+      "loss": 2.0107,
+      "step": 3147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003046798204352272,
+      "loss": 1.958,
+      "step": 3148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030437729179660727,
+      "loss": 1.9824,
+      "step": 3149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030407484769461805,
+      "loss": 1.9785,
+      "step": 3150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003037724882599585,
+      "loss": 2.0459,
+      "step": 3151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003034702136232903,
+      "loss": 1.8926,
+      "step": 3152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003031680239152388,
+      "loss": 1.9297,
+      "step": 3153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030286591926639286,
+      "loss": 1.9434,
+      "step": 3154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003025638998073041,
+      "loss": 2.0234,
+      "step": 3155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003022619656684875,
+      "loss": 1.9102,
+      "step": 3156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003019601169804216,
+      "loss": 1.8672,
+      "step": 3157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030165835387354744,
+      "loss": 1.9951,
+      "step": 3158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003013566764782692,
+      "loss": 1.9033,
+      "step": 3159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030105508492495406,
+      "loss": 1.9404,
+      "step": 3160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030075357934393254,
+      "loss": 1.9668,
+      "step": 3161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030045215986549724,
+      "loss": 1.874,
+      "step": 3162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030015082661990393,
+      "loss": 2.0645,
+      "step": 3163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002998495797373715,
+      "loss": 1.8867,
+      "step": 3164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002995484193480808,
+      "loss": 1.9775,
+      "step": 3165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002992473455821756,
+      "loss": 1.8818,
+      "step": 3166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029894635856976266,
+      "loss": 1.9102,
+      "step": 3167
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002986454584409106,
+      "loss": 2.124,
+      "step": 3168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029834464532565053,
+      "loss": 1.8789,
+      "step": 3169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002980439193539766,
+      "loss": 1.9541,
+      "step": 3170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029774328065584475,
+      "loss": 1.9609,
+      "step": 3171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002974427293611732,
+      "loss": 1.8945,
+      "step": 3172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002971422655998428,
+      "loss": 1.9717,
+      "step": 3173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002968418895016962,
+      "loss": 1.9209,
+      "step": 3174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029654160119653835,
+      "loss": 1.8916,
+      "step": 3175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002962414008141359,
+      "loss": 1.8945,
+      "step": 3176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002959412884842181,
+      "loss": 1.916,
+      "step": 3177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029564126433647586,
+      "loss": 2.0254,
+      "step": 3178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029534132850056173,
+      "loss": 1.9297,
+      "step": 3179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029504148110609063,
+      "loss": 1.916,
+      "step": 3180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029474172228263887,
+      "loss": 2.0625,
+      "step": 3181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002944420521597444,
+      "loss": 1.8877,
+      "step": 3182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029414247086690726,
+      "loss": 1.8398,
+      "step": 3183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029384297853358895,
+      "loss": 1.957,
+      "step": 3184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029354357528921205,
+      "loss": 1.9512,
+      "step": 3185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002932442612631614,
+      "loss": 1.9375,
+      "step": 3186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029294503658478275,
+      "loss": 1.9941,
+      "step": 3187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002926459013833834,
+      "loss": 1.8691,
+      "step": 3188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002923468557882322,
+      "loss": 1.9346,
+      "step": 3189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029204789992855884,
+      "loss": 1.9629,
+      "step": 3190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029174903393355454,
+      "loss": 1.9648,
+      "step": 3191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029145025793237143,
+      "loss": 1.8799,
+      "step": 3192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029115157205412333,
+      "loss": 1.9316,
+      "step": 3193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029085297642788454,
+      "loss": 1.9629,
+      "step": 3194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002905544711826903,
+      "loss": 2.0234,
+      "step": 3195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029025605644753737,
+      "loss": 1.9854,
+      "step": 3196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002899577323513829,
+      "loss": 1.957,
+      "step": 3197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028965949902314475,
+      "loss": 1.9248,
+      "step": 3198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028936135659170215,
+      "loss": 2.04,
+      "step": 3199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002890633051858948,
+      "loss": 2.0146,
+      "step": 3200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028876534493452264,
+      "loss": 1.875,
+      "step": 3201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028846747596634705,
+      "loss": 1.9346,
+      "step": 3202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002881696984100891,
+      "loss": 1.9062,
+      "step": 3203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028787201239443075,
+      "loss": 1.9727,
+      "step": 3204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028757441804801463,
+      "loss": 2.0391,
+      "step": 3205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028727691549944323,
+      "loss": 1.8447,
+      "step": 3206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028697950487728,
+      "loss": 1.9795,
+      "step": 3207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028668218631004786,
+      "loss": 1.9102,
+      "step": 3208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028638495992623094,
+      "loss": 1.9688,
+      "step": 3209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028608782585427275,
+      "loss": 1.9473,
+      "step": 3210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028579078422257686,
+      "loss": 1.9014,
+      "step": 3211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002854938351595078,
+      "loss": 2.0547,
+      "step": 3212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002851969787933889,
+      "loss": 1.9756,
+      "step": 3213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028490021525250434,
+      "loss": 1.9531,
+      "step": 3214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002846035446650981,
+      "loss": 1.9268,
+      "step": 3215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002843069671593734,
+      "loss": 1.9297,
+      "step": 3216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002840104828634935,
+      "loss": 1.9102,
+      "step": 3217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028371409190558196,
+      "loss": 1.9336,
+      "step": 3218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028341779441372095,
+      "loss": 1.9385,
+      "step": 3219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028312159051595314,
+      "loss": 1.9756,
+      "step": 3220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002828254803402807,
+      "loss": 1.8213,
+      "step": 3221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028252946401466464,
+      "loss": 1.9238,
+      "step": 3222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002822335416670257,
+      "loss": 1.916,
+      "step": 3223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028193771342524456,
+      "loss": 2.0801,
+      "step": 3224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002816419794171605,
+      "loss": 1.8779,
+      "step": 3225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002813463397705723,
+      "loss": 1.96,
+      "step": 3226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002810507946132382,
+      "loss": 2.0947,
+      "step": 3227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002807553440728755,
+      "loss": 1.9023,
+      "step": 3228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028045998827716046,
+      "loss": 1.9277,
+      "step": 3229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002801647273537283,
+      "loss": 1.9502,
+      "step": 3230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027986956143017386,
+      "loss": 1.9629,
+      "step": 3231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002795744906340501,
+      "loss": 1.8574,
+      "step": 3232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027927951509286924,
+      "loss": 1.9424,
+      "step": 3233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027898463493410255,
+      "loss": 1.998,
+      "step": 3234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002786898502851801,
+      "loss": 2.0322,
+      "step": 3235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027839516127348997,
+      "loss": 1.9121,
+      "step": 3236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027810056802637993,
+      "loss": 2.0059,
+      "step": 3237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027780607067115563,
+      "loss": 1.9287,
+      "step": 3238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002775116693350813,
+      "loss": 1.9541,
+      "step": 3239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027721736414538037,
+      "loss": 1.9619,
+      "step": 3240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002769231552292336,
+      "loss": 1.957,
+      "step": 3241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002766290427137815,
+      "loss": 1.9268,
+      "step": 3242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027633502672612154,
+      "loss": 1.875,
+      "step": 3243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002760411073933107,
+      "loss": 1.9707,
+      "step": 3244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002757472848423633,
+      "loss": 1.9678,
+      "step": 3245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002754535592002522,
+      "loss": 1.9023,
+      "step": 3246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002751599305939082,
+      "loss": 1.9316,
+      "step": 3247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002748663991502208,
+      "loss": 1.9414,
+      "step": 3248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002745729649960366,
+      "loss": 1.9453,
+      "step": 3249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002742796282581609,
+      "loss": 1.9023,
+      "step": 3250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027398638906335634,
+      "loss": 1.959,
+      "step": 3251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002736932475383436,
+      "loss": 1.8574,
+      "step": 3252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002734002038098015,
+      "loss": 1.9395,
+      "step": 3253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000273107258004366,
+      "loss": 1.9141,
+      "step": 3254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027281441024863115,
+      "loss": 1.9932,
+      "step": 3255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027252166066914874,
+      "loss": 1.9688,
+      "step": 3256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027222900939242783,
+      "loss": 1.9014,
+      "step": 3257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000271936456544935,
+      "loss": 1.8887,
+      "step": 3258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002716440022530942,
+      "loss": 1.9893,
+      "step": 3259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002713516466432874,
+      "loss": 1.8916,
+      "step": 3260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027105938984185306,
+      "loss": 1.9189,
+      "step": 3261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002707672319750877,
+      "loss": 1.9521,
+      "step": 3262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002704751731692448,
+      "loss": 1.8662,
+      "step": 3263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027018321355053485,
+      "loss": 1.9541,
+      "step": 3264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002698913532451256,
+      "loss": 1.9541,
+      "step": 3265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026959959237914214,
+      "loss": 1.9512,
+      "step": 3266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002693079310786661,
+      "loss": 1.9531,
+      "step": 3267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002690163694697363,
+      "loss": 1.8877,
+      "step": 3268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026872490767834863,
+      "loss": 1.9551,
+      "step": 3269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026843354583045597,
+      "loss": 1.9854,
+      "step": 3270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002681422840519674,
+      "loss": 1.8809,
+      "step": 3271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026785112246874944,
+      "loss": 1.9395,
+      "step": 3272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002675600612066249,
+      "loss": 1.8477,
+      "step": 3273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002672691003913732,
+      "loss": 1.9375,
+      "step": 3274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002669782401487307,
+      "loss": 1.874,
+      "step": 3275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002666874806043899,
+      "loss": 1.8789,
+      "step": 3276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026639682188400016,
+      "loss": 1.9863,
+      "step": 3277
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026610626411316683,
+      "loss": 1.9453,
+      "step": 3278
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026581580741745226,
+      "loss": 1.8555,
+      "step": 3279
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002655254519223746,
+      "loss": 1.8975,
+      "step": 3280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002652351977534082,
+      "loss": 1.8818,
+      "step": 3281
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026494504503598394,
+      "loss": 1.9121,
+      "step": 3282
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002646549938954891,
+      "loss": 1.9287,
+      "step": 3283
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002643650444572663,
+      "loss": 1.8633,
+      "step": 3284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026407519684661497,
+      "loss": 1.9326,
+      "step": 3285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026378545118879005,
+      "loss": 1.9326,
+      "step": 3286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026349580760900233,
+      "loss": 1.9004,
+      "step": 3287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002632062662324192,
+      "loss": 1.9443,
+      "step": 3288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000262916827184163,
+      "loss": 1.9492,
+      "step": 3289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002626274905893125,
+      "loss": 1.8955,
+      "step": 3290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000262338256572902,
+      "loss": 1.9795,
+      "step": 3291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026204912525992134,
+      "loss": 1.8789,
+      "step": 3292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026176009677531616,
+      "loss": 1.9502,
+      "step": 3293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026147117124398724,
+      "loss": 1.9238,
+      "step": 3294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002611823487907917,
+      "loss": 1.9131,
+      "step": 3295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026089362954054117,
+      "loss": 1.9033,
+      "step": 3296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026060501361800333,
+      "loss": 1.9717,
+      "step": 3297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002603165011479012,
+      "loss": 1.8662,
+      "step": 3298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026002809225491273,
+      "loss": 1.999,
+      "step": 3299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025973978706367106,
+      "loss": 2.0869,
+      "step": 3300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025945158569876526,
+      "loss": 1.9951,
+      "step": 3301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025916348828473857,
+      "loss": 1.8887,
+      "step": 3302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025887549494608975,
+      "loss": 1.9062,
+      "step": 3303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025858760580727274,
+      "loss": 1.9678,
+      "step": 3304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002582998209926965,
+      "loss": 1.9297,
+      "step": 3305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025801214062672425,
+      "loss": 1.9668,
+      "step": 3306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000257724564833675,
+      "loss": 1.8633,
+      "step": 3307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025743709373782177,
+      "loss": 1.9277,
+      "step": 3308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002571497274633926,
+      "loss": 1.9492,
+      "step": 3309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025686246613457046,
+      "loss": 1.8955,
+      "step": 3310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000256575309875493,
+      "loss": 1.8516,
+      "step": 3311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000256288258810252,
+      "loss": 1.9814,
+      "step": 3312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025600131306289397,
+      "loss": 1.9014,
+      "step": 3313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002557144727574203,
+      "loss": 2.0,
+      "step": 3314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002554277380177864,
+      "loss": 1.8975,
+      "step": 3315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025514110896790183,
+      "loss": 1.8975,
+      "step": 3316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002548545857316311,
+      "loss": 1.9199,
+      "step": 3317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002545681684327928,
+      "loss": 1.9678,
+      "step": 3318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025428185719515945,
+      "loss": 1.874,
+      "step": 3319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002539956521424581,
+      "loss": 1.9053,
+      "step": 3320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002537095533983697,
+      "loss": 1.9258,
+      "step": 3321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025342356108652903,
+      "loss": 1.9883,
+      "step": 3322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025313767533052556,
+      "loss": 1.959,
+      "step": 3323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025285189625390194,
+      "loss": 2.0654,
+      "step": 3324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002525662239801553,
+      "loss": 1.9336,
+      "step": 3325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002522806586327364,
+      "loss": 1.916,
+      "step": 3326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025199520033504985,
+      "loss": 1.9766,
+      "step": 3327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025170984921045383,
+      "loss": 1.9043,
+      "step": 3328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002514246053822601,
+      "loss": 1.9932,
+      "step": 3329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002511394689737346,
+      "loss": 1.9307,
+      "step": 3330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025085444010809635,
+      "loss": 1.9707,
+      "step": 3331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002505695189085181,
+      "loss": 1.9053,
+      "step": 3332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002502847054981262,
+      "loss": 1.9365,
+      "step": 3333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 1.9053,
+      "step": 3334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002497154025371726,
+      "loss": 1.9062,
+      "step": 3335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002494309132326304,
+      "loss": 1.9678,
+      "step": 3336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002491465322093128,
+      "loss": 1.8613,
+      "step": 3337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024886225959011246,
+      "loss": 1.9395,
+      "step": 3338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002485780954978753,
+      "loss": 1.9258,
+      "step": 3339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002482940400554007,
+      "loss": 1.832,
+      "step": 3340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024801009338544023,
+      "loss": 1.9277,
+      "step": 3341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002477262556106994,
+      "loss": 1.9385,
+      "step": 3342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024744252685383596,
+      "loss": 1.8115,
+      "step": 3343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002471589072374606,
+      "loss": 1.8652,
+      "step": 3344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024687539688413723,
+      "loss": 1.9062,
+      "step": 3345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024659199591638267,
+      "loss": 1.9131,
+      "step": 3346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024630870445666573,
+      "loss": 1.9268,
+      "step": 3347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024602552262740834,
+      "loss": 1.9121,
+      "step": 3348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002457424505509853,
+      "loss": 1.9121,
+      "step": 3349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002454594883497235,
+      "loss": 1.9736,
+      "step": 3350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002451766361459025,
+      "loss": 1.9668,
+      "step": 3351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002448938940617546,
+      "loss": 1.9688,
+      "step": 3352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024461126221946423,
+      "loss": 1.8438,
+      "step": 3353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024432874074116815,
+      "loss": 1.9883,
+      "step": 3354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024404632974895575,
+      "loss": 1.9629,
+      "step": 3355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002437640293648683,
+      "loss": 1.9473,
+      "step": 3356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024348183971089927,
+      "loss": 1.9072,
+      "step": 3357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024319976090899482,
+      "loss": 1.9189,
+      "step": 3358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024291779308105237,
+      "loss": 1.9287,
+      "step": 3359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024263593634892233,
+      "loss": 2.0117,
+      "step": 3360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024235419083440613,
+      "loss": 1.8906,
+      "step": 3361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000242072556659258,
+      "loss": 1.8584,
+      "step": 3362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024179103394518349,
+      "loss": 1.9131,
+      "step": 3363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024150962281384005,
+      "loss": 1.9004,
+      "step": 3364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024122832338683726,
+      "loss": 1.9658,
+      "step": 3365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002409471357857359,
+      "loss": 1.8203,
+      "step": 3366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002406660601320489,
+      "loss": 1.8594,
+      "step": 3367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024038509654724077,
+      "loss": 2.0225,
+      "step": 3368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024010424515272732,
+      "loss": 1.9385,
+      "step": 3369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023982350606987585,
+      "loss": 1.9697,
+      "step": 3370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023954287942000552,
+      "loss": 1.9648,
+      "step": 3371
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023926236532438645,
+      "loss": 1.9346,
