diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13261 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 2206,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4925373134328359e-05,
+      "loss": 7.5781,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9850746268656717e-05,
+      "loss": 3.2715,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.477611940298507e-05,
+      "loss": 2.8535,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9701492537313435e-05,
+      "loss": 2.209,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.462686567164179e-05,
+      "loss": 2.1797,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.955223880597014e-05,
+      "loss": 2.1035,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001044776119402985,
+      "loss": 1.9961,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011940298507462687,
+      "loss": 1.9697,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013432835820895522,
+      "loss": 1.8076,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014925373134328358,
+      "loss": 1.7178,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016417910447761195,
+      "loss": 1.6123,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017910447761194028,
+      "loss": 1.5537,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019402985074626867,
+      "loss": 1.498,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000208955223880597,
+      "loss": 1.4639,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022388059701492538,
+      "loss": 1.3496,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023880597014925374,
+      "loss": 1.3018,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002537313432835821,
+      "loss": 1.2578,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00026865671641791044,
+      "loss": 1.2207,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002835820895522388,
+      "loss": 1.2012,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029850746268656717,
+      "loss": 1.123,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00031343283582089556,
+      "loss": 1.085,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003283582089552239,
+      "loss": 1.0986,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00034328358208955223,
+      "loss": 1.0938,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00035820895522388057,
+      "loss": 1.0654,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00037313432835820896,
+      "loss": 1.0527,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00038805970149253735,
+      "loss": 0.9922,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0004029850746268657,
+      "loss": 0.9917,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000417910447761194,
+      "loss": 0.9844,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00043283582089552236,
+      "loss": 0.9717,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00044776119402985075,
+      "loss": 0.936,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00046268656716417914,
+      "loss": 0.9541,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004776119402985075,
+      "loss": 0.9224,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004925373134328358,
+      "loss": 0.9443,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005074626865671642,
+      "loss": 0.916,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005223880597014925,
+      "loss": 0.874,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005373134328358209,
+      "loss": 0.9106,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005522388059701493,
+      "loss": 0.8633,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0005671641791044776,
+      "loss": 0.8662,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005820895522388059,
+      "loss": 0.894,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005970149253731343,
+      "loss": 0.8789,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006119402985074627,
+      "loss": 0.874,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006268656716417911,
+      "loss": 0.8652,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006417910447761194,
+      "loss": 0.856,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006567164179104478,
+      "loss": 0.8188,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006716417910447762,
+      "loss": 0.8428,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0006865671641791045,
+      "loss": 0.8511,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007014925373134329,
+      "loss": 0.8462,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007164179104477611,
+      "loss": 0.8228,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0007313432835820895,
+      "loss": 0.8096,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007462686567164179,
+      "loss": 0.7681,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007611940298507463,
+      "loss": 0.8008,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007761194029850747,
+      "loss": 0.8027,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000791044776119403,
+      "loss": 0.8076,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008059701492537314,
+      "loss": 0.7822,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008208955223880598,
+      "loss": 0.7666,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000835820895522388,
+      "loss": 0.748,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008507462686567164,
+      "loss": 0.7661,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008656716417910447,
+      "loss": 0.8076,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008805970149253731,
+      "loss": 0.7886,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0008955223880597015,
+      "loss": 0.7905,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009104477611940298,
+      "loss": 0.7241,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009253731343283583,
+      "loss": 0.772,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009402985074626867,
+      "loss": 0.7744,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000955223880597015,
+      "loss": 0.728,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009701492537313433,
+      "loss": 0.7378,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009850746268656716,
+      "loss": 0.7583,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.001,
+      "loss": 0.7603,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999994607152905,
+      "loss": 0.731,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000999997842862325,
+      "loss": 0.7666,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009999951464445937,
+      "loss": 0.7832,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000999991371467913,
+      "loss": 0.7212,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000999986517940426,
+      "loss": 0.7134,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999805858726025,
+      "loss": 0.7539,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000999973575277239,
+      "loss": 0.7104,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000999965486169458,
+      "loss": 0.6909,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999563185667088,
+      "loss": 0.7129,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999460724887676,
+      "loss": 0.7305,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999347479577364,
+      "loss": 0.6797,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999223449980435,
+      "loss": 0.6875,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009999088636364441,
+      "loss": 0.7051,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009998943039020195,
+      "loss": 0.6841,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000999878665826177,
+      "loss": 0.6514,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998619494426496,
+      "loss": 0.6846,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998441547874974,
+      "loss": 0.6802,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999825281899106,
+      "loss": 0.6787,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009998053308181865,
+      "loss": 0.6646,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997843015877763,
+      "loss": 0.709,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009997621942532382,
+      "loss": 0.6826,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999739008862261,
+      "loss": 0.6763,
+      "step": 89
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999714745464859,
+      "loss": 0.6685,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009996894041133715,
+      "loss": 0.6582,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000999662984862463,
+      "loss": 0.6841,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009996354877691237,
+      "loss": 0.6616,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009996069128926691,
+      "loss": 0.6377,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995772602947386,
+      "loss": 0.6733,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995465300392972,
+      "loss": 0.6865,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009995147221926342,
+      "loss": 0.6523,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994818368233638,
+      "loss": 0.6392,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994478740024241,
+      "loss": 0.6733,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009994128338030777,
+      "loss": 0.6572,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009993767163009111,
+      "loss": 0.6528,
+      "step": 101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009993395215738348,
+      "loss": 0.6245,
+      "step": 102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000999301249702083,
+      "loss": 0.6748,
+      "step": 103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009992619007682133,
+      "loss": 0.6362,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009992214748571072,
+      "loss": 0.6318,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009991799720559687,
+      "loss": 0.646,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000999137392454325,
+      "loss": 0.6353,
+      "step": 107
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009990937361440268,
+      "loss": 0.6187,
+      "step": 108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000999049003219246,
+      "loss": 0.6357,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000999003193776478,
+      "loss": 0.6401,
+      "step": 110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009989563079145407,
+      "loss": 0.6367,
+      "step": 111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009989083457345727,
+      "loss": 0.6455,
+      "step": 112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009988593073400353,