+      "step": 3372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023898196390424042,
+      "loss": 2.0107,
+      "step": 3373
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023870167528074066,
+      "loss": 1.8604,
+      "step": 3374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002384214995750112,
+      "loss": 1.9443,
+      "step": 3375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002381414369081274,
+      "loss": 1.958,
+      "step": 3376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023786148740111618,
+      "loss": 1.8877,
+      "step": 3377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023758165117495505,
+      "loss": 1.957,
+      "step": 3378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002373019283505727,
+      "loss": 1.9414,
+      "step": 3379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023702231904884895,
+      "loss": 1.8691,
+      "step": 3380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023674282339061477,
+      "loss": 2.0078,
+      "step": 3381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002364634414966515,
+      "loss": 1.9258,
+      "step": 3382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023618417348769138,
+      "loss": 1.9043,
+      "step": 3383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023590501948441805,
+      "loss": 1.8379,
+      "step": 3384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023562597960746524,
+      "loss": 1.9941,
+      "step": 3385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023534705397741735,
+      "loss": 1.8428,
+      "step": 3386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023506824271480987,
+      "loss": 1.8096,
+      "step": 3387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002347895459401288,
+      "loss": 1.9336,
+      "step": 3388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023451096377381014,
+      "loss": 1.8682,
+      "step": 3389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023423249633624106,
+      "loss": 2.0273,
+      "step": 3390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023395414374775865,
+      "loss": 1.9238,
+      "step": 3391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023367590612865036,
+      "loss": 1.8975,
+      "step": 3392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002333977835991545,
+      "loss": 1.9121,
+      "step": 3393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023311977627945885,
+      "loss": 1.9033,
+      "step": 3394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023284188428970222,
+      "loss": 1.8281,
+      "step": 3395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023256410774997284,
+      "loss": 1.8721,
+      "step": 3396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002322864467803097,
+      "loss": 1.9453,
+      "step": 3397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023200890150070147,
+      "loss": 1.8477,
+      "step": 3398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023173147203108663,
+      "loss": 1.96,
+      "step": 3399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023145415849135426,
+      "loss": 1.8496,
+      "step": 3400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023117696100134261,
+      "loss": 1.9766,
+      "step": 3401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023089987968084036,
+      "loss": 1.916,
+      "step": 3402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023062291464958585,
+      "loss": 1.9727,
+      "step": 3403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000230346066027267,
+      "loss": 1.9355,
+      "step": 3404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002300693339335212,
+      "loss": 1.8652,
+      "step": 3405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002297927184879363,
+      "loss": 1.8809,
+      "step": 3406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022951621981004872,
+      "loss": 1.9336,
+      "step": 3407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022923983801934522,
+      "loss": 1.9316,
+      "step": 3408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002289635732352619,
+      "loss": 1.9609,
+      "step": 3409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022868742557718402,
+      "loss": 1.9111,
+      "step": 3410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022841139516444626,
+      "loss": 1.9639,
+      "step": 3411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002281354821163326,
+      "loss": 1.8564,
+      "step": 3412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002278596865520768,
+      "loss": 1.8984,
+      "step": 3413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022758400859086114,
+      "loss": 2.0312,
+      "step": 3414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022730844835181757,
+      "loss": 1.8945,
+      "step": 3415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002270330059540272,
+      "loss": 1.9619,
+      "step": 3416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022675768151651993,
+      "loss": 1.9502,
+      "step": 3417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022648247515827453,
+      "loss": 1.9072,
+      "step": 3418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022620738699821936,
+      "loss": 2.0205,
+      "step": 3419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002259324171552312,
+      "loss": 1.9053,
+      "step": 3420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022565756574813574,
+      "loss": 1.8613,
+      "step": 3421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002253828328957076,
+      "loss": 1.9746,
+      "step": 3422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022510821871667047,
+      "loss": 1.9854,
+      "step": 3423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002248337233296961,
+      "loss": 1.8799,
+      "step": 3424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022455934685340558,
+      "loss": 1.8652,
+      "step": 3425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022428508940636804,
+      "loss": 1.8857,
+      "step": 3426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022401095110710135,
+      "loss": 1.8896,
+      "step": 3427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022373693207407226,
+      "loss": 1.9121,
+      "step": 3428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022346303242569537,
+      "loss": 2.0449,
+      "step": 3429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002231892522803343,
+      "loss": 1.9258,
+      "step": 3430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022291559175630044,
+      "loss": 1.8936,
+      "step": 3431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022264205097185407,
+      "loss": 1.9395,
+      "step": 3432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022236863004520323,
+      "loss": 1.9082,
+      "step": 3433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002220953290945043,
+      "loss": 1.9131,
+      "step": 3434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000221822148237862,
+      "loss": 1.9844,
+      "step": 3435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002215490875933292,
+      "loss": 1.8896,
+      "step": 3436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022127614727890626,
+      "loss": 1.8848,
+      "step": 3437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022100332741254236,
+      "loss": 1.9775,
+      "step": 3438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000220730628112134,
+      "loss": 1.9473,
+      "step": 3439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022045804949552555,
+      "loss": 1.8906,
+      "step": 3440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022018559168050995,
+      "loss": 1.9248,
+      "step": 3441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021991325478482694,
+      "loss": 1.9004,
+      "step": 3442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021964103892616477,
+      "loss": 1.835,
+      "step": 3443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002193689442221593,
+      "loss": 1.9688,
+      "step": 3444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021909697079039363,
+      "loss": 1.8291,
+      "step": 3445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021882511874839878,
+      "loss": 1.9551,
+      "step": 3446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000218553388213653,
+      "loss": 1.8857,
+      "step": 3447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021828177930358251,
+      "loss": 1.9297,
+      "step": 3448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021801029213556052,
+      "loss": 1.9639,
+      "step": 3449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021773892682690782,
+      "loss": 1.957,
+      "step": 3450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021746768349489281,
+      "loss": 1.9707,
+      "step": 3451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002171965622567308,
+      "loss": 1.8408,
+      "step": 3452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021692556322958412,
+      "loss": 1.8887,
+      "step": 3453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021665468653056307,
+      "loss": 1.8164,
+      "step": 3454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021638393227672444,
+      "loss": 1.8701,
+      "step": 3455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021611330058507206,
+      "loss": 1.8818,
+      "step": 3456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002158427915725573,
+      "loss": 1.8828,
+      "step": 3457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021557240535607835,
+      "loss": 2.0039,
+      "step": 3458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021530214205247995,
+      "loss": 1.9863,
+      "step": 3459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021503200177855426,
+      "loss": 1.8877,
+      "step": 3460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002147619846510399,
+      "loss": 1.8652,
+      "step": 3461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021449209078662224,
+      "loss": 1.8467,
+      "step": 3462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021422232030193374,
+      "loss": 1.8848,
+      "step": 3463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002139526733135534,
+      "loss": 1.9561,
+      "step": 3464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021368314993800675,
+      "loss": 1.7588,
+      "step": 3465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021341375029176562,
+      "loss": 1.9668,
+      "step": 3466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021314447449124918,
+      "loss": 1.8691,
+      "step": 3467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021287532265282239,
+      "loss": 1.8525,
+      "step": 3468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002126062948927966,
+      "loss": 1.9512,
+      "step": 3469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021233739132743007,
+      "loss": 1.9561,
+      "step": 3470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021206861207292716,
+      "loss": 1.8857,
+      "step": 3471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021179995724543823,
+      "loss": 1.915,
+      "step": 3472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021153142696106036,
+      "loss": 1.876,
+      "step": 3473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021126302133583647,
+      "loss": 1.877,
+      "step": 3474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002109947404857554,
+      "loss": 1.9785,
+      "step": 3475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021072658452675274,
+      "loss": 1.793,
+      "step": 3476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021045855357470944,
+      "loss": 1.9277,
+      "step": 3477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021019064774545283,
+      "loss": 1.9092,
+      "step": 3478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020992286715475623,
+      "loss": 1.9795,
+      "step": 3479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020965521191833848,
+      "loss": 1.877,
+      "step": 3480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002093876821518645,
+      "loss": 1.8721,
+      "step": 3481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002091202779709447,
+      "loss": 1.9912,
+      "step": 3482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020885299949113578,
+      "loss": 1.8994,
+      "step": 3483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020858584682793946,
+      "loss": 1.9248,
+      "step": 3484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002083188200968036,
+      "loss": 1.959,
+      "step": 3485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020805191941312157,
+      "loss": 1.8428,
+      "step": 3486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020778514489223205,
+      "loss": 1.9062,
+      "step": 3487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000207518496649419,
+      "loss": 1.9434,
+      "step": 3488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020725197479991265,
+      "loss": 1.9385,
+      "step": 3489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020698557945888774,
+      "loss": 1.9229,
+      "step": 3490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002067193107414645,
+      "loss": 1.835,
+      "step": 3491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020645316876270886,
+      "loss": 1.9746,
+      "step": 3492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002061871536376319,
+      "loss": 1.9258,
+      "step": 3493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020592126548118933,
+      "loss": 1.958,
+      "step": 3494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020565550440828278,
+      "loss": 1.9453,
+      "step": 3495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002053898705337583,
+      "loss": 1.8408,
+      "step": 3496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002051243639724071,
+      "loss": 1.8701,
+      "step": 3497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002048589848389657,
+      "loss": 1.8574,
+      "step": 3498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020459373324811548,
+      "loss": 1.9717,
+      "step": 3499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020432860931448243,
+      "loss": 1.9238,
+      "step": 3500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020406361315263732,
+      "loss": 1.8936,
+      "step": 3501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020379874487709633,
+      "loss": 1.9053,
+      "step": 3502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020353400460231975,
+      "loss": 1.9541,
+      "step": 3503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020326939244271253,
+      "loss": 1.9238,
+      "step": 3504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002030049085126247,
+      "loss": 1.9219,
+      "step": 3505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020274055292635086,
+      "loss": 1.9238,
+      "step": 3506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020247632579812957,
+      "loss": 1.8965,
+      "step": 3507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002022122272421446,
+      "loss": 1.9043,
+      "step": 3508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020194825737252366,
+      "loss": 1.873,
+      "step": 3509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020168441630333877,
+      "loss": 1.9014,
+      "step": 3510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020142070414860702,
+      "loss": 1.9707,
+      "step": 3511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020115712102228877,
+      "loss": 1.9365,
+      "step": 3512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020089366703828965,
+      "loss": 1.8818,
+      "step": 3513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020063034231045852,
+      "loss": 1.9316,
+      "step": 3514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020036714695258922,
+      "loss": 2.0107,
+      "step": 3515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020010408107841915,
+      "loss": 1.8604,
+      "step": 3516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019984114480162973,
+      "loss": 1.9785,
+      "step": 3517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019957833823584692,
+      "loss": 1.8213,
+      "step": 3518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019931566149463992,
+      "loss": 1.8486,
+      "step": 3519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019905311469152226,
+      "loss": 1.9648,
+      "step": 3520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019879069793995153,
+      "loss": 1.8408,
+      "step": 3521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019852841135332855,
+      "loss": 1.8604,
+      "step": 3522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019826625504499806,
+      "loss": 1.9932,
+      "step": 3523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001980042291282489,
+      "loss": 1.8242,
+      "step": 3524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001977423337163129,
+      "loss": 1.9434,
+      "step": 3525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019748056892236605,
+      "loss": 1.915,
+      "step": 3526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019721893485952785,
+      "loss": 1.8662,
+      "step": 3527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019695743164086104,
+      "loss": 1.9355,
+      "step": 3528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019669605937937192,
+      "loss": 1.8926,
+      "step": 3529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019643481818801002,
+      "loss": 1.8916,
+      "step": 3530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019617370817966872,
+      "loss": 2.0068,
+      "step": 3531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019591272946718413,
+      "loss": 1.9365,
+      "step": 3532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019565188216333611,
+      "loss": 1.9648,
+      "step": 3533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019539116638084758,
+      "loss": 1.9678,
+      "step": 3534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001951305822323845,
+      "loss": 1.8018,
+      "step": 3535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019487012983055585,
+      "loss": 2.0508,
+      "step": 3536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019460980928791405,
+      "loss": 1.9414,
+      "step": 3537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019434962071695434,
+      "loss": 1.9639,
+      "step": 3538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019408956423011458,
+      "loss": 1.7695,
+      "step": 3539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001938296399397762,
+      "loss": 1.9668,
+      "step": 3540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019356984795826333,
+      "loss": 1.9014,
+      "step": 3541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019331018839784236,
+      "loss": 1.9873,
+      "step": 3542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001930506613707233,
+      "loss": 1.8232,
+      "step": 3543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019279126698905835,
+      "loss": 1.8818,
+      "step": 3544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001925320053649422,
+      "loss": 1.8994,
+      "step": 3545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019227287661041293,
+      "loss": 1.8438,
+      "step": 3546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019201388083745036,
+      "loss": 1.8203,
+      "step": 3547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019175501815797759,
+      "loss": 1.8848,
+      "step": 3548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001914962886838595,
+      "loss": 1.8564,
+      "step": 3549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001912376925269041,
+      "loss": 1.9277,
+      "step": 3550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019097922979886128,
+      "loss": 1.9355,
+      "step": 3551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001907209006114232,
+      "loss": 1.874,
+      "step": 3552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019046270507622505,
+      "loss": 2.0264,
+      "step": 3553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019020464330484323,
+      "loss": 1.918,
+      "step": 3554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018994671540879704,
+      "loss": 1.874,
+      "step": 3555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018968892149954809,
+      "loss": 1.9355,
+      "step": 3556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018943126168849945,
+      "loss": 1.8711,
+      "step": 3557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018917373608699634,
+      "loss": 1.9404,
+      "step": 3558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018891634480632663,
+      "loss": 1.9219,
+      "step": 3559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018865908795771924,
+      "loss": 1.958,
+      "step": 3560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018840196565234575,
+      "loss": 2.0029,
+      "step": 3561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018814497800131937,
+      "loss": 2.0322,
+      "step": 3562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018788812511569492,
+      "loss": 1.9082,
+      "step": 3563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001876314071064691,
+      "loss": 1.8994,
+      "step": 3564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018737482408458012,
+      "loss": 2.0215,
+      "step": 3565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018711837616090855,
+      "loss": 2.041,
+      "step": 3566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001868620634462756,
+      "loss": 1.9834,
+      "step": 3567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018660588605144484,
+      "loss": 1.8223,
+      "step": 3568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018634984408712118,
+      "loss": 1.916,
+      "step": 3569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018609393766395082,
+      "loss": 1.9482,
+      "step": 3570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001858381668925212,
+      "loss": 1.8359,
+      "step": 3571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018558253188336176,
+      "loss": 1.916,
+      "step": 3572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853270327469428,
+      "loss": 1.9629,
+      "step": 3573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018507166959367578,
+      "loss": 1.9688,
+      "step": 3574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001848164425339139,
+      "loss": 1.9229,
+      "step": 3575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018456135167795135,
+      "loss": 1.9492,
+      "step": 3576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018430639713602314,
+      "loss": 1.9844,
+      "step": 3577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018405157901830595,
+      "loss": 1.8994,
+      "step": 3578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000183796897434917,
+      "loss": 1.8691,
+      "step": 3579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018354235249591473,
+      "loss": 1.876,
+      "step": 3580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018328794431129826,
+      "loss": 1.9092,
+      "step": 3581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001830336729910081,
+      "loss": 1.8662,
+      "step": 3582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018277953864492548,
+      "loss": 1.877,
+      "step": 3583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018252554138287191,
+      "loss": 1.9521,
+      "step": 3584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001822716813146106,
+      "loss": 1.7842,
+      "step": 3585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001820179585498446,
+      "loss": 1.9863,
+      "step": 3586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018176437319821786,
+      "loss": 1.874,
+      "step": 3587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018151092536931523,
+      "loss": 1.7959,
+      "step": 3588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001812576151726621,
+      "loss": 1.9062,
+      "step": 3589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000181004442717724,
+      "loss": 1.8896,
+      "step": 3590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018075140811390738,
+      "loss": 1.8467,
+      "step": 3591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001804985114705588,
+      "loss": 2.0049,
+      "step": 3592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001802457528969652,
+      "loss": 1.835,
+      "step": 3593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001799931325023542,
+      "loss": 1.9424,
+      "step": 3594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017974065039589332,
+      "loss": 1.8838,
+      "step": 3595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017948830668669047,
+      "loss": 1.7344,
+      "step": 3596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000179236101483794,
+      "loss": 1.9717,