+      "loss": 0.6499,
+      "step": 113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009988091928367113,
+      "loss": 0.6177,
+      "step": 114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009987580023327045,
+      "loss": 0.5952,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009987057359384399,
+      "loss": 0.6147,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009986523937666636,
+      "loss": 0.6265,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009985979759324418,
+      "loss": 0.6162,
+      "step": 118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009985424825531613,
+      "loss": 0.6152,
+      "step": 119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009984859137485292,
+      "loss": 0.6299,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000998428269640572,
+      "loss": 0.5981,
+      "step": 121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009983695503536364,
+      "loss": 0.6016,
+      "step": 122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000998309756014388,
+      "loss": 0.6177,
+      "step": 123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000998248886751811,
+      "loss": 0.6387,
+      "step": 124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009981869426972097,
+      "loss": 0.6284,
+      "step": 125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009981239239842054,
+      "loss": 0.592,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009980598307487384,
+      "loss": 0.6245,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009979946631290665,
+      "loss": 0.6382,
+      "step": 128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009979284212657657,
+      "loss": 0.6304,
+      "step": 129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009978611053017286,
+      "loss": 0.6079,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009977927153821651,
+      "loss": 0.5767,
+      "step": 131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000997723251654602,
+      "loss": 0.6011,
+      "step": 132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009976527142688818,
+      "loss": 0.6323,
+      "step": 133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009975811033771636,
+      "loss": 0.6143,
+      "step": 134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009975084191339223,
+      "loss": 0.6035,
+      "step": 135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009974346616959476,
+      "loss": 0.5986,
+      "step": 136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009973598312223446,
+      "loss": 0.625,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009972839278745331,
+      "loss": 0.6079,
+      "step": 138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009972069518162472,
+      "loss": 0.6177,
+      "step": 139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009971289032135347,
+      "loss": 0.6152,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009970497822347573,
+      "loss": 0.6074,
+      "step": 141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009969695890505903,
+      "loss": 0.5825,
+      "step": 142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009968883238340213,
+      "loss": 0.5977,
+      "step": 143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009968059867603504,
+      "loss": 0.5693,
+      "step": 144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009967225780071907,
+      "loss": 0.5806,
+      "step": 145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009966380977544658,
+      "loss": 0.5874,
+      "step": 146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009965525461844119,
+      "loss": 0.563,
+      "step": 147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009964659234815753,
+      "loss": 0.583,
+      "step": 148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000996378229832813,
+      "loss": 0.5957,
+      "step": 149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009962894654272927,
+      "loss": 0.5835,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009961996304564916,
+      "loss": 0.5688,
+      "step": 151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000996108725114196,
+      "loss": 0.5703,
+      "step": 152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009960167495965016,
+      "loss": 0.5845,
+      "step": 153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000995923704101812,
+      "loss": 0.5791,
+      "step": 154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009958295888308395,
+      "loss": 0.6069,
+      "step": 155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009957344039866036,
+      "loss": 0.5771,
+      "step": 156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009956381497744317,
+      "loss": 0.5918,
+      "step": 157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009955408264019567,
+      "loss": 0.5864,
+      "step": 158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009954424340791195,
+      "loss": 0.5557,
+      "step": 159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009953429730181654,
+      "loss": 0.5732,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000995242443433646,
+      "loss": 0.5698,
+      "step": 161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009951408455424174,
+      "loss": 0.5576,
+      "step": 162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009950381795636406,
+      "loss": 0.5713,
+      "step": 163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009949344457187799,
+      "loss": 0.5469,
+      "step": 164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000994829644231604,
+      "loss": 0.5457,
+      "step": 165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009947237753281843,
+      "loss": 0.5789,
+      "step": 166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009946168392368946,
+      "loss": 0.5938,
+      "step": 167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000994508836188411,
+      "loss": 0.5552,
+      "step": 168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009943997664157108,
+      "loss": 0.5659,
+      "step": 169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009942896301540728,
+      "loss": 0.5671,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000994178427641076,
+      "loss": 0.541,
+      "step": 171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009940661591166002,
+      "loss": 0.5137,
+      "step": 172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009939528248228237,
+      "loss": 0.5596,
+      "step": 173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009938384250042247,
+      "loss": 0.5667,
+      "step": 174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000993722959907579,
+      "loss": 0.5586,
+      "step": 175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009936064297819613,
+      "loss": 0.5461,
+      "step": 176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000993488834878743,
+      "loss": 0.5503,
+      "step": 177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009933701754515926,
+      "loss": 0.5625,
+      "step": 178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009932504517564753,
+      "loss": 0.5352,
+      "step": 179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009931296640516513,
+      "loss": 0.5547,
+      "step": 180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009930078125976766,
+      "loss": 0.5576,
+      "step": 181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009928848976574018,
+      "loss": 0.5698,
+      "step": 182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009927609194959716,
+      "loss": 0.5479,
+      "step": 183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009926358783808238,
+      "loss": 0.5415,
+      "step": 184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009925097745816895,
+      "loss": 0.5654,
+      "step": 185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009923826083705924,
+      "loss": 0.5298,
+      "step": 186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009922543800218472,
+      "loss": 0.5244,
+      "step": 187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000992125089812061,
+      "loss": 0.5454,
+      "step": 188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009919947380201298,
+      "loss": 0.5388,
+      "step": 189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000991863324927241,
+      "loss": 0.5137,
+      "step": 190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000991730850816871,
+      "loss": 0.5381,
+      "step": 191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009915973159747846,
+      "loss": 0.5371,
+      "step": 192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009914627206890352,
+      "loss": 0.5278,
+      "step": 193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009913270652499634,
+      "loss": 0.4966,
+      "step": 194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000991190349950197,
+      "loss": 0.5115,
+      "step": 195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009910525750846493,
+      "loss": 0.5337,
+      "step": 196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009909137409505205,
+      "loss": 0.5037,
+      "step": 197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009907738478472947,
+      "loss": 0.5339,
+      "step": 198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000990632896076741,
+      "loss": 0.533,
+      "step": 199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009904908859429116,
+      "loss": 0.5088,
+      "step": 200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009903478177521424,
+      "loss": 0.5337,
+      "step": 201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009902036918130513,
+      "loss": 0.5098,
+      "step": 202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000990058508436538,
+      "loss": 0.5232,
+      "step": 203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009899122679357828,
+      "loss": 0.4968,
+      "step": 204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009897649706262473,
+      "loss": 0.5176,
+      "step": 205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000989616616825672,
+      "loss": 0.5178,
+      "step": 206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009894672068540764,
+      "loss": 0.5552,
+      "step": 207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009893167410337592,
+      "loss": 0.5203,
+      "step": 208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009891652196892954,
+      "loss": 0.5278,
+      "step": 209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009890126431475379,
+      "loss": 0.5339,
+      "step": 210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009888590117376154,
+      "loss": 0.5056,
+      "step": 211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009887043257909322,
+      "loss": 0.5037,
+      "step": 212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009885485856411676,
+      "loss": 0.5332,
+      "step": 213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009883917916242743,
+      "loss": 0.4895,
+      "step": 214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000988233944078479,
+      "loss": 0.5134,
+      "step": 215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009880750433442806,
+      "loss": 0.5073,
+      "step": 216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009879150897644504,
+      "loss": 0.5068,
+      "step": 217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009877540836840302,
+      "loss": 0.4871,
+      "step": 218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009875920254503326,
+      "loss": 0.4724,
+      "step": 219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009874289154129395,
+      "loss": 0.4768,
+      "step": 220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009872647539237022,
+      "loss": 0.5049,
+      "step": 221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009870995413367397,
+      "loss": 0.4854,
+      "step": 222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000986933278008438,
+      "loss": 0.4824,
+      "step": 223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000986765964297451,
+      "loss": 0.5012,
+      "step": 224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009865976005646972,
+      "loss": 0.4861,
+      "step": 225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009864281871733607,
+      "loss": 0.5012,
+      "step": 226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009862577244888896,
+      "loss": 0.4692,
+      "step": 227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009860862128789954,
+      "loss": 0.4888,
+      "step": 228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009859136527136526,
+      "loss": 0.4807,
+      "step": 229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009857400443650975,
+      "loss": 0.4854,
+      "step": 230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009855653882078275,
+      "loss": 0.4758,
+      "step": 231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009853896846185998,
+      "loss": 0.468,
+      "step": 232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000985212933976432,
+      "loss": 0.4907,
+      "step": 233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009850351366625992,
+      "loss": 0.4834,
+      "step": 234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009848562930606353,
+      "loss": 0.4829,
+      "step": 235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009846764035563306,
+      "loss": 0.4519,
+      "step": 236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009844954685377316,
+      "loss": 0.4937,
+      "step": 237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009843134883951406,
+      "loss": 0.4932,
+      "step": 238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000984130463521114,
+      "loss": 0.4817,
+      "step": 239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009839463943104613,
+      "loss": 0.4963,
+      "step": 240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000983761281160246,
+      "loss": 0.5032,
+      "step": 241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009835751244697828,
+      "loss": 0.4888,
+      "step": 242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009833879246406371,
+      "loss": 0.4827,
+      "step": 243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009831996820766254,
+      "loss": 0.4822,
+      "step": 244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009830103971838126,
+      "loss": 0.4788,
+      "step": 245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009828200703705127,
+      "loss": 0.5149,
+      "step": 246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009826287020472872,
+      "loss": 0.4983,
+      "step": 247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000982436292626944,
+      "loss": 0.4624,
+      "step": 248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000982242842524537,
+      "loss": 0.4758,
+      "step": 249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000982048352157365,
+      "loss": 0.4636,
+      "step": 250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009818528219449705,
+      "loss": 0.4751,
+      "step": 251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009816562523091393,
+      "loss": 0.4707,
+      "step": 252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009814586436738997,
+      "loss": 0.4634,
+      "step": 253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009812599964655207,
+      "loss": 0.498,
+      "step": 254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000981060311112512,
+      "loss": 0.4604,
+      "step": 255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009808595880456226,
+      "loss": 0.4812,
+      "step": 256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00098065782769784,
+      "loss": 0.489,
+      "step": 257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009804550305043895,
+      "loss": 0.5,
+      "step": 258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009802511969027325,
+      "loss": 0.4463,
+      "step": 259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009800463273325665,
+      "loss": 0.4514,
+      "step": 260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009798404222358236,
+      "loss": 0.4629,
+      "step": 261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009796334820566697,
+      "loss": 0.4958,
+      "step": 262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009794255072415035,
+      "loss": 0.4678,
+      "step": 263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009792164982389557,
+      "loss": 0.47,
+      "step": 264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009790064554998875,
+      "loss": 0.4658,
+      "step": 265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009787953794773902,
+      "loss": 0.4521,
+      "step": 266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009785832706267843,
+      "loss": 0.4639,
+      "step": 267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000978370129405618,
+      "loss": 0.4636,
+      "step": 268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009781559562736665,
+      "loss": 0.4639,
+      "step": 269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0009779407516929308,
+      "loss": 0.4373,
+      "step": 270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000977724516127637,
+      "loss": 0.4612,
+      "step": 271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009775072500442358,
+      "loss": 0.4387,
+      "step": 272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009772889539114,
+      "loss": 0.4299,
+      "step": 273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009770696282000244,
+      "loss": 0.4475,
+      "step": 274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009768492733832252,
+      "loss": 0.4451,
+      "step": 275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009766278899363385,
+      "loss": 0.4485,
+      "step": 276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009764054783369191,
+      "loss": 0.478,
+      "step": 277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009761820390647393,
+      "loss": 0.4624,
+      "step": 278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009759575726017891,
+      "loss": 0.4407,
+      "step": 279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009757320794322736,
+      "loss": 0.4604,
+      "step": 280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000975505560042613,
+      "loss": 0.4507,
+      "step": 281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009752780149214408,
+      "loss": 0.4419,
+      "step": 282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009750494445596037,
+      "loss": 0.4487,
+      "step": 283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009748198494501597,
+      "loss": 0.4299,
+      "step": 284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009745892300883772,
+      "loss": 0.457,
+      "step": 285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009743575869717342,
+      "loss": 0.4502,
+      "step": 286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000974124920599917,
+      "loss": 0.4224,
+      "step": 287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009738912314748194,
+      "loss": 0.4609,
+      "step": 288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009736565201005413,
+      "loss": 0.4446,
+      "step": 289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009734207869833878,
+      "loss": 0.4568,
+      "step": 290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009731840326318675,
+      "loss": 0.4355,
+      "step": 291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000972946257556693,
+      "loss": 0.4343,
+      "step": 292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009727074622707778,
+      "loss": 0.4397,
+      "step": 293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009724676472892365,
+      "loss": 0.4443,
+      "step": 294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009722268131293835,
+      "loss": 0.4578,
+      "step": 295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009719849603107314,
+      "loss": 0.4294,
+      "step": 296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009717420893549903,
+      "loss": 0.4487,
+      "step": 297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009714982007860665,
+      "loss": 0.4392,
+      "step": 298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009712532951300615,
+      "loss": 0.4277,
+      "step": 299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000971007372915271,
+      "loss": 0.4568,
+      "step": 300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009707604346721832,
+      "loss": 0.4539,
+      "step": 301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009705124809334782,
+      "loss": 0.4194,
+      "step": 302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009702635122340268,
+      "loss": 0.4497,
+      "step": 303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009700135291108888,
+      "loss": 0.4395,
+      "step": 304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009697625321033127,
+      "loss": 0.4651,
+      "step": 305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009695105217527337,
+      "loss": 0.4287,
+      "step": 306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009692574986027734,
+      "loss": 0.4224,