+      "step": 3597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017898403489619198,
+      "loss": 1.8896,
+      "step": 3598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017873210703281284,
+      "loss": 1.9502,
+      "step": 3599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001784803180025247,
+      "loss": 1.873,
+      "step": 3600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017822866791413638,
+      "loss": 1.9121,
+      "step": 3601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017797715687639594,
+      "loss": 1.8994,
+      "step": 3602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001777257849979918,
+      "loss": 1.916,
+      "step": 3603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017747455238755223,
+      "loss": 1.8867,
+      "step": 3604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017722345915364507,
+      "loss": 1.9668,
+      "step": 3605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017697250540477788,
+      "loss": 1.9365,
+      "step": 3606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001767216912493984,
+      "loss": 1.8994,
+      "step": 3607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017647101679589373,
+      "loss": 1.8525,
+      "step": 3608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017622048215259028,
+      "loss": 1.8984,
+      "step": 3609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001759700874277546,
+      "loss": 1.8809,
+      "step": 3610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017571983272959284,
+      "loss": 1.8457,
+      "step": 3611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017546971816624995,
+      "loss": 1.9189,
+      "step": 3612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017521974384581112,
+      "loss": 1.9316,
+      "step": 3613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001749699098763003,
+      "loss": 1.9951,
+      "step": 3614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017472021636568108,
+      "loss": 1.9102,
+      "step": 3615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001744706634218562,
+      "loss": 1.9346,
+      "step": 3616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017422125115266785,
+      "loss": 1.7627,
+      "step": 3617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017397197966589756,
+      "loss": 1.9785,
+      "step": 3618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017372284906926543,
+      "loss": 1.8164,
+      "step": 3619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017347385947043143,
+      "loss": 1.8447,
+      "step": 3620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001732250109769941,
+      "loss": 1.8643,
+      "step": 3621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017297630369649087,
+      "loss": 1.9375,
+      "step": 3622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001727277377363986,
+      "loss": 1.8643,
+      "step": 3623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017247931320413308,
+      "loss": 1.9355,
+      "step": 3624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017223103020704838,
+      "loss": 1.9668,
+      "step": 3625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017198288885243818,
+      "loss": 1.8525,
+      "step": 3626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001717348892475345,
+      "loss": 1.8545,
+      "step": 3627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017148703149950784,
+      "loss": 1.8525,
+      "step": 3628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017123931571546826,
+      "loss": 1.8545,
+      "step": 3629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001709917420024635,
+      "loss": 1.9131,
+      "step": 3630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017074431046748073,
+      "loss": 1.9521,
+      "step": 3631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000170497021217445,
+      "loss": 1.9072,
+      "step": 3632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017024987435922053,
+      "loss": 1.8945,
+      "step": 3633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001700028699996094,
+      "loss": 1.8887,
+      "step": 3634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016975600824535226,
+      "loss": 1.9229,
+      "step": 3635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016950928920312857,
+      "loss": 1.8027,
+      "step": 3636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001692627129795555,
+      "loss": 1.8887,
+      "step": 3637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016901627968118888,
+      "loss": 1.8662,
+      "step": 3638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016876998941452292,
+      "loss": 1.9814,
+      "step": 3639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016852384228598967,
+      "loss": 1.9238,
+      "step": 3640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016827783840195915,
+      "loss": 1.8379,
+      "step": 3641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016803197786874013,
+      "loss": 1.9619,
+      "step": 3642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016778626079257903,
+      "loss": 1.8809,
+      "step": 3643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016754068727966003,
+      "loss": 1.8154,
+      "step": 3644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016729525743610575,
+      "loss": 1.9131,
+      "step": 3645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016704997136797673,
+      "loss": 1.9258,
+      "step": 3646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016680482918127087,
+      "loss": 1.8535,
+      "step": 3647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001665598309819245,
+      "loss": 1.958,
+      "step": 3648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016631497687581126,
+      "loss": 1.8252,
+      "step": 3649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016607026696874262,
+      "loss": 1.8262,
+      "step": 3650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016582570136646784,
+      "loss": 1.8926,
+      "step": 3651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016558128017467406,
+      "loss": 1.748,
+      "step": 3652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016533700349898552,
+      "loss": 1.8984,
+      "step": 3653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001650928714449641,
+      "loss": 1.874,
+      "step": 3654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016484888411810966,
+      "loss": 1.8672,
+      "step": 3655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016460504162385902,
+      "loss": 2.0068,
+      "step": 3656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016436134406758635,
+      "loss": 1.9111,
+      "step": 3657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001641177915546036,
+      "loss": 1.9453,
+      "step": 3658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016387438419016,
+      "loss": 1.9736,
+      "step": 3659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016363112207944152,
+      "loss": 1.9795,
+      "step": 3660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001633880053275722,
+      "loss": 1.998,
+      "step": 3661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016314503403961256,
+      "loss": 1.8604,
+      "step": 3662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001629022083205603,
+      "loss": 1.916,
+      "step": 3663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001626595282753508,
+      "loss": 1.8154,
+      "step": 3664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016241699400885575,
+      "loss": 1.959,
+      "step": 3665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016217460562588455,
+      "loss": 1.9307,
+      "step": 3666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001619323632311828,
+      "loss": 1.8711,
+      "step": 3667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016169026692943388,
+      "loss": 1.7812,
+      "step": 3668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016144831682525734,
+      "loss": 1.9678,
+      "step": 3669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001612065130232096,
+      "loss": 1.9834,
+      "step": 3670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016096485562778446,
+      "loss": 1.9248,
+      "step": 3671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016072334474341154,
+      "loss": 1.9336,
+      "step": 3672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000160481980474458,
+      "loss": 1.9385,
+      "step": 3673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016024076292522733,
+      "loss": 1.96,
+      "step": 3674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001599996921999594,
+      "loss": 1.8779,
+      "step": 3675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001597587684028307,
+      "loss": 1.8896,
+      "step": 3676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001595179916379546,
+      "loss": 1.8584,
+      "step": 3677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001592773620093804,
+      "loss": 1.8408,
+      "step": 3678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015903687962109437,
+      "loss": 1.8525,
+      "step": 3679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015879654457701846,
+      "loss": 2.0322,
+      "step": 3680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015855635698101177,
+      "loss": 1.8496,
+      "step": 3681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015831631693686904,
+      "loss": 1.8828,
+      "step": 3682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015807642454832133,
+      "loss": 1.792,
+      "step": 3683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015783667991903634,
+      "loss": 1.8789,
+      "step": 3684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001575970831526172,
+      "loss": 1.9746,
+      "step": 3685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015735763435260382,
+      "loss": 1.9365,
+      "step": 3686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015711833362247202,
+      "loss": 1.7979,
+      "step": 3687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015687918106563326,
+      "loss": 1.9541,
+      "step": 3688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015664017678543519,
+      "loss": 1.79,
+      "step": 3689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015640132088516162,
+      "loss": 1.9336,
+      "step": 3690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001561626134680319,
+      "loss": 1.9307,
+      "step": 3691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015592405463720117,
+      "loss": 1.8896,
+      "step": 3692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015568564449576066,
+      "loss": 1.9336,
+      "step": 3693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015544738314673745,
+      "loss": 1.9814,
+      "step": 3694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001552092706930937,
+      "loss": 1.9229,
+      "step": 3695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015497130723772797,
+      "loss": 1.7764,
+      "step": 3696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001547334928834739,
+      "loss": 1.875,
+      "step": 3697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015449582773310067,
+      "loss": 1.8115,
+      "step": 3698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015425831188931356,
+      "loss": 1.7969,
+      "step": 3699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015402094545475254,
+      "loss": 1.9326,
+      "step": 3700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015378372853199385,
+      "loss": 1.918,
+      "step": 3701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001535466612235484,
+      "loss": 1.9355,
+      "step": 3702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001533097436318629,
+      "loss": 1.8447,
+      "step": 3703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015307297585931916,
+      "loss": 1.8379,
+      "step": 3704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015283635800823414,
+      "loss": 1.9443,
+      "step": 3705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001525998901808604,
+      "loss": 1.8711,
+      "step": 3706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001523635724793852,
+      "loss": 1.9375,
+      "step": 3707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015212740500593125,
+      "loss": 1.9932,
+      "step": 3708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015189138786255642,
+      "loss": 1.8789,
+      "step": 3709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015165552115125325,
+      "loss": 1.9785,
+      "step": 3710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015141980497394936,
+      "loss": 1.8301,
+      "step": 3711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001511842394325077,
+      "loss": 1.9268,
+      "step": 3712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015094882462872555,
+      "loss": 1.9434,
+      "step": 3713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001507135606643355,
+      "loss": 2.0225,
+      "step": 3714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000150478447641005,
+      "loss": 1.9561,
+      "step": 3715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001502434856603358,
+      "loss": 1.8193,
+      "step": 3716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015000867482386476,
+      "loss": 1.8555,
+      "step": 3717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001497740152330631,
+      "loss": 1.8408,
+      "step": 3718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014953950698933722,
+      "loss": 1.8486,
+      "step": 3719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001493051501940275,
+      "loss": 1.8555,
+      "step": 3720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014907094494840927,
+      "loss": 1.8984,
+      "step": 3721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014883689135369255,
+      "loss": 1.8291,
+      "step": 3722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014860298951102126,
+      "loss": 1.8311,
+      "step": 3723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014836923952147386,
+      "loss": 1.9053,
+      "step": 3724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014813564148606374,
+      "loss": 1.8896,
+      "step": 3725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014790219550573798,
+      "loss": 1.7988,
+      "step": 3726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014766890168137813,
+      "loss": 1.8301,
+      "step": 3727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014743576011380029,
+      "loss": 1.9062,
+      "step": 3728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001472027709037545,
+      "loss": 1.9463,
+      "step": 3729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014696993415192484,
+      "loss": 1.8555,
+      "step": 3730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014673724995892983,
+      "loss": 1.8027,
+      "step": 3731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014650471842532194,
+      "loss": 1.7715,
+      "step": 3732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014627233965158743,
+      "loss": 1.9746,
+      "step": 3733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014604011373814667,
+      "loss": 1.9297,
+      "step": 3734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014580804078535424,
+      "loss": 1.9932,
+      "step": 3735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014557612089349854,
+      "loss": 1.8701,
+      "step": 3736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014534435416280135,
+      "loss": 1.8857,
+      "step": 3737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014511274069341896,
+      "loss": 1.958,
+      "step": 3738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014488128058544097,
+      "loss": 1.8643,
+      "step": 3739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014464997393889067,
+      "loss": 1.9141,
+      "step": 3740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014441882085372532,
+      "loss": 1.8926,
+      "step": 3741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014418782142983587,
+      "loss": 1.8945,
+      "step": 3742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014395697576704636,
+      "loss": 1.9424,
+      "step": 3743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000143726283965115,
+      "loss": 1.9648,
+      "step": 3744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001434957461237331,
+      "loss": 1.8691,
+      "step": 3745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001432653623425254,
+      "loss": 1.8535,
+      "step": 3746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014303513272105056,
+      "loss": 1.8428,
+      "step": 3747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001428050573587999,
+      "loss": 1.8857,
+      "step": 3748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014257513635519865,
+      "loss": 1.793,
+      "step": 3749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014234536980960532,
+      "loss": 1.874,
+      "step": 3750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001421157578213113,
+      "loss": 1.8867,
+      "step": 3751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014188630048954148,
+      "loss": 1.918,
+      "step": 3752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001416569979134536,
+      "loss": 1.8623,
+      "step": 3753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014142785019213906,
+      "loss": 1.8721,
+      "step": 3754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014119885742462169,
+      "loss": 1.9316,
+      "step": 3755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014097001970985896,
+      "loss": 1.9297,
+      "step": 3756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014074133714674113,
+      "loss": 1.9814,
+      "step": 3757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014051280983409125,
+      "loss": 1.792,
+      "step": 3758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001402844378706652,
+      "loss": 1.8926,
+      "step": 3759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014005622135515223,
+      "loss": 1.9629,
+      "step": 3760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013982816038617398,
+      "loss": 1.918,
+      "step": 3761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001396002550622848,
+      "loss": 1.8369,
+      "step": 3762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001393725054819721,
+      "loss": 1.8379,
+      "step": 3763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013914491174365606,
+      "loss": 1.8115,
+      "step": 3764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000138917473945689,
+      "loss": 1.9082,
+      "step": 3765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013869019218635642,
+      "loss": 1.9219,
+      "step": 3766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013846306656387604,
+      "loss": 1.9766,
+      "step": 3767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001382360971763982,
+      "loss": 1.9062,
+      "step": 3768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013800928412200546,
+      "loss": 1.8721,
+      "step": 3769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013778262749871334,
+      "loss": 2.0176,
+      "step": 3770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013755612740446965,
+      "loss": 1.9453,
+      "step": 3771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013732978393715407,
+      "loss": 1.918,
+      "step": 3772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013710359719457926,
+      "loss": 1.7227,
+      "step": 3773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013687756727448968,
+      "loss": 1.9131,
+      "step": 3774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013665169427456203,
+      "loss": 1.8555,
+      "step": 3775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013642597829240544,
+      "loss": 1.8633,
+      "step": 3776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013620041942556126,
+      "loss": 1.9824,
+      "step": 3777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013597501777150252,
+      "loss": 1.9609,
+      "step": 3778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001357497734276348,
+      "loss": 1.8516,
+      "step": 3779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001355246864912953,
+      "loss": 1.8857,
+      "step": 3780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013529975705975329,
+      "loss": 1.876,
+      "step": 3781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013507498523021022,
+      "loss": 1.8291,
+      "step": 3782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013485037109979908,
+      "loss": 1.8945,
+      "step": 3783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013462591476558512,
+      "loss": 1.9043,
+      "step": 3784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013440161632456482,
+      "loss": 1.8809,
+      "step": 3785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013417747587366719,
+      "loss": 1.8682,
+      "step": 3786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001339534935097524,
+      "loss": 1.7871,
+      "step": 3787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013372966932961218,
+      "loss": 1.8398,
+      "step": 3788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013350600342997061,
+      "loss": 1.9209,
+      "step": 3789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013328249590748253,
+      "loss": 1.8799,
+      "step": 3790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013305914685873499,
+      "loss": 1.8975,
+      "step": 3791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013283595638024636,
+      "loss": 1.9404,
+      "step": 3792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013261292456846646,
+      "loss": 1.915,
+      "step": 3793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013239005151977618,
+      "loss": 1.7666,
+      "step": 3794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013216733733048859,
+      "loss": 1.9648,
+      "step": 3795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013194478209684746,
+      "loss": 1.876,
+      "step": 3796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013172238591502793,
+      "loss": 1.9141,
+      "step": 3797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001315001488811367,
+      "loss": 1.9404,
+      "step": 3798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013127807109121164,
+      "loss": 1.7627,
+      "step": 3799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013105615264122155,
+      "loss": 1.8721,
+      "step": 3800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013083439362706673,
+      "loss": 1.9004,
+      "step": 3801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013061279414457823,
+      "loss": 1.916,
+      "step": 3802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001303913542895182,
+      "loss": 1.751,
+      "step": 3803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001301700741575801,
+      "loss": 1.8682,
+      "step": 3804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012994895384438825,
+      "loss": 1.9785,
+      "step": 3805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001297279934454978,
+      "loss": 1.9795,
+      "step": 3806
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012950719305639463,
+      "loss": 1.877,
+      "step": 3807
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012928655277249596,
+      "loss": 1.8438,
+      "step": 3808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012906607268914949,
+      "loss": 1.8506,
+      "step": 3809
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012884575290163353,
+      "loss": 1.9336,
+      "step": 3810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012862559350515745,
+      "loss": 1.8916,
+      "step": 3811
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012840559459486144,
+      "loss": 1.7725,
+      "step": 3812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012818575626581574,
+      "loss": 1.8467,
+      "step": 3813
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012796607861302183,
+      "loss": 1.8594,
+      "step": 3814
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012774656173141142,
+      "loss": 1.7812,
+      "step": 3815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012752720571584648,
+      "loss": 1.8877,
+      "step": 3816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012730801066112026,
+      "loss": 1.9336,
+      "step": 3817
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012708897666195552,
+      "loss": 1.8291,
+      "step": 3818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012687010381300624,
+      "loss": 1.8662,
+      "step": 3819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012665139220885614,
+      "loss": 1.8887,
+      "step": 3820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001264328419440197,
+      "loss": 1.8125,
+      "step": 3821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001262144531129415,
+      "loss": 1.9746,