+      "step": 307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009690034631992375,
+      "loss": 0.4482,
+      "step": 308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009687484160901159,
+      "loss": 0.4312,
+      "step": 309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009684923578255804,
+      "loss": 0.4229,
+      "step": 310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009682352889579846,
+      "loss": 0.4253,
+      "step": 311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009679772100418614,
+      "loss": 0.4353,
+      "step": 312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000967718121633923,
+      "loss": 0.4504,
+      "step": 313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000967458024293059,
+      "loss": 0.4019,
+      "step": 314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009671969185803355,
+      "loss": 0.4539,
+      "step": 315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009669348050589939,
+      "loss": 0.4187,
+      "step": 316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009666716842944492,
+      "loss": 0.4255,
+      "step": 317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009664075568542896,
+      "loss": 0.4263,
+      "step": 318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009661424233082747,
+      "loss": 0.4197,
+      "step": 319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009658762842283342,
+      "loss": 0.4189,
+      "step": 320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009656091401885671,
+      "loss": 0.4172,
+      "step": 321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009653409917652404,
+      "loss": 0.4375,
+      "step": 322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009650718395367873,
+      "loss": 0.4556,
+      "step": 323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009648016840838064,
+      "loss": 0.4194,
+      "step": 324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009645305259890606,
+      "loss": 0.4084,
+      "step": 325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009642583658374756,
+      "loss": 0.4031,
+      "step": 326
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009639852042161387,
+      "loss": 0.4292,
+      "step": 327
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009637110417142974,
+      "loss": 0.4146,
+      "step": 328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009634358789233581,
+      "loss": 0.4412,
+      "step": 329
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009631597164368855,
+      "loss": 0.4109,
+      "step": 330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009628825548506001,
+      "loss": 0.4304,
+      "step": 331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009626043947623781,
+      "loss": 0.4094,
+      "step": 332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009623252367722494,
+      "loss": 0.4255,
+      "step": 333
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009620450814823965,
+      "loss": 0.4414,
+      "step": 334
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009617639294971532,
+      "loss": 0.4255,
+      "step": 335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009614817814230036,
+      "loss": 0.415,
+      "step": 336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00096119863786858,
+      "loss": 0.437,
+      "step": 337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009609144994446625,
+      "loss": 0.4211,
+      "step": 338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000960629366764177,
+      "loss": 0.4482,
+      "step": 339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009603432404421946,
+      "loss": 0.4087,
+      "step": 340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009600561210959291,
+      "loss": 0.4067,
+      "step": 341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000959768009344737,
+      "loss": 0.4182,
+      "step": 342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009594789058101153,
+      "loss": 0.4363,
+      "step": 343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009591888111157006,
+      "loss": 0.4163,
+      "step": 344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009588977258872672,
+      "loss": 0.4143,
+      "step": 345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009586056507527265,
+      "loss": 0.3923,
+      "step": 346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009583125863421251,
+      "loss": 0.4365,
+      "step": 347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009580185332876437,
+      "loss": 0.4155,
+      "step": 348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009577234922235954,
+      "loss": 0.4143,
+      "step": 349
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009574274637864247,
+      "loss": 0.4312,
+      "step": 350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009571304486147063,
+      "loss": 0.4165,
+      "step": 351
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000956832447349143,
+      "loss": 0.4004,
+      "step": 352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009565334606325647,
+      "loss": 0.4124,
+      "step": 353
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009562334891099277,
+      "loss": 0.4187,
+      "step": 354
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000955932533428312,
+      "loss": 0.4019,
+      "step": 355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009556305942369207,
+      "loss": 0.3896,
+      "step": 356
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009553276721870786,
+      "loss": 0.4253,
+      "step": 357
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009550237679322307,
+      "loss": 0.4094,
+      "step": 358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009547188821279407,
+      "loss": 0.416,
+      "step": 359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009544130154318895,
+      "loss": 0.3987,
+      "step": 360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009541061685038741,
+      "loss": 0.3977,
+      "step": 361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009537983420058058,
+      "loss": 0.4121,
+      "step": 362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009534895366017093,
+      "loss": 0.4001,
+      "step": 363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009531797529577205,
+      "loss": 0.4004,
+      "step": 364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000952868991742086,
+      "loss": 0.416,
+      "step": 365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009525572536251606,
+      "loss": 0.3999,
+      "step": 366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009522445392794069,
+      "loss": 0.3835,
+      "step": 367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009519308493793931,
+      "loss": 0.3894,
+      "step": 368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009516161846017919,
+      "loss": 0.3994,
+      "step": 369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009513005456253788,
+      "loss": 0.3691,
+      "step": 370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009509839331310311,
+      "loss": 0.4062,
+      "step": 371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009506663478017256,
+      "loss": 0.4026,
+      "step": 372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009503477903225382,
+      "loss": 0.3909,
+      "step": 373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009500282613806416,
+      "loss": 0.4072,
+      "step": 374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000949707761665304,
+      "loss": 0.4067,
+      "step": 375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009493862918678879,
+      "loss": 0.4114,
+      "step": 376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009490638526818481,
+      "loss": 0.3787,
+      "step": 377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009487404448027308,
+      "loss": 0.3892,
+      "step": 378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009484160689281716,
+      "loss": 0.3955,
+      "step": 379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009480907257578946,
+      "loss": 0.4121,
+      "step": 380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009477644159937098,
+      "loss": 0.4062,
+      "step": 381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009474371403395129,
+      "loss": 0.387,
+      "step": 382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009471088995012829,
+      "loss": 0.3936,
+      "step": 383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009467796941870808,
+      "loss": 0.396,
+      "step": 384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009464495251070482,
+      "loss": 0.4031,
+      "step": 385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009461183929734055,
+      "loss": 0.4084,
+      "step": 386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009457862985004509,
+      "loss": 0.3882,
+      "step": 387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009454532424045583,
+      "loss": 0.3718,
+      "step": 388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009451192254041758,
+      "loss": 0.3816,
+      "step": 389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009447842482198245,
+      "loss": 0.3899,
+      "step": 390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009444483115740967,
+      "loss": 0.364,
+      "step": 391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009441114161916545,
+      "loss": 0.3853,
+      "step": 392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009437735627992277,
+      "loss": 0.3762,
+      "step": 393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000943434752125613,
+      "loss": 0.4016,
+      "step": 394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009430949849016726,
+      "loss": 0.4041,
+      "step": 395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000942754261860331,
+      "loss": 0.3555,
+      "step": 396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009424125837365754,
+      "loss": 0.377,
+      "step": 397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009420699512674526,
+      "loss": 0.3992,
+      "step": 398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009417263651920689,
+      "loss": 0.3704,
+      "step": 399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009413818262515869,
+      "loss": 0.3738,
+      "step": 400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000941036335189225,
+      "loss": 0.4009,
+      "step": 401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009406898927502554,
+      "loss": 0.3804,
+      "step": 402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009403424996820022,
+      "loss": 0.3604,
+      "step": 403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009399941567338411,
+      "loss": 0.384,
+      "step": 404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009396448646571958,
+      "loss": 0.3943,
+      "step": 405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009392946242055378,
+      "loss": 0.3762,
+      "step": 406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009389434361343844,
+      "loss": 0.3894,
+      "step": 407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009385913012012972,
+      "loss": 0.4089,
+      "step": 408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009382382201658799,
+      "loss": 0.3794,
+      "step": 409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009378841937897776,
+      "loss": 0.3955,
+      "step": 410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009375292228366741,
+      "loss": 0.3918,
+      "step": 411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009371733080722911,
+      "loss": 0.3862,
+      "step": 412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009368164502643861,
+      "loss": 0.416,
+      "step": 413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009364586501827512,
+      "loss": 0.3845,
+      "step": 414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009360999085992105,
+      "loss": 0.3735,
+      "step": 415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009357402262876198,
+      "loss": 0.3867,
+      "step": 416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009353796040238633,
+      "loss": 0.3728,
+      "step": 417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009350180425858537,
+      "loss": 0.3923,
+      "step": 418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000934655542753529,
+      "loss": 0.3794,
+      "step": 419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009342921053088518,
+      "loss": 0.3728,
+      "step": 420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000933927731035807,
+      "loss": 0.3735,
+      "step": 421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009335624207204006,
+      "loss": 0.405,
+      "step": 422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009331961751506576,
+      "loss": 0.3892,
+      "step": 423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009328289951166204,
+      "loss": 0.4019,
+      "step": 424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009324608814103477,
+      "loss": 0.4229,
+      "step": 425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009320918348259115,
+      "loss": 0.387,
+      "step": 426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009317218561593969,
+      "loss": 0.3735,
+      "step": 427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009313509462088989,
+      "loss": 0.3674,
+      "step": 428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009309791057745218,
+      "loss": 0.3882,
+      "step": 429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009306063356583771,
+      "loss": 0.3535,
+      "step": 430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009302326366645819,
+      "loss": 0.3823,
+      "step": 431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009298580095992566,
+      "loss": 0.3506,
+      "step": 432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009294824552705238,
+      "loss": 0.3567,
+      "step": 433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009291059744885063,
+      "loss": 0.3892,
+      "step": 434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009287285680653254,
+      "loss": 0.373,
+      "step": 435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009283502368150994,
+      "loss": 0.375,
+      "step": 436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009279709815539412,
+      "loss": 0.3601,
+      "step": 437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009275908030999569,
+      "loss": 0.3813,
+      "step": 438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009272097022732444,
+      "loss": 0.3799,
+      "step": 439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009268276798958907,
+      "loss": 0.3799,
+      "step": 440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009264447367919717,
+      "loss": 0.3704,
+      "step": 441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009260608737875486,
+      "loss": 0.3796,
+      "step": 442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009256760917106671,
+      "loss": 0.3809,
+      "step": 443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009252903913913556,
+      "loss": 0.3721,
+      "step": 444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009249037736616234,
+      "loss": 0.3586,
+      "step": 445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009245162393554582,
+      "loss": 0.3794,
+      "step": 446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009241277893088258,
+      "loss": 0.3618,
+      "step": 447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009237384243596667,
+      "loss": 0.384,
+      "step": 448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000923348145347895,
+      "loss": 0.3582,
+      "step": 449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000922956953115397,
+      "loss": 0.3499,
+      "step": 450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009225648485060283,
+      "loss": 0.396,
+      "step": 451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009221718323656132,
+      "loss": 0.3677,
+      "step": 452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009217779055419421,
+      "loss": 0.3618,
+      "step": 453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009213830688847699,
+      "loss": 0.375,
+      "step": 454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009209873232458139,
+      "loss": 0.3687,
+      "step": 455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009205906694787525,
+      "loss": 0.3652,
+      "step": 456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009201931084392229,
+      "loss": 0.3701,
+      "step": 457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009197946409848195,
+      "loss": 0.3577,
+      "step": 458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009193952679750919,
+      "loss": 0.3438,
+      "step": 459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009189949902715431,
+      "loss": 0.3447,
+      "step": 460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009185938087376277,
+      "loss": 0.3687,
+      "step": 461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009181917242387501,
+      "loss": 0.3579,
+      "step": 462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009177887376422624,
+      "loss": 0.3823,
+      "step": 463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009173848498174623,
+      "loss": 0.3801,
+      "step": 464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009169800616355921,
+      "loss": 0.3643,
+      "step": 465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009165743739698363,
+      "loss": 0.3689,
+      "step": 466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009161677876953193,
+      "loss": 0.3708,
+      "step": 467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009157603036891042,
+      "loss": 0.3645,
+      "step": 468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009153519228301905,
+      "loss": 0.3594,
+      "step": 469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009149426459995126,
+      "loss": 0.3596,
+      "step": 470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009145324740799372,
+      "loss": 0.3682,
+      "step": 471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009141214079562625,
+      "loss": 0.3567,
+      "step": 472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009137094485152146,
+      "loss": 0.3916,
+      "step": 473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009132965966454476,
+      "loss": 0.3625,
+      "step": 474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009128828532375404,
+      "loss": 0.3765,
+      "step": 475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009124682191839946,
+      "loss": 0.3809,
+      "step": 476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009120526953792336,
+      "loss": 0.3813,
+      "step": 477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009116362827196001,
+      "loss": 0.3616,
+      "step": 478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009112189821033538,
+      "loss": 0.3533,
+      "step": 479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009108007944306702,
+      "loss": 0.3765,
+      "step": 480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009103817206036382,
+      "loss": 0.3447,
+      "step": 481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009099617615262581,
+      "loss": 0.3687,
+      "step": 482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009095409181044398,
+      "loss": 0.364,
+      "step": 483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009091191912460014,
+      "loss": 0.3564,
+      "step": 484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000908696581860666,
+      "loss": 0.3613,
+      "step": 485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009082730908600607,
+      "loss": 0.3494,
+      "step": 486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009078487191577144,
+      "loss": 0.3481,
+      "step": 487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000907423467669056,
+      "loss": 0.3657,
+      "step": 488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009069973373114118,
+      "loss": 0.356,
+      "step": 489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009065703290040042,
+      "loss": 0.3313,
+      "step": 490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009061424436679494,
+      "loss": 0.344,
+      "step": 491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009057136822262555,
+      "loss": 0.3557,
+      "step": 492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009052840456038203,
+      "loss": 0.3521,
+      "step": 493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009048535347274298,
+      "loss": 0.3855,
+      "step": 494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009044221505257558,
+      "loss": 0.3547,
+      "step": 495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009039898939293538,
+      "loss": 0.3606,
+      "step": 496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009035567658706613,
+      "loss": 0.3608,
+      "step": 497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009031227672839957,
+      "loss": 0.3584,
+      "step": 498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000902687899105552,
+      "loss": 0.3611,
+      "step": 499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009022521622734017,