+      "step": 3822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012599622580999605,
+      "loss": 1.8721,
+      "step": 3823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012577816012948872,
+      "loss": 1.918,
+      "step": 3824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012556025616565443,
+      "loss": 1.8066,
+      "step": 3825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001253425140126585,
+      "loss": 1.8379,
+      "step": 3826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001251249337645965,
+      "loss": 1.9141,
+      "step": 3827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012490751551549368,
+      "loss": 1.9854,
+      "step": 3828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012469025935930522,
+      "loss": 1.9209,
+      "step": 3829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012447316538991682,
+      "loss": 1.7939,
+      "step": 3830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012425623370114353,
+      "loss": 1.9141,
+      "step": 3831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001240394643867303,
+      "loss": 1.8203,
+      "step": 3832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012382285754035228,
+      "loss": 1.8232,
+      "step": 3833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012360641325561434,
+      "loss": 1.8711,
+      "step": 3834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001233901316260508,
+      "loss": 1.9238,
+      "step": 3835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001231740127451258,
+      "loss": 1.8574,
+      "step": 3836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001229580567062334,
+      "loss": 1.8486,
+      "step": 3837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012274226360269687,
+      "loss": 1.9727,
+      "step": 3838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001225266335277694,
+      "loss": 1.873,
+      "step": 3839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001223111665746338,
+      "loss": 1.8428,
+      "step": 3840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012209586283640206,
+      "loss": 1.8643,
+      "step": 3841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001218807224061157,
+      "loss": 1.8848,
+      "step": 3842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012166574537674602,
+      "loss": 1.8662,
+      "step": 3843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012145093184119326,
+      "loss": 1.8604,
+      "step": 3844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012123628189228713,
+      "loss": 1.8711,
+      "step": 3845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012102179562278682,
+      "loss": 1.8125,
+      "step": 3846
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012080747312538082,
+      "loss": 1.8682,
+      "step": 3847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012059331449268645,
+      "loss": 1.8457,
+      "step": 3848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012037931981725076,
+      "loss": 1.8545,
+      "step": 3849
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012016548919154946,
+      "loss": 1.877,
+      "step": 3850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011995182270798766,
+      "loss": 1.792,
+      "step": 3851
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011973832045889927,
+      "loss": 1.9238,
+      "step": 3852
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011952498253654753,
+      "loss": 1.9111,
+      "step": 3853
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011931180903312467,
+      "loss": 1.9414,
+      "step": 3854
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011909880004075152,
+      "loss": 1.9238,
+      "step": 3855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011888595565147836,
+      "loss": 1.873,
+      "step": 3856
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011867327595728372,
+      "loss": 1.8037,
+      "step": 3857
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011846076105007526,
+      "loss": 1.7227,
+      "step": 3858
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011824841102168965,
+      "loss": 1.9561,
+      "step": 3859
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011803622596389185,
+      "loss": 1.8506,
+      "step": 3860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011782420596837585,
+      "loss": 1.8545,
+      "step": 3861
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011761235112676433,
+      "loss": 1.918,
+      "step": 3862
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011740066153060847,
+      "loss": 1.8252,
+      "step": 3863
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001171891372713878,
+      "loss": 1.8936,
+      "step": 3864
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011697777844051105,
+      "loss": 1.8086,
+      "step": 3865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011676658512931477,
+      "loss": 1.7822,
+      "step": 3866
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011655555742906437,
+      "loss": 1.8945,
+      "step": 3867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011634469543095383,
+      "loss": 1.8047,
+      "step": 3868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011613399922610512,
+      "loss": 1.8691,
+      "step": 3869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011592346890556876,
+      "loss": 1.7744,
+      "step": 3870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011571310456032341,
+      "loss": 2.0449,
+      "step": 3871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011550290628127653,
+      "loss": 1.9395,
+      "step": 3872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011529287415926309,
+      "loss": 1.8545,
+      "step": 3873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011508300828504681,
+      "loss": 1.8486,
+      "step": 3874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011487330874931951,
+      "loss": 1.9023,
+      "step": 3875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011466377564270086,
+      "loss": 2.042,
+      "step": 3876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011445440905573857,
+      "loss": 1.8652,
+      "step": 3877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011424520907890889,
+      "loss": 1.7891,
+      "step": 3878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001140361758026156,
+      "loss": 1.8076,
+      "step": 3879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011382730931719042,
+      "loss": 1.8418,
+      "step": 3880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001136186097128934,
+      "loss": 1.8379,
+      "step": 3881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011341007707991225,
+      "loss": 1.8301,
+      "step": 3882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011320171150836228,
+      "loss": 1.874,
+      "step": 3883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011299351308828709,
+      "loss": 1.7539,
+      "step": 3884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011278548190965777,
+      "loss": 1.8848,
+      "step": 3885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011257761806237299,
+      "loss": 1.9297,
+      "step": 3886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011236992163625932,
+      "loss": 1.8662,
+      "step": 3887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011216239272107093,
+      "loss": 1.8281,
+      "step": 3888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011195503140648983,
+      "loss": 1.8604,
+      "step": 3889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001117478377821251,
+      "loss": 1.96,
+      "step": 3890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011154081193751387,
+      "loss": 1.8467,
+      "step": 3891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011133395396212048,
+      "loss": 1.7969,
+      "step": 3892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011112726394533662,
+      "loss": 1.9297,
+      "step": 3893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011092074197648178,
+      "loss": 1.9277,
+      "step": 3894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011071438814480239,
+      "loss": 1.9551,
+      "step": 3895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011050820253947259,
+      "loss": 1.8574,
+      "step": 3896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011030218524959373,
+      "loss": 1.917,
+      "step": 3897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011009633636419431,
+      "loss": 1.9639,
+      "step": 3898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010989065597222996,
+      "loss": 1.8936,
+      "step": 3899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001096851441625839,
+      "loss": 1.8594,
+      "step": 3900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010947980102406597,
+      "loss": 1.9004,
+      "step": 3901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010927462664541377,
+      "loss": 1.9502,
+      "step": 3902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010906962111529118,
+      "loss": 1.9209,
+      "step": 3903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010886478452228987,
+      "loss": 1.8438,
+      "step": 3904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010866011695492812,
+      "loss": 1.7139,
+      "step": 3905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010845561850165093,
+      "loss": 1.9111,
+      "step": 3906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010825128925083089,
+      "loss": 1.8584,
+      "step": 3907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010804712929076676,
+      "loss": 1.8184,
+      "step": 3908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010784313870968465,
+      "loss": 1.833,
+      "step": 3909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010763931759573737,
+      "loss": 1.9111,
+      "step": 3910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010743566603700433,
+      "loss": 1.9727,
+      "step": 3911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001072321841214916,
+      "loss": 1.8154,
+      "step": 3912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001070288719371324,
+      "loss": 1.8721,
+      "step": 3913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010682572957178616,
+      "loss": 1.8438,
+      "step": 3914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010662275711323888,
+      "loss": 1.876,
+      "step": 3915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010641995464920345,
+      "loss": 1.8232,
+      "step": 3916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010621732226731939,
+      "loss": 1.8984,
+      "step": 3917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010601486005515216,
+      "loss": 1.7236,
+      "step": 3918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010581256810019435,
+      "loss": 1.7373,
+      "step": 3919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010561044648986445,
+      "loss": 1.9219,
+      "step": 3920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010540849531150754,
+      "loss": 1.9453,
+      "step": 3921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010520671465239496,
+      "loss": 1.876,
+      "step": 3922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010500510459972451,
+      "loss": 1.9541,
+      "step": 3923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010480366524062041,
+      "loss": 1.8506,
+      "step": 3924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010460239666213261,
+      "loss": 1.8711,
+      "step": 3925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001044012989512378,
+      "loss": 1.8877,
+      "step": 3926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010420037219483841,
+      "loss": 1.9512,
+      "step": 3927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010399961647976313,
+      "loss": 1.7539,
+      "step": 3928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010379903189276684,
+      "loss": 1.8545,
+      "step": 3929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001035986185205306,
+      "loss": 1.8301,
+      "step": 3930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000103398376449661,
+      "loss": 1.8291,
+      "step": 3931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001031983057666911,
+      "loss": 1.8955,
+      "step": 3932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010299840655807957,
+      "loss": 1.7949,
+      "step": 3933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000102798678910211,
+      "loss": 1.9316,
+      "step": 3934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010259912290939616,
+      "loss": 1.918,
+      "step": 3935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010239973864187118,
+      "loss": 1.8428,
+      "step": 3936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010220052619379844,
+      "loss": 1.8574,
+      "step": 3937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010200148565126566,
+      "loss": 1.8076,
+      "step": 3938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010180261710028665,
+      "loss": 1.9551,
+      "step": 3939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010160392062680058,
+      "loss": 1.8838,
+      "step": 3940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010140539631667228,
+      "loss": 1.7568,
+      "step": 3941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010120704425569249,
+      "loss": 1.7656,
+      "step": 3942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010100886452957714,
+      "loss": 1.9424,
+      "step": 3943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010081085722396788,
+      "loss": 1.9512,
+      "step": 3944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010061302242443204,
+      "loss": 1.7568,
+      "step": 3945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010041536021646203,
+      "loss": 1.835,
+      "step": 3946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010021787068547567,
+      "loss": 1.8711,
+      "step": 3947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010002055391681658,
+      "loss": 1.9043,
+      "step": 3948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.982340999575334e-05,
+      "loss": 1.9629,
+      "step": 3949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.962643900747992e-05,
+      "loss": 1.9023,
+      "step": 3950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.94296410371156e-05,
+      "loss": 1.9395,
+      "step": 3951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.923301616970509e-05,
+      "loss": 1.8887,
+      "step": 3952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.903656449021797e-05,
+      "loss": 1.9482,
+      "step": 3953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.884028608354895e-05,
+      "loss": 1.8242,
+      "step": 3954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.864418103451828e-05,
+      "loss": 1.8271,
+      "step": 3955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.844824942787072e-05,
+      "loss": 1.8711,
+      "step": 3956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.825249134827663e-05,
+      "loss": 1.8936,
+      "step": 3957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.805690688033114e-05,
+      "loss": 1.8154,
+      "step": 3958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.786149610855439e-05,
+      "loss": 1.7695,
+      "step": 3959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.766625911739113e-05,
+      "loss": 1.8926,
+      "step": 3960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.747119599121163e-05,
+      "loss": 1.8779,
+      "step": 3961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.72763068143106e-05,
+      "loss": 1.8838,
+      "step": 3962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.708159167090753e-05,
+      "loss": 1.9326,
+      "step": 3963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.688705064514702e-05,
+      "loss": 1.8418,
+      "step": 3964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.66926838210983e-05,
+      "loss": 1.9795,
+      "step": 3965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.649849128275512e-05,
+      "loss": 1.9844,
+      "step": 3966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.630447311403623e-05,
+      "loss": 1.8594,
+      "step": 3967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.61106293987849e-05,
+      "loss": 1.834,
+      "step": 3968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.591696022076868e-05,
+      "loss": 1.8066,
+      "step": 3969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.57234656636804e-05,
+      "loss": 1.8965,
+      "step": 3970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.553014581113667e-05,
+      "loss": 1.8838,
+      "step": 3971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.533700074667927e-05,
+      "loss": 1.9121,
+      "step": 3972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.514403055377385e-05,
+      "loss": 1.8037,
+      "step": 3973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.495123531581113e-05,
+      "loss": 1.9482,
+      "step": 3974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.475861511610562e-05,
+      "loss": 1.8545,
+      "step": 3975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.456617003789642e-05,
+      "loss": 1.8105,
+      "step": 3976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.43739001643471e-05,
+      "loss": 1.9395,
+      "step": 3977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.418180557854528e-05,
+      "loss": 1.8555,
+      "step": 3978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.398988636350303e-05,
+      "loss": 1.8672,
+      "step": 3979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.379814260215663e-05,
+      "loss": 1.9043,
+      "step": 3980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.360657437736636e-05,
+      "loss": 1.9199,
+      "step": 3981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.34151817719166e-05,
+      "loss": 1.8154,
+      "step": 3982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.322396486851626e-05,
+      "loss": 1.9678,
+      "step": 3983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.303292374979794e-05,
+      "loss": 1.9141,
+      "step": 3984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.284205849831817e-05,
+      "loss": 1.8789,
+      "step": 3985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.265136919655787e-05,
+      "loss": 1.8936,
+      "step": 3986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.246085592692183e-05,
+      "loss": 1.9062,
+      "step": 3987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.227051877173859e-05,
+      "loss": 1.8818,
+      "step": 3988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.208035781326057e-05,
+      "loss": 1.9062,
+      "step": 3989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.189037313366439e-05,
+      "loss": 1.8438,
+      "step": 3990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.170056481504996e-05,
+      "loss": 1.9062,
+      "step": 3991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.151093293944141e-05,
+      "loss": 1.7617,
+      "step": 3992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.132147758878668e-05,
+      "loss": 1.792,
+      "step": 3993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.113219884495699e-05,
+      "loss": 1.8252,
+      "step": 3994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.09430967897475e-05,
+      "loss": 1.9346,
+      "step": 3995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.075417150487713e-05,
+      "loss": 1.8408,
+      "step": 3996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.056542307198823e-05,
+      "loss": 1.874,
+      "step": 3997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.037685157264658e-05,
+      "loss": 1.9814,
+      "step": 3998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.018845708834189e-05,
+      "loss": 1.8232,
+      "step": 3999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.000023970048726e-05,
+      "loss": 1.8701,
+      "step": 4000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.981219949041891e-05,
+      "loss": 1.7861,
+      "step": 4001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.962433653939705e-05,
+      "loss": 1.8701,
+      "step": 4002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.943665092860487e-05,
+      "loss": 1.8574,
+      "step": 4003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.924914273914903e-05,
+      "loss": 1.8633,
+      "step": 4004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.906181205205943e-05,
+      "loss": 1.9053,
+      "step": 4005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.88746589482895e-05,
+      "loss": 1.9092,
+      "step": 4006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.868768350871592e-05,
+      "loss": 1.7559,
+      "step": 4007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.850088581413829e-05,
+      "loss": 1.959,
+      "step": 4008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.831426594527975e-05,
+      "loss": 1.8926,
+      "step": 4009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.81278239827864e-05,
+      "loss": 1.8594,
+      "step": 4010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.794156000722731e-05,
+      "loss": 1.9141,
+      "step": 4011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.775547409909501e-05,
+      "loss": 1.8037,
+      "step": 4012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.756956633880475e-05,
+      "loss": 1.9043,
+      "step": 4013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.738383680669504e-05,
+      "loss": 1.9648,
+      "step": 4014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.719828558302733e-05,
+      "loss": 1.8701,
+      "step": 4015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.701291274798584e-05,
+      "loss": 1.9277,
+      "step": 4016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.682771838167774e-05,
+      "loss": 1.9893,
+      "step": 4017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.664270256413331e-05,
+      "loss": 1.8428,
+      "step": 4018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.645786537530526e-05,
+      "loss": 1.8477,
+      "step": 4019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.627320689506956e-05,
+      "loss": 1.8633,
+      "step": 4020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.608872720322481e-05,
+      "loss": 1.8008,
+      "step": 4021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.59044263794922e-05,
+      "loss": 1.9248,
+      "step": 4022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.572030450351565e-05,
+      "loss": 1.8096,
+      "step": 4023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.553636165486173e-05,
+      "loss": 1.8828,
+      "step": 4024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.535259791301996e-05,
+      "loss": 1.9131,
+      "step": 4025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.516901335740196e-05,
+      "loss": 1.915,
+      "step": 4026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.498560806734229e-05,
+      "loss": 1.8135,
+      "step": 4027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.480238212209812e-05,
+      "loss": 1.7637,
+      "step": 4028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.461933560084878e-05,
+      "loss": 1.7793,
+      "step": 4029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.443646858269604e-05,
+      "loss": 1.9092,
+      "step": 4030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.425378114666455e-05,
+      "loss": 1.7979,
+      "step": 4031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.407127337170096e-05,
+      "loss": 1.8691,
+      "step": 4032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.388894533667429e-05,
+      "loss": 1.9287,
+      "step": 4033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.37067971203761e-05,
+      "loss": 1.9863,
+      "step": 4034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.352482880152023e-05,
+      "loss": 1.9609,
+      "step": 4035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.334304045874247e-05,
+      "loss": 1.9248,
+      "step": 4036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.316143217060129e-05,
+      "loss": 1.9561,
+      "step": 4037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.298000401557704e-05,
+      "loss": 1.748,
+      "step": 4038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.279875607207226e-05,
+      "loss": 1.8818,
+      "step": 4039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.26176884184115e-05,
+      "loss": 1.8584,
+      "step": 4040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.243680113284169e-05,
+      "loss": 1.8174,
+      "step": 4041
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.225609429353187e-05,
+      "loss": 1.9326,
+      "step": 4042
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.207556797857252e-05,
+      "loss": 1.8936,
+      "step": 4043
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.189522226597689e-05,
+      "loss": 1.9111,
+      "step": 4044
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.171505723367961e-05,
+      "loss": 1.8311,
+      "step": 4045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.153507295953727e-05,