+      "loss": 0.3596,
+      "step": 500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009018155577274891,
+      "loss": 0.3738,
+      "step": 501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009013780864096313,
+      "loss": 0.3677,
+      "step": 502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009009397492635143,
+      "loss": 0.3625,
+      "step": 503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009005005472346923,
+      "loss": 0.3616,
+      "step": 504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009000604812705853,
+      "loss": 0.3674,
+      "step": 505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008996195523204762,
+      "loss": 0.3552,
+      "step": 506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008991777613355103,
+      "loss": 0.3542,
+      "step": 507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008987351092686922,
+      "loss": 0.3582,
+      "step": 508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008982915970748836,
+      "loss": 0.344,
+      "step": 509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000897847225710802,
+      "loss": 0.3743,
+      "step": 510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000897401996135018,
+      "loss": 0.3569,
+      "step": 511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0008969559093079538,
+      "loss": 0.3459,
+      "step": 512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008965089661918805,
+      "loss": 0.3562,
+      "step": 513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008960611677509165,
+      "loss": 0.3523,
+      "step": 514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008956125149510252,
+      "loss": 0.3373,
+      "step": 515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008951630087600129,
+      "loss": 0.3472,
+      "step": 516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000894712650147527,
+      "loss": 0.3542,
+      "step": 517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008942614400850535,
+      "loss": 0.3674,
+      "step": 518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008938093795459151,
+      "loss": 0.3459,
+      "step": 519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008933564695052692,
+      "loss": 0.3733,
+      "step": 520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008929027109401055,
+      "loss": 0.3647,
+      "step": 521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008924481048292446,
+      "loss": 0.3496,
+      "step": 522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0008919926521533346,
+      "loss": 0.3394,
+      "step": 523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008915363538948502,
+      "loss": 0.3618,
+      "step": 524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008910792110380903,
+      "loss": 0.3416,
+      "step": 525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008906212245691754,
+      "loss": 0.322,
+      "step": 526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008901623954760459,
+      "loss": 0.3401,
+      "step": 527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00088970272474846,
+      "loss": 0.3538,
+      "step": 528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008892422133779909,
+      "loss": 0.334,
+      "step": 529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000888780862358026,
+      "loss": 0.355,
+      "step": 530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008883186726837631,
+      "loss": 0.3323,
+      "step": 531
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008878556453522099,
+      "loss": 0.3313,
+      "step": 532
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008873917813621803,
+      "loss": 0.3655,
+      "step": 533
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0008869270817142936,
+      "loss": 0.3491,
+      "step": 534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008864615474109714,
+      "loss": 0.342,
+      "step": 535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008859951794564356,
+      "loss": 0.3621,
+      "step": 536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008855279788567068,
+      "loss": 0.3589,
+      "step": 537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008850599466196016,
+      "loss": 0.3813,
+      "step": 538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008845910837547305,
+      "loss": 0.3491,
+      "step": 539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008841213912734958,
+      "loss": 0.3499,
+      "step": 540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008836508701890892,
+      "loss": 0.3389,
+      "step": 541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008831795215164901,
+      "loss": 0.3467,
+      "step": 542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008827073462724633,
+      "loss": 0.3359,
+      "step": 543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008822343454755561,
+      "loss": 0.3398,
+      "step": 544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0008817605201460969,
+      "loss": 0.3562,
+      "step": 545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008812858713061927,
+      "loss": 0.3586,
+      "step": 546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000880810399979727,
+      "loss": 0.354,
+      "step": 547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008803341071923575,
+      "loss": 0.3481,
+      "step": 548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008798569939715139,
+      "loss": 0.3357,
+      "step": 549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008793790613463954,
+      "loss": 0.3481,
+      "step": 550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008789003103479694,
+      "loss": 0.3511,
+      "step": 551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000878420742008968,
+      "loss": 0.3367,
+      "step": 552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008779403573638869,
+      "loss": 0.3486,
+      "step": 553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008774591574489823,
+      "loss": 0.3674,
+      "step": 554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008769771433022693,
+      "loss": 0.343,
+      "step": 555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008764943159635193,
+      "loss": 0.3486,
+      "step": 556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0008760106764742579,
+      "loss": 0.3308,
+      "step": 557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008755262258777626,
+      "loss": 0.3552,
+      "step": 558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008750409652190607,
+      "loss": 0.3386,
+      "step": 559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008745548955449269,
+      "loss": 0.3352,
+      "step": 560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008740680179038807,
+      "loss": 0.3521,
+      "step": 561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000873580333346185,
+      "loss": 0.3335,
+      "step": 562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008730918429238428,
+      "loss": 0.3333,
+      "step": 563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008726025476905961,
+      "loss": 0.343,
+      "step": 564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008721124487019225,
+      "loss": 0.334,
+      "step": 565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008716215470150335,
+      "loss": 0.345,
+      "step": 566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008711298436888723,
+      "loss": 0.3503,
+      "step": 567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0008706373397841114,
+      "loss": 0.3369,
+      "step": 568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008701440363631499,
+      "loss": 0.3403,
+      "step": 569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008696499344901115,
+      "loss": 0.355,
+      "step": 570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000869155035230843,
+      "loss": 0.3477,
+      "step": 571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008686593396529108,
+      "loss": 0.3552,
+      "step": 572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008681628488255986,
+      "loss": 0.3215,
+      "step": 573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008676655638199067,
+      "loss": 0.3398,
+      "step": 574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008671674857085473,
+      "loss": 0.3425,
+      "step": 575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008666686155659445,
+      "loss": 0.334,
+      "step": 576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008661689544682301,
+      "loss": 0.324,
+      "step": 577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008656685034932428,
+      "loss": 0.3579,
+      "step": 578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0008651672637205245,
+      "loss": 0.335,
+      "step": 579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008646652362313193,
+      "loss": 0.3311,
+      "step": 580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008641624221085699,
+      "loss": 0.3589,
+      "step": 581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008636588224369163,
+      "loss": 0.3108,
+      "step": 582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008631544383026931,
+      "loss": 0.3472,
+      "step": 583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008626492707939264,
+      "loss": 0.3523,
+      "step": 584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008621433210003333,
+      "loss": 0.3345,
+      "step": 585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008616365900133174,
+      "loss": 0.3325,
+      "step": 586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008611290789259676,
+      "loss": 0.3391,
+      "step": 587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008606207888330562,
+      "loss": 0.3323,
+      "step": 588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008601117208310352,
+      "loss": 0.3665,
+      "step": 589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0008596018760180351,
+      "loss": 0.3416,
+      "step": 590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008590912554938619,
+      "loss": 0.3318,
+      "step": 591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000858579860359995,
+      "loss": 0.3474,
+      "step": 592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008580676917195846,
+      "loss": 0.3357,
+      "step": 593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008575547506774497,
+      "loss": 0.3423,
+      "step": 594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008570410383400753,
+      "loss": 0.3523,
+      "step": 595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008565265558156101,