+      "loss": 1.8555,
+      "step": 4046
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.135526952132877e-05,
+      "loss": 1.9023,
+      "step": 4047
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.117564699675434e-05,
+      "loss": 1.9326,
+      "step": 4048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.099620546343634e-05,
+      "loss": 1.9004,
+      "step": 4049
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.081694499891901e-05,
+      "loss": 1.8887,
+      "step": 4050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.063786568066805e-05,
+      "loss": 1.8945,
+      "step": 4051
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.045896758607091e-05,
+      "loss": 1.8896,
+      "step": 4052
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.028025079243705e-05,
+      "loss": 1.874,
+      "step": 4053
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.010171537699713e-05,
+      "loss": 1.8564,
+      "step": 4054
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.992336141690393e-05,
+      "loss": 1.874,
+      "step": 4055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.974518898923117e-05,
+      "loss": 1.9141,
+      "step": 4056
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.95671981709749e-05,
+      "loss": 1.9346,
+      "step": 4057
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.938938903905207e-05,
+      "loss": 1.8848,
+      "step": 4058
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.921176167030125e-05,
+      "loss": 1.874,
+      "step": 4059
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.903431614148276e-05,
+      "loss": 1.8008,
+      "step": 4060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.885705252927788e-05,
+      "loss": 1.8457,
+      "step": 4061
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.867997091028966e-05,
+      "loss": 1.8877,
+      "step": 4062
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.850307136104246e-05,
+      "loss": 1.8193,
+      "step": 4063
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.832635395798176e-05,
+      "loss": 1.8809,
+      "step": 4064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.814981877747434e-05,
+      "loss": 1.835,
+      "step": 4065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.79734658958085e-05,
+      "loss": 1.8135,
+      "step": 4066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.779729538919351e-05,
+      "loss": 1.8438,
+      "step": 4067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.762130733375971e-05,
+      "loss": 1.9443,
+      "step": 4068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.744550180555903e-05,
+      "loss": 1.8896,
+      "step": 4069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.726987888056431e-05,
+      "loss": 1.8906,
+      "step": 4070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.709443863466925e-05,
+      "loss": 1.8291,
+      "step": 4071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.691918114368906e-05,
+      "loss": 1.9512,
+      "step": 4072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.674410648335956e-05,
+      "loss": 1.8379,
+      "step": 4073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.65692147293377e-05,
+      "loss": 1.8818,
+      "step": 4074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.639450595720137e-05,
+      "loss": 1.9053,
+      "step": 4075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.621998024244948e-05,
+      "loss": 1.8799,
+      "step": 4076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.604563766050199e-05,
+      "loss": 1.9473,
+      "step": 4077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.587147828669915e-05,
+      "loss": 1.9619,
+      "step": 4078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.569750219630272e-05,
+      "loss": 1.877,
+      "step": 4079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.552370946449483e-05,
+      "loss": 1.7861,
+      "step": 4080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.535010016637827e-05,
+      "loss": 1.8369,
+      "step": 4081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.517667437697695e-05,
+      "loss": 1.8711,
+      "step": 4082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.500343217123546e-05,
+      "loss": 1.915,
+      "step": 4083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.483037362401856e-05,
+      "loss": 1.8945,
+      "step": 4084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.46574988101123e-05,
+      "loss": 1.8018,
+      "step": 4085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.448480780422285e-05,
+      "loss": 1.8047,
+      "step": 4086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4312300680977e-05,
+      "loss": 1.8682,
+      "step": 4087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.41399775149224e-05,
+      "loss": 1.875,
+      "step": 4088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.396783838052679e-05,
+      "loss": 1.9199,
+      "step": 4089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.379588335217874e-05,
+      "loss": 1.8008,
+      "step": 4090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3624112504187e-05,
+      "loss": 1.8408,
+      "step": 4091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.345252591078088e-05,
+      "loss": 1.8408,
+      "step": 4092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.328112364611012e-05,
+      "loss": 1.834,
+      "step": 4093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.310990578424437e-05,
+      "loss": 1.8223,
+      "step": 4094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.293887239917429e-05,
+      "loss": 1.8359,
+      "step": 4095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.276802356481016e-05,
+      "loss": 1.8643,
+      "step": 4096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.259735935498295e-05,
+      "loss": 1.8447,
+      "step": 4097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.242687984344382e-05,
+      "loss": 1.8838,
+      "step": 4098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.225658510386385e-05,
+      "loss": 1.8438,
+      "step": 4099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.208647520983425e-05,
+      "loss": 1.8477,
+      "step": 4100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.191655023486682e-05,
+      "loss": 1.8389,
+      "step": 4101
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.174681025239299e-05,
+      "loss": 1.8623,
+      "step": 4102
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.157725533576426e-05,
+      "loss": 1.8174,
+      "step": 4103
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.14078855582525e-05,
+      "loss": 1.9238,
+      "step": 4104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.123870099304941e-05,
+      "loss": 1.833,
+      "step": 4105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.106970171326649e-05,
+      "loss": 1.9785,
+      "step": 4106
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.090088779193516e-05,
+      "loss": 1.7646,
+      "step": 4107
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.073225930200722e-05,
+      "loss": 1.8799,
+      "step": 4108
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.05638163163536e-05,
+      "loss": 1.9141,
+      "step": 4109
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.039555890776578e-05,
+      "loss": 1.9014,
+      "step": 4110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.022748714895444e-05,
+      "loss": 1.8662,
+      "step": 4111
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.00596011125506e-05,
+      "loss": 1.8525,
+      "step": 4112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.989190087110442e-05,
+      "loss": 1.8701,
+      "step": 4113
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.97243864970864e-05,
+      "loss": 1.8867,
+      "step": 4114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.955705806288626e-05,
+      "loss": 1.8135,
+      "step": 4115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.938991564081332e-05,
+      "loss": 1.9307,
+      "step": 4116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.922295930309692e-05,
+      "loss": 1.8018,
+      "step": 4117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.905618912188582e-05,
+      "loss": 1.8262,
+      "step": 4118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.888960516924808e-05,
+      "loss": 1.8467,
+      "step": 4119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.872320751717175e-05,
+      "loss": 1.749,
+      "step": 4120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.855699623756395e-05,
+      "loss": 1.8877,
+      "step": 4121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.839097140225136e-05,
+      "loss": 1.8232,
+      "step": 4122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.822513308298028e-05,
+      "loss": 1.7822,
+      "step": 4123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.805948135141616e-05,
+      "loss": 1.7988,
+      "step": 4124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.78940162791441e-05,
+      "loss": 1.8271,
+      "step": 4125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.772873793766815e-05,
+      "loss": 1.9014,
+      "step": 4126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.756364639841218e-05,
+      "loss": 1.834,
+      "step": 4127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.739874173271882e-05,
+      "loss": 1.8965,
+      "step": 4128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.723402401185014e-05,
+      "loss": 1.917,
+      "step": 4129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.70694933069877e-05,
+      "loss": 1.8809,
+      "step": 4130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.690514968923167e-05,
+      "loss": 1.8945,
+      "step": 4131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.674099322960186e-05,
+      "loss": 1.7842,
+      "step": 4132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.65770239990372e-05,
+      "loss": 1.8174,
+      "step": 4133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.641324206839539e-05,
+      "loss": 1.9375,
+      "step": 4134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.624964750845325e-05,
+      "loss": 1.8594,
+      "step": 4135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.608624038990696e-05,
+      "loss": 1.9766,
+      "step": 4136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.59230207833713e-05,
+      "loss": 1.8135,
+      "step": 4137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.575998875938016e-05,
+      "loss": 1.8623,
+      "step": 4138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.559714438838644e-05,
+      "loss": 1.8613,
+      "step": 4139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5434487740762e-05,
+      "loss": 1.874,
+      "step": 4140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.527201888679746e-05,
+      "loss": 1.916,
+      "step": 4141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.51097378967021e-05,
+      "loss": 1.8604,
+      "step": 4142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.494764484060444e-05,
+      "loss": 1.8682,
+      "step": 4143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.478573978855146e-05,
+      "loss": 1.8896,
+      "step": 4144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4624022810509e-05,
+      "loss": 1.791,
+      "step": 4145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.446249397636172e-05,
+      "loss": 1.8818,
+      "step": 4146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.430115335591291e-05,
+      "loss": 1.8145,
+      "step": 4147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.414000101888428e-05,
+      "loss": 1.8506,
+      "step": 4148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.397903703491664e-05,
+      "loss": 1.8486,
+      "step": 4149
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.381826147356907e-05,
+      "loss": 1.8438,
+      "step": 4150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.365767440431914e-05,
+      "loss": 1.877,
+      "step": 4151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.349727589656329e-05,
+      "loss": 1.8359,
+      "step": 4152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.333706601961642e-05,
+      "loss": 1.8066,
+      "step": 4153
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.317704484271164e-05,
+      "loss": 1.873,
+      "step": 4154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.301721243500092e-05,
+      "loss": 1.8643,
+      "step": 4155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.285756886555422e-05,
+      "loss": 1.8594,
+      "step": 4156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.269811420336019e-05,
+      "loss": 1.7295,
+      "step": 4157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.25388485173256e-05,
+      "loss": 1.8164,
+      "step": 4158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.237977187627586e-05,
+      "loss": 1.917,
+      "step": 4159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.222088434895462e-05,
+      "loss": 1.8057,
+      "step": 4160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.206218600402352e-05,
+      "loss": 1.8994,
+      "step": 4161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.190367691006288e-05,
+      "loss": 1.8018,
+      "step": 4162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.174535713557089e-05,
+      "loss": 1.9492,
+      "step": 4163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1587226748964e-05,
+      "loss": 1.9307,
+      "step": 4164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.142928581857693e-05,
+      "loss": 1.7812,
+      "step": 4165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.127153441266236e-05,
+      "loss": 1.9248,
+      "step": 4166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.111397259939128e-05,
+      "loss": 1.7568,
+      "step": 4167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.095660044685275e-05,
+      "loss": 1.9043,
+      "step": 4168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.079941802305361e-05,
+      "loss": 1.7744,
+      "step": 4169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.064242539591874e-05,
+      "loss": 1.8662,
+      "step": 4170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.048562263329138e-05,
+      "loss": 1.9658,
+      "step": 4171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.032900980293221e-05,
+      "loss": 1.9092,
+      "step": 4172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.017258697252037e-05,
+      "loss": 1.7451,
+      "step": 4173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.001635420965235e-05,
+      "loss": 1.7432,
+      "step": 4174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.986031158184296e-05,
+      "loss": 1.9326,
+      "step": 4175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.970445915652456e-05,
+      "loss": 1.7871,
+      "step": 4176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.954879700104732e-05,
+      "loss": 1.9434,
+      "step": 4177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.939332518267943e-05,
+      "loss": 1.918,
+      "step": 4178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9238043768606474e-05,
+      "loss": 1.9062,
+      "step": 4179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.908295282593207e-05,
+      "loss": 1.7803,
+      "step": 4180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.892805242167748e-05,
+      "loss": 1.876,
+      "step": 4181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.87733426227815e-05,
+      "loss": 1.7881,
+      "step": 4182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8618823496100426e-05,
+      "loss": 1.8643,
+      "step": 4183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8464495108408625e-05,
+      "loss": 1.9014,
+      "step": 4184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.831035752639763e-05,
+      "loss": 1.9482,
+      "step": 4185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.81564108166765e-05,
+      "loss": 1.8564,
+      "step": 4186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.800265504577201e-05,
+      "loss": 1.8428,
+      "step": 4187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.784909028012858e-05,
+      "loss": 1.9805,
+      "step": 4188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.769571658610761e-05,
+      "loss": 1.8047,
+      "step": 4189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7542534029988436e-05,
+      "loss": 1.8877,
+      "step": 4190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7389542677967365e-05,
+      "loss": 1.8184,
+      "step": 4191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.72367425961583e-05,
+      "loss": 1.9238,
+      "step": 4192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.708413385059241e-05,
+      "loss": 1.9844,
+      "step": 4193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6931716507218236e-05,
+      "loss": 1.8086,
+      "step": 4194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.677949063190169e-05,
+      "loss": 1.7568,
+      "step": 4195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.66274562904257e-05,
+      "loss": 1.9102,
+      "step": 4196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.647561354849079e-05,
+      "loss": 1.8711,
+      "step": 4197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.632396247171428e-05,
+      "loss": 1.8887,
+      "step": 4198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.617250312563082e-05,
+      "loss": 1.8418,
+      "step": 4199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.602123557569239e-05,
+      "loss": 1.9268,
+      "step": 4200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.587015988726774e-05,
+      "loss": 1.7549,
+      "step": 4201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.571927612564298e-05,
+      "loss": 1.9414,
+      "step": 4202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.55685843560213e-05,
+      "loss": 1.8135,
+      "step": 4203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.541808464352277e-05,
+      "loss": 1.9072,
+      "step": 4204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.526777705318442e-05,
+      "loss": 1.9746,
+      "step": 4205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.511766164996046e-05,
+      "loss": 1.96,
+      "step": 4206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.496773849872183e-05,
+      "loss": 1.8213,
+      "step": 4207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4818007664256656e-05,
+      "loss": 1.8564,
+      "step": 4208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.466846921126961e-05,
+      "loss": 1.8516,
+      "step": 4209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.451912320438263e-05,
+      "loss": 1.8545,
+      "step": 4210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.436996970813418e-05,
+      "loss": 1.8076,
+      "step": 4211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.42210087869795e-05,
+      "loss": 1.8145,
+      "step": 4212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.407224050529097e-05,
+      "loss": 1.9375,
+      "step": 4213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.392366492735723e-05,
+      "loss": 1.8428,
+      "step": 4214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3775282117384086e-05,
+      "loss": 1.9766,
+      "step": 4215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.362709213949396e-05,
+      "loss": 1.749,
+      "step": 4216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.347909505772569e-05,
+      "loss": 1.8721,
+      "step": 4217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3331290936034895e-05,
+      "loss": 1.9082,
+      "step": 4218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.318367983829392e-05,
+      "loss": 1.8086,
+      "step": 4219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.303626182829158e-05,
+      "loss": 1.8711,
+      "step": 4220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.288903696973313e-05,
+      "loss": 1.9238,
+      "step": 4221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.274200532624057e-05,
+      "loss": 1.873,
+      "step": 4222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2595166961352527e-05,
+      "loss": 1.7959,
+      "step": 4223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.244852193852373e-05,
+      "loss": 1.917,
+      "step": 4224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.23020703211255e-05,
+      "loss": 1.915,
+      "step": 4225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2155812172445695e-05,
+      "loss": 1.8955,
+      "step": 4226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.200974755568849e-05,
+      "loss": 1.8398,
+      "step": 4227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.186387653397434e-05,
+      "loss": 1.8896,
+      "step": 4228
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1718199170340094e-05,
+      "loss": 1.8232,
+      "step": 4229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.157271552773918e-05,
+      "loss": 1.9385,
+      "step": 4230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1427425669040795e-05,
+      "loss": 1.9404,
+      "step": 4231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.128232965703095e-05,
+      "loss": 1.8652,
+      "step": 4232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1137427554411365e-05,
+      "loss": 1.75,
+      "step": 4233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.099271942380024e-05,
+      "loss": 1.8789,
+      "step": 4234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.084820532773199e-05,
+      "loss": 1.7793,
+      "step": 4235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.070388532865716e-05,
+      "loss": 1.8652,
+      "step": 4236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.055975948894226e-05,
+      "loss": 1.8604,
+      "step": 4237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.041582787087007e-05,
+      "loss": 1.8389,
+      "step": 4238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.027209053663939e-05,
+      "loss": 1.9043,
+      "step": 4239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0128547548364876e-05,
+      "loss": 1.6826,
+      "step": 4240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9985198968077626e-05,
+      "loss": 1.8555,
+      "step": 4241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.98420448577242e-05,
+      "loss": 1.9688,
+      "step": 4242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9699085279167665e-05,
+      "loss": 1.8398,
+      "step": 4243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.955632029418644e-05,
+      "loss": 1.9648,
+      "step": 4244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.941374996447545e-05,
+      "loss": 1.916,
+      "step": 4245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.927137435164503e-05,
+      "loss": 1.834,
+      "step": 4246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9129193517221504e-05,
+      "loss": 1.79,
+      "step": 4247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.898720752264729e-05,
+      "loss": 1.8369,
+      "step": 4248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.884541642928009e-05,
+      "loss": 1.9551,
+      "step": 4249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.870382029839387e-05,
+      "loss": 1.8916,
+      "step": 4250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.856241919117821e-05,
+      "loss": 1.9033,
+      "step": 4251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.842121316873821e-05,
+      "loss": 1.7773,
+      "step": 4252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8280202292094844e-05,
+      "loss": 1.9121,
+      "step": 4253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8139386622184755e-05,
+      "loss": 1.8994,
+      "step": 4254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7998766219860214e-05,
+      "loss": 1.9111,
+      "step": 4255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.785834114588894e-05,
+      "loss": 1.8525,
+      "step": 4256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7718111460954506e-05,
+      "loss": 1.9678,
+      "step": 4257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.757807722565605e-05,
+      "loss": 1.8701,
+      "step": 4258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7438238500507994e-05,
+      "loss": 1.8281,
+      "step": 4259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.729859534594033e-05,
+      "loss": 1.7881,
+      "step": 4260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7159147822298795e-05,
+      "loss": 1.8916,
+      "step": 4261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7019895989844355e-05,
+      "loss": 1.8682,
+      "step": 4262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.688083990875336e-05,
+      "loss": 1.8467,
+      "step": 4263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.674197963911775e-05,
+      "loss": 1.8574,
+      "step": 4264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.660331524094491e-05,
+      "loss": 1.8398,
+      "step": 4265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6464846774157135e-05,
+      "loss": 1.7881,
+      "step": 4266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.632657429859266e-05,
+      "loss": 1.7881,
+      "step": 4267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6188497874004535e-05,
+      "loss": 1.8789,
+      "step": 4268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6050617560061236e-05,
+      "loss": 1.8135,
+      "step": 4269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.591293341634661e-05,
+      "loss": 1.8193,
+      "step": 4270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5775445502359736e-05,
+      "loss": 1.8066,
+      "step": 4271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.563815387751463e-05,
+      "loss": 1.8594,
+      "step": 4272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5501058601140843e-05,
+      "loss": 1.8125,
+      "step": 4273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5364159732482756e-05,
+      "loss": 1.8896,
+      "step": 4274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.522745733070005e-05,
+      "loss": 1.8701,
+      "step": 4275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.509095145486736e-05,
+      "loss": 1.9014,
+      "step": 4276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.49546421639746e-05,
+      "loss": 1.8223,
+      "step": 4277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.481852951692672e-05,
+      "loss": 1.8652,
+      "step": 4278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.468261357254338e-05,
+      "loss": 1.8438,
+      "step": 4279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4546894389559725e-05,
+      "loss": 1.9209,
+      "step": 4280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.441137202662543e-05,
+      "loss": 1.8438,
+      "step": 4281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.427604654230527e-05,
+      "loss": 1.7793,
+      "step": 4282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4140917995079076e-05,
+      "loss": 1.7803,
+      "step": 4283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.400598644334136e-05,
+      "loss": 1.6885,
+      "step": 4284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3871251945401594e-05,
+      "loss": 1.8213,
+      "step": 4285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3736714559484245e-05,
+      "loss": 1.8008,
+      "step": 4286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.360237434372838e-05,
+      "loss": 1.8975,
+      "step": 4287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.346823135618788e-05,
+      "loss": 1.8008,
+      "step": 4288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.333428565483155e-05,
+      "loss": 1.8369,
+      "step": 4289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.320053729754281e-05,
+      "loss": 1.8145,
+      "step": 4290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.306698634211975e-05,
+      "loss": 1.8008,
+      "step": 4291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2933632846275326e-05,
+      "loss": 1.8594,
+      "step": 4292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.280047686763716e-05,
+      "loss": 1.7852,
+      "step": 4293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2667518463747334e-05,
+      "loss": 1.8975,
+      "step": 4294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.253475769206255e-05,
+      "loss": 1.8164,
+      "step": 4295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2402194609954425e-05,
+      "loss": 1.7861,
+      "step": 4296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2269829274708735e-05,
+      "loss": 1.9775,
+      "step": 4297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2137661743526065e-05,
+      "loss": 1.9199,
+      "step": 4298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2005692073521614e-05,
+      "loss": 1.874,
+      "step": 4299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.187392032172471e-05,
+      "loss": 1.791,
+      "step": 4300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.17423465450793e-05,
+      "loss": 1.8496,
+      "step": 4301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.161097080044407e-05,
+      "loss": 1.8232,
+      "step": 4302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.147979314459177e-05,
+      "loss": 1.8643,
+      "step": 4303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134881363420956e-05,
+      "loss": 1.834,
+      "step": 4304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.121803232589916e-05,
+      "loss": 1.7539,
+      "step": 4305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1087449276176695e-05,
+      "loss": 1.9111,
+      "step": 4306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.09570645414723e-05,
+      "loss": 1.8633,
+      "step": 4307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0826878178130744e-05,
+      "loss": 1.7334,
+      "step": 4308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.06968902424108e-05,
+      "loss": 1.8896,
+      "step": 4309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.05671007904857e-05,
+      "loss": 1.7666,
+      "step": 4310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0437509878442635e-05,
+      "loss": 1.9619,
+      "step": 4311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.030811756228331e-05,
+      "loss": 1.7979,
+      "step": 4312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0178923897923546e-05,
+      "loss": 1.9072,
+      "step": 4313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.004992894119303e-05,
+      "loss": 1.7285,
+      "step": 4314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.99211327478361e-05,
+      "loss": 1.8438,
+      "step": 4315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.979253537351068e-05,
+      "loss": 1.8213,
+      "step": 4316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9664136873788946e-05,
+      "loss": 1.8291,
+      "step": 4317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9535937304157446e-05,
+      "loss": 1.8154,
+      "step": 4318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.940793672001613e-05,
+      "loss": 1.8418,
+      "step": 4319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.928013517667961e-05,
+      "loss": 1.9004,
+      "step": 4320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9152532729376136e-05,
+      "loss": 1.9434,
+      "step": 4321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.902512943324793e-05,
+      "loss": 1.7422,
+      "step": 4322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.889792534335118e-05,
+      "loss": 1.9082,
+      "step": 4323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.877092051465614e-05,
+      "loss": 1.8076,
+      "step": 4324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.864411500204679e-05,
+      "loss": 1.7725,
+      "step": 4325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.851750886032085e-05,
+      "loss": 1.7285,
+      "step": 4326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.839110214419017e-05,
+      "loss": 1.9307,
+      "step": 4327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.826489490828039e-05,
+      "loss": 1.9131,
+      "step": 4328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8138887207130767e-05,
+      "loss": 1.9199,
+      "step": 4329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8013079095194306e-05,
+      "loss": 1.9072,
+      "step": 4330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.788747062683812e-05,
+      "loss": 1.7725,
+      "step": 4331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7762061856342567e-05,
+      "loss": 1.7334,
+      "step": 4332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.763685283790208e-05,
+      "loss": 1.8486,
+      "step": 4333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7511843625624654e-05,
+      "loss": 1.877,
+      "step": 4334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.738703427353191e-05,
+      "loss": 1.8506,
+      "step": 4335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7262424835558913e-05,
+      "loss": 1.8418,
+      "step": 4336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.713801536555483e-05,
+      "loss": 1.7939,
+      "step": 4337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7013805917281985e-05,
+      "loss": 1.8633,
+      "step": 4338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.688979654441627e-05,
+      "loss": 1.7783,
+      "step": 4339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6765987300547366e-05,
+      "loss": 1.7686,
+      "step": 4340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.664237823917843e-05,
+      "loss": 1.8564,
+      "step": 4341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.65189694137259e-05,
+      "loss": 1.7852,
+      "step": 4342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.639576087751989e-05,
+      "loss": 1.8623,
+      "step": 4343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.627275268380392e-05,
+      "loss": 1.8018,
+      "step": 4344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6149944885734755e-05,
+      "loss": 1.9365,
+      "step": 4345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.602733753638277e-05,
+      "loss": 1.877,
+      "step": 4346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.590493068873163e-05,
+      "loss": 1.8818,
+      "step": 4347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5782724395678476e-05,
+      "loss": 1.8662,
+      "step": 4348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.566071871003351e-05,
+      "loss": 1.7959,
+      "step": 4349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.553891368452061e-05,
+      "loss": 1.9053,
+      "step": 4350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5417309371776674e-05,
+      "loss": 1.8691,
+      "step": 4351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5295905824351826e-05,
+      "loss": 1.7979,
+      "step": 4352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.517470309470972e-05,
+      "loss": 1.835,
+      "step": 4353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5053701235226886e-05,
+      "loss": 1.8242,
+      "step": 4354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.493290029819324e-05,
+      "loss": 1.793,
+      "step": 4355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.481230033581201e-05,
+      "loss": 1.7334,
+      "step": 4356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.469190140019923e-05,
+      "loss": 1.8691,
+      "step": 4357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4571703543384295e-05,
+      "loss": 1.8867,
+      "step": 4358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.445170681730964e-05,
+      "loss": 1.9443,
+      "step": 4359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.433191127383078e-05,
+      "loss": 1.8799,
+      "step": 4360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4212316964716386e-05,
+      "loss": 1.8369,
+      "step": 4361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4092923941647926e-05,
+      "loss": 1.8896,
+      "step": 4362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.397373225622025e-05,
+      "loss": 1.7812,
+      "step": 4363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3854741959940926e-05,
+      "loss": 1.8457,
+      "step": 4364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.373595310423039e-05,
+      "loss": 1.8965,
+      "step": 4365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.361736574042246e-05,
+      "loss": 1.7334,
+      "step": 4366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.349897991976336e-05,
+      "loss": 1.7959,
+      "step": 4367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.338079569341268e-05,
+      "loss": 1.8535,
+      "step": 4368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3262813112442667e-05,
+      "loss": 1.8311,
+      "step": 4369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3145032227838435e-05,
+      "loss": 1.957,
+      "step": 4370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3027453090497874e-05,
+      "loss": 1.8213,
+      "step": 4371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2910075751231895e-05,
+      "loss": 1.8389,
+      "step": 4372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2792900260764035e-05,
+      "loss": 1.7861,
+      "step": 4373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.267592666973057e-05,
+      "loss": 1.8105,
+      "step": 4374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.255915502868062e-05,
+      "loss": 1.9258,
+      "step": 4375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.244258538807621e-05,
+      "loss": 2.0205,
+      "step": 4376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.232621779829176e-05,
+      "loss": 1.8838,
+      "step": 4377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2210052309614436e-05,
+      "loss": 1.8936,
+      "step": 4378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.209408897224419e-05,
+      "loss": 1.8262,
+      "step": 4379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.197832783629362e-05,
+      "loss": 1.9365,
+      "step": 4380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.186276895178774e-05,
+      "loss": 1.7373,
+      "step": 4381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1747412368664354e-05,
+      "loss": 1.8447,
+      "step": 4382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.16322581367739e-05,
+      "loss": 1.8857,
+      "step": 4383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.151730630587912e-05,
+      "loss": 1.915,
+      "step": 4384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.140255692565558e-05,
+      "loss": 1.8018,
+      "step": 4385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1288010045691095e-05,
+      "loss": 1.7686,
+      "step": 4386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1173665715486075e-05,
+      "loss": 1.7578,
+      "step": 4387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.10595239844535e-05,
+      "loss": 1.8242,
+      "step": 4388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.094558490191873e-05,
+      "loss": 1.748,
+      "step": 4389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.083184851711945e-05,
+      "loss": 1.9355,
+      "step": 4390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0718314879205954e-05,
+      "loss": 1.8174,
+      "step": 4391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.060498403724071e-05,
+      "loss": 1.8389,
+      "step": 4392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0491856040198606e-05,
+      "loss": 1.9336,
+      "step": 4393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.037893093696703e-05,
+      "loss": 1.8145,
+      "step": 4394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0266208776345395e-05,
+      "loss": 1.8555,
+      "step": 4395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0153689607045842e-05,
+      "loss": 1.8008,
+      "step": 4396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0041373477692314e-05,
+      "loss": 1.9883,
+      "step": 4397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.992926043682137e-05,
+      "loss": 1.8281,
+      "step": 4398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9817350532881648e-05,
+      "loss": 1.833,
+      "step": 4399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9705643814233907e-05,
+      "loss": 1.8506,
+      "step": 4400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.959414032915142e-05,
+      "loss": 1.8936,
+      "step": 4401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9482840125819254e-05,
+      "loss": 1.9287,
+      "step": 4402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9371743252334936e-05,
+      "loss": 1.8916,
+      "step": 4403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9260849756708007e-05,
+      "loss": 1.8564,
+      "step": 4404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9150159686860022e-05,
+      "loss": 1.8428,
+      "step": 4405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9039673090624775e-05,
+      "loss": 1.8721,
+      "step": 4406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8929390015748125e-05,
+      "loss": 1.9053,
+      "step": 4407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.881931050988784e-05,
+      "loss": 1.9111,
+      "step": 4408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8709434620613762e-05,
+      "loss": 1.9131,
+      "step": 4409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.85997623954079e-05,
+      "loss": 1.958,
+      "step": 4410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8490293881664186e-05,
+      "loss": 1.832,
+      "step": 4411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8381029126688384e-05,
+      "loss": 1.7461,
+      "step": 4412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8271968177698282e-05,
+      "loss": 1.875,
+      "step": 4413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.816311108182368e-05,
+      "loss": 1.8564,
+      "step": 4414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8054457886106167e-05,
+      "loss": 1.9717,
+      "step": 4415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7946008637499244e-05,
+      "loss": 1.8613,
+      "step": 4416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7837763382868252e-05,
+      "loss": 1.8057,
+      "step": 4417
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.772972216899061e-05,
+      "loss": 1.8721,
+      "step": 4418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7621885042555196e-05,
+      "loss": 1.917,
+      "step": 4419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7514252050163003e-05,
+      "loss": 1.877,
+      "step": 4420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.740682323832666e-05,
+      "loss": 1.707,
+      "step": 4421
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7299598653470425e-05,
+      "loss": 1.8652,
+      "step": 4422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7192578341930617e-05,
+      "loss": 1.7959,
+      "step": 4423
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7085762349955135e-05,
+      "loss": 1.7783,
+      "step": 4424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.697915072370344e-05,
+      "loss": 1.875,
+      "step": 4425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6872743509246954e-05,
+      "loss": 1.7324,
+      "step": 4426
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6766540752568514e-05,
+      "loss": 1.7871,
+      "step": 4427
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6660542499562788e-05,
+      "loss": 1.8916,
+      "step": 4428
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6554748796035865e-05,
+      "loss": 1.917,
+      "step": 4429
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6449159687705615e-05,
+      "loss": 1.8223,
+      "step": 4430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6343775220201537e-05,
+      "loss": 1.7842,
+      "step": 4431
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6238595439064485e-05,
+      "loss": 1.8721,
+      "step": 4432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6133620389747094e-05,
+      "loss": 1.8301,
+      "step": 4433
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.60288501176133e-05,
+      "loss": 1.8223,
+      "step": 4434
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5924284667938613e-05,
+      "loss": 1.792,
+      "step": 4435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5819924085910217e-05,
+      "loss": 1.915,
+      "step": 4436
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5715768416626485e-05,
+      "loss": 1.8711,
+      "step": 4437
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5611817705097406e-05,
+      "loss": 1.8066,
+      "step": 4438
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5508071996244497e-05,
+      "loss": 1.9229,
+      "step": 4439
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5404531334900448e-05,
+      "loss": 1.9453,
+      "step": 4440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.530119576580936e-05,
+      "loss": 1.8438,
+      "step": 4441
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.519806533362695e-05,
+      "loss": 1.8486,
+      "step": 4442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5095140082920022e-05,
+      "loss": 1.7246,
+      "step": 4443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4992420058166765e-05,
+      "loss": 1.9385,
+      "step": 4444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4889905303756844e-05,
+      "loss": 1.8184,
+      "step": 4445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4787595863991153e-05,
+      "loss": 1.8809,
+      "step": 4446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4685491783081714e-05,
+      "loss": 1.8555,
+      "step": 4447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4583593105151846e-05,
+      "loss": 1.8867,
+      "step": 4448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4481899874236323e-05,
+      "loss": 1.9111,
+      "step": 4449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4380412134280883e-05,
+      "loss": 1.8018,
+      "step": 4450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.427912992914255e-05,
+      "loss": 1.8525,
+      "step": 4451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4178053302589598e-05,
+      "loss": 1.8457,
+      "step": 4452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4077182298301414e-05,
+      "loss": 1.96,
+      "step": 4453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3976516959868412e-05,
+      "loss": 1.8545,
+      "step": 4454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3876057330792345e-05,
+      "loss": 1.8018,
+      "step": 4455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3775803454485935e-05,
+      "loss": 1.7871,
+      "step": 4456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3675755374272856e-05,
+      "loss": 1.8818,
+      "step": 4457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3575913133388195e-05,
+      "loss": 1.8779,
+      "step": 4458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.347627677497788e-05,
+      "loss": 1.8604,
+      "step": 4459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3376846342098744e-05,
+      "loss": 1.8135,
+      "step": 4460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3277621877718924e-05,
+      "loss": 1.9072,
+      "step": 4461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3178603424717394e-05,
+      "loss": 1.8057,
+      "step": 4462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3079791025884033e-05,
+      "loss": 1.876,
+      "step": 4463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.298118472391969e-05,
+      "loss": 1.9736,
+      "step": 4464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2882784561436277e-05,
+      "loss": 1.7578,
+      "step": 4465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2784590580956664e-05,
+      "loss": 1.8447,
+      "step": 4466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.268660282491436e-05,
+      "loss": 1.8311,
+      "step": 4467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.258882133565404e-05,
+      "loss": 1.8965,
+      "step": 4468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2491246155431076e-05,
+      "loss": 1.8916,
+      "step": 4469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2393877326411573e-05,
+      "loss": 1.9033,
+      "step": 4470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2296714890672808e-05,
+      "loss": 1.8906,
+      "step": 4471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2199758890202594e-05,
+      "loss": 1.8799,
+      "step": 4472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2103009366899575e-05,
+      "loss": 1.9609,
+      "step": 4473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2006466362573307e-05,
+      "loss": 1.8184,
+      "step": 4474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.191012991894392e-05,
+      "loss": 1.9629,
+      "step": 4475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1814000077642338e-05,
+      "loss": 1.8096,
+      "step": 4476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1718076880210326e-05,
+      "loss": 1.8379,
+      "step": 4477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1622360368100226e-05,
+      "loss": 1.8779,
+      "step": 4478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.152685058267495e-05,
+      "loss": 1.7227,
+      "step": 4479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.143154756520832e-05,
+      "loss": 1.8027,
+      "step": 4480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.133645135688478e-05,
+      "loss": 1.8008,
+      "step": 4481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1241561998799174e-05,
+      "loss": 1.7686,
+      "step": 4482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1146879531957153e-05,
+      "loss": 1.8877,
+      "step": 4483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.105240399727493e-05,
+      "loss": 1.7969,
+      "step": 4484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.095813543557923e-05,
+      "loss": 1.7812,
+      "step": 4485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0864073887607416e-05,
+      "loss": 1.8193,
+      "step": 4486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0770219394007362e-05,
+      "loss": 1.8115,
+      "step": 4487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0676571995337512e-05,
+      "loss": 1.8789,