+      "loss": 0.3345,
+      "step": 596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008560113042138647,
+      "loss": 0.3105,
+      "step": 597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000855495284646308,
+      "loss": 0.3391,
+      "step": 598
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008549784982260658,
+      "loss": 0.3682,
+      "step": 599
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008544609460679184,
+      "loss": 0.3467,
+      "step": 600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0008539426292882975,
+      "loss": 0.3428,
+      "step": 601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008534235490052846,
+      "loss": 0.3513,
+      "step": 602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008529037063386076,
+      "loss": 0.3489,
+      "step": 603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008523831024096395,
+      "loss": 0.3298,
+      "step": 604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008518617383413954,
+      "loss": 0.3335,
+      "step": 605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008513396152585296,
+      "loss": 0.3303,
+      "step": 606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008508167342873342,
+      "loss": 0.3411,
+      "step": 607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008502930965557363,
+      "loss": 0.3284,
+      "step": 608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000849768703193295,
+      "loss": 0.3342,
+      "step": 609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008492435553311995,
+      "loss": 0.324,
+      "step": 610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000848717654102267,
+      "loss": 0.3298,
+      "step": 611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000848191000640939,
+      "loss": 0.3262,
+      "step": 612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008476635960832804,
+      "loss": 0.3262,
+      "step": 613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008471354415669761,
+      "loss": 0.3335,
+      "step": 614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008466065382313286,
+      "loss": 0.3538,
+      "step": 615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008460768872172557,
+      "loss": 0.322,
+      "step": 616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008455464896672887,
+      "loss": 0.3645,
+      "step": 617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008450153467255681,
+      "loss": 0.3335,
+      "step": 618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008444834595378434,
+      "loss": 0.3264,
+      "step": 619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008439508292514688,
+      "loss": 0.3364,
+      "step": 620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000843417457015402,
+      "loss": 0.3381,
+      "step": 621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008428833439802011,
+      "loss": 0.3521,
+      "step": 622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008423484912980217,
+      "loss": 0.3416,
+      "step": 623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008418129001226156,
+      "loss": 0.3208,
+      "step": 624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008412765716093271,
+      "loss": 0.3127,
+      "step": 625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008407395069150914,
+      "loss": 0.3303,
+      "step": 626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008402017071984316,
+      "loss": 0.3333,
+      "step": 627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008396631736194562,
+      "loss": 0.3074,
+      "step": 628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008391239073398573,
+      "loss": 0.3318,
+      "step": 629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008385839095229066,
+      "loss": 0.3376,
+      "step": 630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008380431813334548,
+      "loss": 0.3374,
+      "step": 631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008375017239379276,
+      "loss": 0.3254,
+      "step": 632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008369595385043235,
+      "loss": 0.3352,
+      "step": 633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008364166262022119,
+      "loss": 0.3374,
+      "step": 634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008358729882027302,
+      "loss": 0.3542,
+      "step": 635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008353286256785811,
+      "loss": 0.3379,
+      "step": 636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008347835398040297,
+      "loss": 0.3357,
+      "step": 637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008342377317549022,
+      "loss": 0.3337,
+      "step": 638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008336912027085821,
+      "loss": 0.3333,
+      "step": 639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008331439538440088,
+      "loss": 0.3147,
+      "step": 640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008325959863416741,
+      "loss": 0.3403,
+      "step": 641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008320473013836196,
+      "loss": 0.3313,
+      "step": 642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000831497900153435,
+      "loss": 0.3311,
+      "step": 643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008309477838362554,
+      "loss": 0.3557,
+      "step": 644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008303969536187577,
+      "loss": 0.3259,
+      "step": 645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008298454106891593,
+      "loss": 0.3511,
+      "step": 646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008292931562372148,
+      "loss": 0.3352,
+      "step": 647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008287401914542138,
+      "loss": 0.3442,
+      "step": 648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008281865175329782,
+      "loss": 0.3225,
+      "step": 649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008276321356678593,
+      "loss": 0.3225,
+      "step": 650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008270770470547359,
+      "loss": 0.333,
+      "step": 651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008265212528910113,
+      "loss": 0.3462,
+      "step": 652
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008259647543756105,
+      "loss": 0.3286,
+      "step": 653
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008254075527089781,
+      "loss": 0.2991,
+      "step": 654
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008248496490930754,
+      "loss": 0.3124,
+      "step": 655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008242910447313781,
+      "loss": 0.334,
+      "step": 656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008237317408288731,
+      "loss": 0.3247,
+      "step": 657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008231717385920569,
+      "loss": 0.3376,
+      "step": 658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000822611039228932,
+      "loss": 0.3347,
+      "step": 659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008220496439490046,
+      "loss": 0.3308,
+      "step": 660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008214875539632824,
+      "loss": 0.3328,
+      "step": 661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008209247704842715,
+      "loss": 0.3252,
+      "step": 662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000820361294725974,
+      "loss": 0.3284,
+      "step": 663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008197971279038854,
+      "loss": 0.3452,
+      "step": 664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008192322712349917,
+      "loss": 0.3296,
+      "step": 665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008186667259377675,
+      "loss": 0.344,
+      "step": 666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008181004932321721,
+      "loss": 0.3181,
+      "step": 667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008175335743396483,
+      "loss": 0.323,
+      "step": 668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008169659704831187,
+      "loss": 0.3391,
+      "step": 669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008163976828869839,
+      "loss": 0.3374,
+      "step": 670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008158287127771191,
+      "loss": 0.3247,
+      "step": 671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008152590613808714,
+      "loss": 0.322,
+      "step": 672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008146887299270584,
+      "loss": 0.3112,
+      "step": 673
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008141177196459643,
+      "loss": 0.3472,
+      "step": 674
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000813546031769337,
+      "loss": 0.332,
+      "step": 675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008129736675303871,
+      "loss": 0.3082,
+      "step": 676
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008124006281637838,
+      "loss": 0.3301,
+      "step": 677
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008118269149056521,
+      "loss": 0.3132,
+      "step": 678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008112525289935715,
+      "loss": 0.3464,
+      "step": 679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008106774716665721,
+      "loss": 0.3547,
+      "step": 680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008101017441651324,
+      "loss": 0.3464,
+      "step": 681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008095253477311765,
+      "loss": 0.3135,
+      "step": 682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008089482836080717,
+      "loss": 0.311,
+      "step": 683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008083705530406251,
+      "loss": 0.3333,
+      "step": 684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000807792157275082,
+      "loss": 0.3254,
+      "step": 685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008072130975591223,
+      "loss": 0.3423,
+      "step": 686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008066333751418582,
+      "loss": 0.3186,
+      "step": 687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008060529912738315,
+      "loss": 0.3203,
+      "step": 688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008054719472070106,
+      "loss": 0.3116,
+      "step": 689
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008048902441947885,
+      "loss": 0.3113,
+      "step": 690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008043078834919792,
+      "loss": 0.3074,
+      "step": 691
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008037248663548156,
+      "loss": 0.3428,