+      "step": 4488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0583131732066606e-05,
+      "loss": 1.9131,
+      "step": 4489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.048989864457418e-05,
+      "loss": 1.7998,
+      "step": 4490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.039687277315011e-05,
+      "loss": 1.8867,
+      "step": 4491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0304054157994523e-05,
+      "loss": 1.8105,
+      "step": 4492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0211442839218275e-05,
+      "loss": 1.7578,
+      "step": 4493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.011903885684263e-05,
+      "loss": 1.9268,
+      "step": 4494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0026842250799037e-05,
+      "loss": 1.7891,
+      "step": 4495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9934853060929458e-05,
+      "loss": 1.9775,
+      "step": 4496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.984307132698626e-05,
+      "loss": 1.8115,
+      "step": 4497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9751497088632054e-05,
+      "loss": 1.915,
+      "step": 4498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9660130385439846e-05,
+      "loss": 1.9023,
+      "step": 4499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9568971256892942e-05,
+      "loss": 1.7969,
+      "step": 4500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.947801974238511e-05,
+      "loss": 1.8193,
+      "step": 4501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9387275881220022e-05,
+      "loss": 1.873,
+      "step": 4502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9296739712611977e-05,
+      "loss": 1.835,
+      "step": 4503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.92064112756854e-05,
+      "loss": 1.8867,
+      "step": 4504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9116290609474786e-05,
+      "loss": 1.8467,
+      "step": 4505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9026377752925095e-05,
+      "loss": 1.8535,
+      "step": 4506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.893667274489136e-05,
+      "loss": 1.9307,
+      "step": 4507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.884717562413879e-05,
+      "loss": 1.8926,
+      "step": 4508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8757886429342895e-05,
+      "loss": 1.8574,
+      "step": 4509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.866880519908909e-05,
+      "loss": 1.9375,
+      "step": 4510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8579931971872966e-05,
+      "loss": 1.8545,
+      "step": 4511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8491266786100525e-05,
+      "loss": 1.7656,
+      "step": 4512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.840280968008745e-05,
+      "loss": 1.7705,
+      "step": 4513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8314560692059833e-05,
+      "loss": 1.9014,
+      "step": 4514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.822651986015361e-05,
+      "loss": 1.8516,
+      "step": 4515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8138687222414962e-05,
+      "loss": 1.8047,
+      "step": 4516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8051062816799912e-05,
+      "loss": 1.8193,
+      "step": 4517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.796364668117445e-05,
+      "loss": 1.875,
+      "step": 4518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7876438853314968e-05,
+      "loss": 1.8086,
+      "step": 4519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7789439370907325e-05,
+      "loss": 1.8408,
+      "step": 4520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.770264827154766e-05,
+      "loss": 1.7646,
+      "step": 4521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7616065592742035e-05,
+      "loss": 1.793,
+      "step": 4522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7529691371906354e-05,
+      "loss": 1.7539,
+      "step": 4523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.744352564636642e-05,
+      "loss": 1.915,
+      "step": 4524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.735756845335812e-05,
+      "loss": 1.7832,
+      "step": 4525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7271819830026957e-05,
+      "loss": 1.8086,
+      "step": 4526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.718627981342852e-05,
+      "loss": 1.8604,
+      "step": 4527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.710094844052812e-05,
+      "loss": 1.9727,
+      "step": 4528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.701582574820104e-05,
+      "loss": 1.8564,
+      "step": 4529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6930911773232306e-05,
+      "loss": 1.9082,
+      "step": 4530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6846206552316613e-05,
+      "loss": 1.915,
+      "step": 4531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.676171012205874e-05,
+      "loss": 1.8174,
+      "step": 4532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.667742251897303e-05,
+      "loss": 1.8564,
+      "step": 4533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6593343779483517e-05,
+      "loss": 1.8408,
+      "step": 4534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6509473939924135e-05,
+      "loss": 1.8135,
+      "step": 4535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6425813036538616e-05,
+      "loss": 1.8857,
+      "step": 4536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6342361105480096e-05,
+      "loss": 1.7676,
+      "step": 4537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6259118182811785e-05,
+      "loss": 1.8164,
+      "step": 4538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.617608430450629e-05,
+      "loss": 1.8525,
+      "step": 4539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.609325950644591e-05,
+      "loss": 1.8848,
+      "step": 4540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.60106438244228e-05,
+      "loss": 1.7812,
+      "step": 4541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5928237294138394e-05,
+      "loss": 1.8945,
+      "step": 4542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5846039951204095e-05,
+      "loss": 1.7861,
+      "step": 4543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.576405183114077e-05,
+      "loss": 1.8105,
+      "step": 4544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.568227296937885e-05,
+      "loss": 1.8818,
+      "step": 4545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5600703401258297e-05,
+      "loss": 1.8486,
+      "step": 4546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.551934316202869e-05,
+      "loss": 1.9404,
+      "step": 4547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.543819228684912e-05,
+      "loss": 1.8877,
+      "step": 4548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5357250810788315e-05,
+      "loss": 1.8535,
+      "step": 4549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.52765187688243e-05,
+      "loss": 1.9316,
+      "step": 4550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5195996195844885e-05,
+      "loss": 1.8506,
+      "step": 4551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5115683126647072e-05,
+      "loss": 2.0049,
+      "step": 4552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5035579595937377e-05,
+      "loss": 1.9131,
+      "step": 4553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4955685638331995e-05,
+      "loss": 1.8301,
+      "step": 4554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4876001288356311e-05,
+      "loss": 1.8838,
+      "step": 4555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.479652658044517e-05,
+      "loss": 1.8906,
+      "step": 4556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.471726154894304e-05,
+      "loss": 1.8154,
+      "step": 4557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4638206228103413e-05,
+      "loss": 1.8457,
+      "step": 4558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4559360652089404e-05,
+      "loss": 1.8809,
+      "step": 4559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4480724854973536e-05,
+      "loss": 1.8047,
+      "step": 4560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4402298870737517e-05,
+      "loss": 1.8633,
+      "step": 4561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4324082733272348e-05,
+      "loss": 1.7119,
+      "step": 4562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4246076476378489e-05,
+      "loss": 1.9268,
+      "step": 4563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.416828013376581e-05,
+      "loss": 1.8926,
+      "step": 4564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.409069373905314e-05,
+      "loss": 1.8193,
+      "step": 4565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4013317325768826e-05,
+      "loss": 1.8896,
+      "step": 4566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3936150927350399e-05,
+      "loss": 1.8193,
+      "step": 4567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.385919457714463e-05,
+      "loss": 1.8926,
+      "step": 4568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.378244830840747e-05,
+      "loss": 1.9102,
+      "step": 4569
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3705912154304224e-05,
+      "loss": 1.8867,
+      "step": 4570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3629586147909323e-05,
+      "loss": 1.8516,
+      "step": 4571
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.355347032220633e-05,
+      "loss": 1.8623,
+      "step": 4572
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3477564710088097e-05,
+      "loss": 1.8652,
+      "step": 4573
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.34018693443565e-05,
+      "loss": 1.8496,
+      "step": 4574
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3326384257722645e-05,
+      "loss": 1.8652,
+      "step": 4575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3251109482806666e-05,
+      "loss": 1.9219,
+      "step": 4576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3176045052138097e-05,
+      "loss": 1.7881,
+      "step": 4577
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3101190998155155e-05,
+      "loss": 1.7939,
+      "step": 4578
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.302654735320552e-05,
+      "loss": 1.7373,
+      "step": 4579
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2952114149545724e-05,
+      "loss": 1.8525,
+      "step": 4580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2877891419341426e-05,
+      "loss": 1.8652,
+      "step": 4581
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2803879194667245e-05,
+      "loss": 1.8242,
+      "step": 4582
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2730077507506987e-05,
+      "loss": 1.7959,
+      "step": 4583
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2656486389753418e-05,
+      "loss": 1.8418,
+      "step": 4584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2583105873208266e-05,
+      "loss": 1.7949,
+      "step": 4585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2509935989582332e-05,
+      "loss": 1.8467,
+      "step": 4586
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2436976770495268e-05,
+      "loss": 1.833,
+      "step": 4587
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2364228247475684e-05,
+      "loss": 1.8711,
+      "step": 4588
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2291690451961435e-05,
+      "loss": 1.8398,
+      "step": 4589
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2219363415298833e-05,
+      "loss": 1.7578,
+      "step": 4590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2147247168743547e-05,
+      "loss": 1.7891,
+      "step": 4591
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2075341743459978e-05,
+      "loss": 1.8809,
+      "step": 4592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2003647170521381e-05,
+      "loss": 1.8066,
+      "step": 4593
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1932163480909864e-05,
+      "loss": 1.8545,
+      "step": 4594
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1860890705516547e-05,
+      "loss": 1.8818,
+      "step": 4595
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1789828875141351e-05,
+      "loss": 1.8506,
+      "step": 4596
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1718978020492987e-05,
+      "loss": 1.957,
+      "step": 4597
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1648338172188966e-05,
+      "loss": 1.79,
+      "step": 4598
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1577909360755812e-05,
+      "loss": 1.7197,
+      "step": 4599
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.150769161662868e-05,
+      "loss": 1.8135,
+      "step": 4600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1437684970151407e-05,
+      "loss": 1.7754,
+      "step": 4601
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1367889451576962e-05,
+      "loss": 1.8789,
+      "step": 4602
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1298305091066662e-05,
+      "loss": 1.8525,
+      "step": 4603
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.12289319186909e-05,
+      "loss": 1.8555,
+      "step": 4604
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1159769964428745e-05,
+      "loss": 1.8369,
+      "step": 4605
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1090819258167795e-05,
+      "loss": 1.8428,
+      "step": 4606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1022079829704435e-05,
+      "loss": 1.8467,
+      "step": 4607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0953551708744014e-05,
+      "loss": 1.7627,
+      "step": 4608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0885234924900234e-05,
+      "loss": 1.835,
+      "step": 4609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0817129507695534e-05,
+      "loss": 1.8711,
+      "step": 4610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0749235486561149e-05,
+      "loss": 1.7959,
+      "step": 4611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0681552890836943e-05,
+      "loss": 1.8857,
+      "step": 4612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0614081749771187e-05,
+      "loss": 1.8281,
+      "step": 4613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0546822092521113e-05,
+      "loss": 1.9248,
+      "step": 4614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0479773948152305e-05,
+      "loss": 1.8369,
+      "step": 4615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0412937345639029e-05,
+      "loss": 1.791,
+      "step": 4616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0346312313864126e-05,
+      "loss": 1.8838,
+      "step": 4617
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.027989888161901e-05,
+      "loss": 1.8887,
+      "step": 4618
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0213697077603722e-05,
+      "loss": 1.7959,
+      "step": 4619
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0147706930426715e-05,
+      "loss": 1.8887,
+      "step": 4620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0081928468605117e-05,
+      "loss": 1.8408,
+      "step": 4621
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0016361720564472e-05,
+      "loss": 1.9463,
+      "step": 4622
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.951006714638833e-06,
+      "loss": 1.8721,
+      "step": 4623
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.88586347907089e-06,
+      "loss": 1.8877,
+      "step": 4624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.820932042011622e-06,
+      "loss": 1.8418,
+      "step": 4625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.75621243152064e-06,
+      "loss": 1.8154,
+      "step": 4626
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.691704675565904e-06,
+      "loss": 1.8623,
+      "step": 4627
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.627408802024007e-06,
+      "loss": 1.8125,
+      "step": 4628
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.56332483867961e-06,
+      "loss": 1.9355,
+      "step": 4629
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.499452813226283e-06,
+      "loss": 1.8809,
+      "step": 4630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.43579275326556e-06,
+      "loss": 1.7764,
+      "step": 4631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.372344686307655e-06,
+      "loss": 1.8291,
+      "step": 4632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.309108639771136e-06,
+      "loss": 1.9531,
+      "step": 4633
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.246084640982866e-06,
+      "loss": 1.9414,
+      "step": 4634
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.183272717178059e-06,
+      "loss": 1.8535,
+      "step": 4635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.120672895500393e-06,
+      "loss": 1.8516,
+      "step": 4636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.058285203001837e-06,
+      "loss": 1.8691,
+      "step": 4637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.99610966664266e-06,
+      "loss": 1.832,
+      "step": 4638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.934146313291425e-06,
+      "loss": 1.8379,
+      "step": 4639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.872395169725156e-06,
+      "loss": 1.834,
+      "step": 4640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.810856262629064e-06,
+      "loss": 1.7988,
+      "step": 4641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.749529618596485e-06,
+      "loss": 1.8643,
+      "step": 4642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.68841526412939e-06,
+      "loss": 1.7891,
+      "step": 4643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.627513225637651e-06,
+      "loss": 1.9717,
+      "step": 4644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.56682352943966e-06,
+      "loss": 1.9082,
+      "step": 4645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.506346201761828e-06,
+      "loss": 1.7881,
+      "step": 4646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.44608126873897e-06,
+      "loss": 1.8564,
+      "step": 4647
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.386028756414089e-06,
+      "loss": 1.7744,
+      "step": 4648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.326188690738257e-06,
+      "loss": 1.877,
+      "step": 4649
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.266561097570902e-06,
+      "loss": 1.8057,
+      "step": 4650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.207146002679522e-06,
+      "loss": 1.79,
+      "step": 4651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.147943431739801e-06,
+      "loss": 1.7832,
+      "step": 4652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.088953410335609e-06,
+      "loss": 1.8828,
+      "step": 4653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.030175963959108e-06,
+      "loss": 1.8604,
+      "step": 4654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.971611118010258e-06,
+      "loss": 1.8301,
+      "step": 4655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.913258897797481e-06,
+      "loss": 1.7783,
+      "step": 4656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.855119328537108e-06,
+      "loss": 1.918,
+      "step": 4657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.797192435353651e-06,
+      "loss": 1.8848,
+      "step": 4658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.739478243279696e-06,
+      "loss": 1.8721,
+      "step": 4659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.681976777255962e-06,
+      "loss": 1.7979,
+      "step": 4660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.624688062131124e-06,
+      "loss": 1.8096,
+      "step": 4661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.5676121226621594e-06,
+      "loss": 1.8076,
+      "step": 4662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.5107489835137825e-06,
+      "loss": 1.8145,
+      "step": 4663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.454098669258891e-06,
+      "loss": 1.8516,
+      "step": 4664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.397661204378514e-06,
+      "loss": 1.8223,
+      "step": 4665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.341436613261532e-06,
+      "loss": 1.8662,
+      "step": 4666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.285424920204953e-06,
+      "loss": 1.8564,
+      "step": 4667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.229626149413693e-06,
+      "loss": 1.9404,
+      "step": 4668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.174040325000797e-06,
+      "loss": 1.8213,
+      "step": 4669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.118667470987106e-06,
+      "loss": 1.8037,
+      "step": 4670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.063507611301423e-06,
+      "loss": 1.875,
+      "step": 4671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.008560769780792e-06,
+      "loss": 1.8184,
+      "step": 4672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.953826970169886e-06,
+      "loss": 1.8447,
+      "step": 4673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.8993062361213946e-06,
+      "loss": 1.7588,
+      "step": 4674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.844998591196139e-06,
+      "loss": 1.8037,
+      "step": 4675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.790904058862568e-06,
+      "loss": 1.8125,
+      "step": 4676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.737022662497094e-06,
+      "loss": 1.7773,
+      "step": 4677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.683354425384259e-06,
+      "loss": 1.8926,
+      "step": 4678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.629899370716231e-06,
+      "loss": 1.7744,
+      "step": 4679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.576657521593088e-06,
+      "loss": 1.8984,
+      "step": 4680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.523628901022927e-06,
+      "loss": 1.7324,
+      "step": 4681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.470813531921527e-06,
+      "loss": 1.8857,
+      "step": 4682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.418211437112631e-06,
+      "loss": 1.8887,
+      "step": 4683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.365822639327723e-06,
+      "loss": 1.8584,
+      "step": 4684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.313647161206193e-06,
+      "loss": 1.9355,
+      "step": 4685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.261685025295283e-06,
+      "loss": 1.8711,
+      "step": 4686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.209936254049808e-06,
+      "loss": 1.7725,
+      "step": 4687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.158400869832604e-06,
+      "loss": 1.8506,
+      "step": 4688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.107078894914353e-06,
+      "loss": 1.8721,
+      "step": 4689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.055970351473261e-06,
+      "loss": 1.7637,
+      "step": 4690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.005075261595494e-06,
+      "loss": 1.8223,
+      "step": 4691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.954393647274958e-06,
+      "loss": 1.8936,
+      "step": 4692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.903925530413135e-06,
+      "loss": 1.7568,
+      "step": 4693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.853670932819466e-06,
+      "loss": 1.9336,
+      "step": 4694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.803629876211025e-06,
+      "loss": 1.8672,
+      "step": 4695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.753802382212625e-06,
+      "loss": 1.8389,
+      "step": 4696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7041884723567636e-06,
+      "loss": 1.8633,
+      "step": 4697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.65478816808368e-06,
+      "loss": 1.8848,
+      "step": 4698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.605601490741241e-06,
+      "loss": 1.8057,
+      "step": 4699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.556628461585001e-06,
+      "loss": 1.8154,
+      "step": 4700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.507869101778307e-06,
+      "loss": 1.7988,
+      "step": 4701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.459323432392083e-06,
+      "loss": 1.9014,
+      "step": 4702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.410991474404825e-06,
+      "loss": 1.9053,
+      "step": 4703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.362873248702827e-06,
+      "loss": 1.792,
+      "step": 4704
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.314968776080009e-06,
+      "loss": 1.8252,
+      "step": 4705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.267278077237758e-06,
+      "loss": 1.8076,
+      "step": 4706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.219801172785255e-06,
+      "loss": 1.8975,
+      "step": 4707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.172538083239198e-06,
+      "loss": 1.8838,
+      "step": 4708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.125488829023916e-06,
+      "loss": 1.7305,
+      "step": 4709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.078653430471425e-06,
+      "loss": 1.8623,
+      "step": 4710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.032031907821088e-06,
+      "loss": 1.9043,
+      "step": 4711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.985624281220014e-06,
+      "loss": 1.8066,
+      "step": 4712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.939430570722936e-06,
+      "loss": 1.6982,
+      "step": 4713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.893450796291998e-06,
+      "loss": 1.8154,
+      "step": 4714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8476849777969735e-06,
+      "loss": 1.8955,
+      "step": 4715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.802133135015096e-06,
+      "loss": 1.7881,
+      "step": 4716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.756795287631288e-06,
+      "loss": 1.8086,
+      "step": 4717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.71167145523782e-06,
+      "loss": 1.8828,
+      "step": 4718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666761657334595e-06,
+      "loss": 1.835,
+      "step": 4719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.622065913329032e-06,
+      "loss": 1.9336,
+      "step": 4720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5775842425359595e-06,
+      "loss": 1.9033,
+      "step": 4721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.533316664177667e-06,
+      "loss": 1.8701,
+      "step": 4722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.48926319738413e-06,
+      "loss": 1.8154,
+      "step": 4723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.445423861192566e-06,
+      "loss": 1.7764,
+      "step": 4724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4017986745478186e-06,
+      "loss": 1.8652,
+      "step": 4725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.358387656302143e-06,
+      "loss": 1.7852,
+      "step": 4726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.315190825215143e-06,
+      "loss": 1.7578,
+      "step": 4727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.272208199953997e-06,
+      "loss": 1.7891,
+      "step": 4728
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.229439799093293e-06,
+      "loss": 1.8408,
+      "step": 4729
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.186885641115023e-06,
+      "loss": 1.7646,
+      "step": 4730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.144545744408535e-06,
+      "loss": 1.9521,
+      "step": 4731
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1024201272706894e-06,
+      "loss": 1.7705,
+      "step": 4732
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.06050880790565e-06,
+      "loss": 1.835,
+      "step": 4733
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.018811804425093e-06,
+      "loss": 1.8379,
+      "step": 4734
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.97732913484794e-06,
+      "loss": 1.8184,
+      "step": 4735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.936060817100628e-06,
+      "loss": 1.8408,
+      "step": 4736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8950068690168375e-06,
+      "loss": 1.8594,
+      "step": 4737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8541673083377085e-06,
+      "loss": 1.8457,
+      "step": 4738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.813542152711735e-06,
+      "loss": 1.8516,
+      "step": 4739
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.773131419694653e-06,
+      "loss": 1.8096,
+      "step": 4740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.732935126749604e-06,
+      "loss": 1.7656,
+      "step": 4741
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.692953291247081e-06,
+      "loss": 1.8389,
+      "step": 4742
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6531859304649305e-06,
+      "loss": 1.9375,
+      "step": 4743
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.613633061588184e-06,
+      "loss": 1.874,
+      "step": 4744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5742947017092795e-06,
+      "loss": 1.79,
+      "step": 4745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.535170867828008e-06,
+      "loss": 1.7988,
+      "step": 4746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.496261576851345e-06,
+      "loss": 1.751,
+      "step": 4747
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.457566845593618e-06,
+      "loss": 1.8193,
+      "step": 4748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.419086690776396e-06,
+      "loss": 1.8418,
+      "step": 4749
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3808211290284885e-06,
+      "loss": 1.957,
+      "step": 4750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.342770176886112e-06,
+      "loss": 1.8799,
+      "step": 4751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.304933850792613e-06,
+      "loss": 1.9141,
+      "step": 4752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2673121670986906e-06,
+      "loss": 1.7852,
+      "step": 4753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2299051420620617e-06,
+      "loss": 1.7363,
+      "step": 4754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1927127918479624e-06,
+      "loss": 1.7734,
+      "step": 4755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1557351325287587e-06,
+      "loss": 1.957,
+      "step": 4756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1189721800839456e-06,
+      "loss": 1.8057,
+      "step": 4757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.08242395040037e-06,
+      "loss": 1.8975,
+      "step": 4758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0460904592719553e-06,
+      "loss": 1.8604,
+      "step": 4759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0099717223999733e-06,
+      "loss": 1.8301,
+      "step": 4760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9740677553928285e-06,
+      "loss": 1.9141,
+      "step": 4761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9383785737659984e-06,
+      "loss": 1.8643,
+      "step": 4762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9029041929423684e-06,
+      "loss": 1.876,
+      "step": 4763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8676446282517864e-06,
+      "loss": 1.9043,
+      "step": 4764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8325998949314534e-06,
+      "loss": 1.9541,
+      "step": 4765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.797770008125533e-06,
+      "loss": 1.8877,
+      "step": 4766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.763154982885541e-06,
+      "loss": 1.7559,
+      "step": 4767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7287548341700663e-06,
+      "loss": 1.793,
+      "step": 4768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.694569576844774e-06,
+      "loss": 1.8936,
+      "step": 4769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.660599225682514e-06,
+      "loss": 1.8086,
+      "step": 4770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6268437953633205e-06,
+      "loss": 1.8281,
+      "step": 4771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5933033004743032e-06,
+      "loss": 1.7832,
+      "step": 4772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.559977755509646e-06,
+      "loss": 1.8291,
+      "step": 4773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5268671748707173e-06,
+      "loss": 1.8955,
+      "step": 4774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4939715728659607e-06,
+      "loss": 1.8096,
+      "step": 4775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4612909637109495e-06,
+      "loss": 1.7363,
+      "step": 4776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4288253615282753e-06,
+      "loss": 1.7256,
+      "step": 4777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3965747803477155e-06,
+      "loss": 1.8564,
+      "step": 4778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.364539234106011e-06,
+      "loss": 1.6895,
+      "step": 4779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.332718736647088e-06,
+      "loss": 1.8682,
+      "step": 4780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.301113301721891e-06,
+      "loss": 1.876,
+      "step": 4781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2697229429883837e-06,
+      "loss": 1.8271,
+      "step": 4782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.238547674011715e-06,
+      "loss": 1.8115,
+      "step": 4783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2075875082639417e-06,
+      "loss": 1.8242,
+      "step": 4784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1768424591241954e-06,
+      "loss": 1.7559,
+      "step": 4785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1463125398787365e-06,
+      "loss": 1.7783,
+      "step": 4786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.115997763720845e-06,
+      "loss": 1.8848,
+      "step": 4787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.085898143750653e-06,
+      "loss": 1.7803,
+      "step": 4788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0560136929755893e-06,
+      "loss": 1.793,
+      "step": 4789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.026344424309823e-06,
+      "loss": 1.876,
+      "step": 4790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.996890350574765e-06,
+      "loss": 1.8701,
+      "step": 4791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.967651484498734e-06,
+      "loss": 1.7773,
+      "step": 4792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9386278387169553e-06,
+      "loss": 1.7881,
+      "step": 4793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9098194257718414e-06,
+      "loss": 1.7402,
+      "step": 4794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.881226258112656e-06,
+      "loss": 1.9521,
+      "step": 4795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8528483480956814e-06,
+      "loss": 1.8262,
+      "step": 4796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.824685707984164e-06,
+      "loss": 1.8301,
+      "step": 4797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7967383499483681e-06,
+      "loss": 1.8242,
+      "step": 4798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7690062860654666e-06,
+      "loss": 1.8174,
+      "step": 4799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7414895283197063e-06,
+      "loss": 1.8438,
+      "step": 4800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.714188088602131e-06,
+      "loss": 1.8818,
+      "step": 4801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6871019787108587e-06,
+      "loss": 1.8135,
+      "step": 4802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6602312103508598e-06,
+      "loss": 1.8691,
+      "step": 4803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.633575795134179e-06,
+      "loss": 1.7725,
+      "step": 4804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6071357445796576e-06,
+      "loss": 1.916,
+      "step": 4805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5809110701131557e-06,
+      "loss": 1.9189,
+      "step": 4806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5549017830674418e-06,
+      "loss": 1.9463,
+      "step": 4807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5291078946821912e-06,
+      "loss": 1.8105,
+      "step": 4808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.503529416103988e-06,
+      "loss": 1.9365,
+      "step": 4809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4781663583863792e-06,
+      "loss": 1.7656,
+      "step": 4810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4530187324897082e-06,
+      "loss": 1.875,
+      "step": 4811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4280865492814487e-06,
+      "loss": 1.9717,
+      "step": 4812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4033698195357048e-06,
+      "loss": 1.9414,
+      "step": 4813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3788685539335989e-06,
+      "loss": 1.9326,
+      "step": 4814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3545827630632168e-06,
+      "loss": 1.7832,
+      "step": 4815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3305124574193306e-06,
+      "loss": 1.8076,
+      "step": 4816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3066576474038417e-06,
+      "loss": 1.876,
+      "step": 4817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2830183433252818e-06,
+      "loss": 1.8877,
+      "step": 4818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2595945553992573e-06,
+      "loss": 1.8369,
+      "step": 4819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2363862937481152e-06,
+      "loss": 1.9229,
+      "step": 4820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2133935684010556e-06,
+      "loss": 1.9248,
+      "step": 4821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1906163892942967e-06,
+      "loss": 1.8359,
+      "step": 4822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1680547662706875e-06,
+      "loss": 1.8047,
+      "step": 4823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.14570870908004e-06,
+      "loss": 1.8691,
+      "step": 4824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.123578227379074e-06,
+      "loss": 1.8223,
+      "step": 4825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1016633307312507e-06,
+      "loss": 1.7725,
+      "step": 4826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0799640286068834e-06,
+      "loss": 1.8799,
+      "step": 4827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0584803303831381e-06,
+      "loss": 1.9551,
+      "step": 4828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0372122453440324e-06,
+      "loss": 1.9004,
+      "step": 4829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.016159782680326e-06,
+      "loss": 1.8145,
+      "step": 4830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.95322951489741e-07,
+      "loss": 1.8564,
+      "step": 4831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.747017607766862e-07,
+      "loss": 1.9453,
+      "step": 4832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.542962194524217e-07,
+      "loss": 1.8877,
+      "step": 4833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.341063363349501e-07,
+      "loss": 1.7793,
+      "step": 4834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.141321201492914e-07,
+      "loss": 1.7559,
+      "step": 4835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.943735795270969e-07,
+      "loss": 1.9229,
+      "step": 4836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.748307230067587e-07,
+      "loss": 1.6689,
+      "step": 4837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.555035590336324e-07,
+      "loss": 1.8584,
+      "step": 4838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.363920959597038e-07,
+      "loss": 1.8428,
+      "step": 4839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.174963420439219e-07,
+      "loss": 1.8496,
+      "step": 4840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.98816305451866e-07,
+      "loss": 1.8516,
+      "step": 4841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.803519942559123e-07,
+      "loss": 1.7734,
+      "step": 4842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.621034164353446e-07,
+      "loss": 1.8643,
+      "step": 4843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.440705798760217e-07,
+      "loss": 1.9531,
+      "step": 4844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.262534923707654e-07,
+      "loss": 1.7949,
+      "step": 4845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.086521616190278e-07,
+      "loss": 1.8965,
+      "step": 4846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.91266595227058e-07,
+      "loss": 1.8789,
+      "step": 4847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.740968007079018e-07,
+      "loss": 1.7422,
+      "step": 4848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.571427854813461e-07,
+      "loss": 1.8301,
+      "step": 4849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.404045568738637e-07,
+      "loss": 1.8105,
+      "step": 4850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.238821221187796e-07,
+      "loss": 1.874,
+      "step": 4851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.075754883561601e-07,
+      "loss": 1.8457,
+      "step": 4852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.914846626326464e-07,
+      "loss": 1.8232,
+      "step": 4853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.756096519018428e-07,
+      "loss": 1.8828,
+      "step": 4854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.599504630239838e-07,
+      "loss": 1.7539,
+      "step": 4855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.445071027659898e-07,
+      "loss": 1.874,
+      "step": 4856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.29279577801689e-07,
+      "loss": 1.6904,
+      "step": 4857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.142678947114288e-07,
+      "loss": 1.7881,
+      "step": 4858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.994720599824087e-07,
+      "loss": 1.9111,
+      "step": 4859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.848920800085143e-07,
+      "loss": 1.8643,
+      "step": 4860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7052796109031683e-07,
+      "loss": 1.7578,
+      "step": 4861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5637970943523953e-07,
+      "loss": 1.9434,
+      "step": 4862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4244733115722525e-07,
+      "loss": 1.8271,
+      "step": 4863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2873083227706887e-07,
+      "loss": 1.8682,
+      "step": 4864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.152302187222512e-07,
+      "loss": 1.7979,
+      "step": 4865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0194549632693867e-07,
+      "loss": 1.9092,
+      "step": 4866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.888766708319835e-07,
+      "loss": 1.8604,
+      "step": 4867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7602374788497927e-07,
+      "loss": 1.7676,
+      "step": 4868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.633867330402052e-07,
+      "loss": 1.7734,
+      "step": 4869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.509656317585708e-07,
+      "loss": 1.7607,
+      "step": 4870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.387604494078933e-07,
+      "loss": 1.8037,
+      "step": 4871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.267711912623983e-07,
+      "loss": 1.8232,
+      "step": 4872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.14997862503219e-07,
+      "loss": 1.8477,
+      "step": 4873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.034404682180081e-07,
+      "loss": 1.8438,
+      "step": 4874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9209901340132574e-07,
+      "loss": 1.8564,
+      "step": 4875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.809735029541405e-07,
+      "loss": 1.7773,
+      "step": 4876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.700639416843287e-07,
+      "loss": 1.8408,
+      "step": 4877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.593703343063414e-07,
+      "loss": 1.8418,
+      "step": 4878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.488926854413154e-07,
+      "loss": 1.8652,
+      "step": 4879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3863099961707323e-07,
+      "loss": 1.9023,
+      "step": 4880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2858528126812327e-07,
+      "loss": 1.751,
+      "step": 4881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1875553473565956e-07,
+      "loss": 1.8057,
+      "step": 4882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0914176426750642e-07,
+      "loss": 1.917,
+      "step": 4883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9974397401811838e-07,
+      "loss": 1.7793,
+      "step": 4884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.905621680487468e-07,
+      "loss": 1.8271,
+      "step": 4885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8159635032716227e-07,
+      "loss": 1.7451,
+      "step": 4886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7284652472787654e-07,
+      "loss": 1.8174,
+      "step": 4887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6431269503208723e-07,
+      "loss": 1.8193,
+      "step": 4888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5599486492756664e-07,
+      "loss": 1.7148,
+      "step": 4889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4789303800882835e-07,
+      "loss": 1.8848,
+      "step": 4890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4000721777701618e-07,
+      "loss": 1.7568,
+      "step": 4891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3233740763990422e-07,
+      "loss": 1.7686,
+      "step": 4892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.248836109118967e-07,
+      "loss": 1.7725,
+      "step": 4893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1764583081408375e-07,
+      "loss": 1.8115,
+      "step": 4894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1062407047429668e-07,
+      "loss": 1.8242,
+      "step": 4895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0381833292683052e-07,
+      "loss": 1.8799,
+      "step": 4896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.722862111277708e-08,
+      "loss": 1.8896,
+      "step": 4897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.085493787980293e-08,
+      "loss": 1.8711,
+      "step": 4898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.469728598220483e-08,
+      "loss": 1.8135,
+      "step": 4899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.875566808107637e-08,
+      "loss": 1.8672,
+      "step": 4900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.30300867439193e-08,
+      "loss": 1.8516,
+      "step": 4901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.752054444497669e-08,
+      "loss": 1.833,
+      "step": 4902
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.22270435652883e-08,
+      "loss": 1.7939,
+      "step": 4903
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.714958639224666e-08,
+      "loss": 1.7266,
+      "step": 4904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.228817512015205e-08,
+      "loss": 1.8477,
+      "step": 4905
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7642811849712974e-08,
+      "loss": 1.8848,
+      "step": 4906
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3213498588434706e-08,
+      "loss": 1.8584,
+      "step": 4907
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.900023725039725e-08,
+      "loss": 1.8008,
+      "step": 4908
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.500302965636637e-08,
+      "loss": 1.8379,
+      "step": 4909
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.122187753362704e-08,
+      "loss": 1.7432,
+      "step": 4910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7656782516205513e-08,
+      "loss": 1.8809,
+      "step": 4911
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4307746144702768e-08,
+      "loss": 1.8721,
+      "step": 4912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1174769866461053e-08,
+      "loss": 1.8232,
+      "step": 4913
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8257855035230807e-08,
+      "loss": 1.6973,
+      "step": 4914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5557002911670282e-08,
+      "loss": 1.8818,
+      "step": 4915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3072214662845916e-08,
+      "loss": 1.8936,
+      "step": 4916
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.080349136250991e-08,
+      "loss": 1.8418,
+      "step": 4917
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.750833991155727e-09,
+      "loss": 1.9121,
+      "step": 4918
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.914243435796053e-09,
+      "loss": 1.8457,
+      "step": 4919
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2937204900738165e-09,
+      "loss": 1.8447,
+      "step": 4920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.889265854317703e-09,
+      "loss": 1.748,
+      "step": 4921
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7008801353756163e-09,
+      "loss": 1.876,
+      "step": 4922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7285638468922393e-09,
+      "loss": 1.7412,
+      "step": 4923
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.723174089759645e-10,
+      "loss": 1.8496,
+      "step": 4924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.321411484764504e-10,
+      "loss": 1.8525,
+      "step": 4925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0803529876257656e-10,
+      "loss": 1.8291,
+      "step": 4926
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.7939,
+      "step": 4927
+    },
+    {
+      "epoch": 1.0,
+      "step": 4927,
+      "total_flos": 3.748687970375303e+18,
+      "train_loss": 2.0709812480972194,
+      "train_runtime": 81625.4527,
+      "train_samples_per_second": 15.45,
+      "train_steps_per_second": 0.06
+    }
+  ],
+  "max_steps": 4927,
+  "num_train_epochs": 1,
+  "total_flos": 3.748687970375303e+18,
+  "trial_name": null,
+  "trial_params": null
+}