+      "step": 692
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008031411940409466,
+      "loss": 0.3293,
+      "step": 693
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008025568678094345,
+      "loss": 0.3037,
+      "step": 694
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008019718889207521,
+      "loss": 0.3359,
+      "step": 695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008013862586367798,
+      "loss": 0.3259,
+      "step": 696
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008007999782208039,
+      "loss": 0.3137,
+      "step": 697
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008002130489375123,
+      "loss": 0.3389,
+      "step": 698
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007996254720529929,
+      "loss": 0.313,
+      "step": 699
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0007990372488347311,
+      "loss": 0.3386,
+      "step": 700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007984483805516055,
+      "loss": 0.3054,
+      "step": 701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007978588684738871,
+      "loss": 0.3357,
+      "step": 702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007972687138732351,
+      "loss": 0.3331,
+      "step": 703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007966779180226951,
+      "loss": 0.3359,
+      "step": 704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007960864821966954,
+      "loss": 0.3337,
+      "step": 705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007954944076710458,
+      "loss": 0.3191,
+      "step": 706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007949016957229327,
+      "loss": 0.3274,
+      "step": 707
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007943083476309183,
+      "loss": 0.3406,
+      "step": 708
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007937143646749368,
+      "loss": 0.3293,
+      "step": 709
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007931197481362917,
+      "loss": 0.3242,
+      "step": 710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0007925244992976537,
+      "loss": 0.3079,
+      "step": 711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007919286194430573,
+      "loss": 0.325,
+      "step": 712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007913321098578978,
+      "loss": 0.3193,
+      "step": 713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007907349718289291,
+      "loss": 0.2977,
+      "step": 714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007901372066442614,
+      "loss": 0.3057,
+      "step": 715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007895388155933564,
+      "loss": 0.311,
+      "step": 716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007889397999670271,
+      "loss": 0.3064,
+      "step": 717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007883401610574337,
+      "loss": 0.2943,
+      "step": 718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00078773990015808,
+      "loss": 0.2998,
+      "step": 719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007871390185638123,
+      "loss": 0.3079,
+      "step": 720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007865375175708157,
+      "loss": 0.2969,
+      "step": 721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0007859353984766114,
+      "loss": 0.2947,
+      "step": 722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007853326625800536,
+      "loss": 0.3381,
+      "step": 723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007847293111813276,
+      "loss": 0.3132,
+      "step": 724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007841253455819462,
+      "loss": 0.2882,
+      "step": 725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007835207670847466,
+      "loss": 0.3049,
+      "step": 726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007829155769938889,
+      "loss": 0.3232,
+      "step": 727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007823097766148522,
+      "loss": 0.3235,
+      "step": 728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007817033672544321,
+      "loss": 0.3098,
+      "step": 729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007810963502207373,
+      "loss": 0.3284,
+      "step": 730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007804887268231883,
+      "loss": 0.3169,
+      "step": 731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000779880498372513,
+      "loss": 0.3154,
+      "step": 732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0007792716661807442,
+      "loss": 0.3096,
+      "step": 733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007786622315612182,
+      "loss": 0.3116,
+      "step": 734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007780521958285697,
+      "loss": 0.317,
+      "step": 735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007774415602987304,
+      "loss": 0.3083,
+      "step": 736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007768303262889261,
+      "loss": 0.312,
+      "step": 737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007762184951176733,
+      "loss": 0.3115,
+      "step": 738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007756060681047768,
+      "loss": 0.3046,
+      "step": 739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007749930465713266,
+      "loss": 0.2964,
+      "step": 740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007743794318396955,
+      "loss": 0.3096,
+      "step": 741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007737652252335355,
+      "loss": 0.3193,
+      "step": 742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007731504280777757,
+      "loss": 0.3328,
+      "step": 743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0007725350416986187,
+      "loss": 0.3091,
+      "step": 744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007719190674235384,
+      "loss": 0.3354,
+      "step": 745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007713025065812769,
+      "loss": 0.3379,
+      "step": 746
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007706853605018416,
+      "loss": 0.3059,
+      "step": 747
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000770067630516502,
+      "loss": 0.3203,
+      "step": 748
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007694493179577879,
+      "loss": 0.321,
+      "step": 749
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007688304241594847,
+      "loss": 0.3164,
+      "step": 750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007682109504566331,
+      "loss": 0.3303,
+      "step": 751
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000767590898185523,
+      "loss": 0.322,
+      "step": 752
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007669702686836936,
+      "loss": 0.3103,
+      "step": 753
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007663490632899292,
+      "loss": 0.3284,
+      "step": 754
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0007657272833442559,
+      "loss": 0.2974,
+      "step": 755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007651049301879391,
+      "loss": 0.323,
+      "step": 756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007644820051634812,
+      "loss": 0.3181,
+      "step": 757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000763858509614618,
+      "loss": 0.311,
+      "step": 758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007632344448863157,
+      "loss": 0.3125,
+      "step": 759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000762609812324769,
+      "loss": 0.3215,
+      "step": 760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007619846132773968,
+      "loss": 0.3098,
+      "step": 761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00076135884909284,
+      "loss": 0.3525,
+      "step": 762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007607325211209593,
+      "loss": 0.3145,
+      "step": 763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007601056307128308,
+      "loss": 0.3129,
+      "step": 764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007594781792207441,
+      "loss": 0.2913,
+      "step": 765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0007588501679981996,
+      "loss": 0.3044,
+      "step": 766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007582215983999043,
+      "loss": 0.281,
+      "step": 767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00075759247178177,
+      "loss": 0.3147,
+      "step": 768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007569627895009103,
+      "loss": 0.2937,
+      "step": 769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007563325529156377,
+      "loss": 0.3157,
+      "step": 770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007557017633854595,
+      "loss": 0.307,
+      "step": 771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007550704222710764,
+      "loss": 0.311,
+      "step": 772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007544385309343792,
+      "loss": 0.334,
+      "step": 773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007538060907384447,
+      "loss": 0.3452,
+      "step": 774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007531731030475345,
+      "loss": 0.3237,
+      "step": 775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000752539569227091,
+      "loss": 0.3037,
+      "step": 776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0007519054906437344,
+      "loss": 0.3042,
+      "step": 777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007512708686652603,
+      "loss": 0.3325,
+      "step": 778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007506357046606367,
+      "loss": 0.307,
+      "step": 779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00075,
+      "loss": 0.2982,
+      "step": 780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007493637560546538,
+      "loss": 0.3123,
+      "step": 781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007487269741970647,
+      "loss": 0.3152,
+      "step": 782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007480896558008593,
+      "loss": 0.2865,
+      "step": 783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000747451802240822,
+      "loss": 0.2998,
+      "step": 784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007468134148928916,
+      "loss": 0.3313,
+      "step": 785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007461744951341579,
+      "loss": 0.3201,
+      "step": 786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007455350443428598,
+      "loss": 0.303,
+      "step": 787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0007448950638983815,
+      "loss": 0.2913,
+      "step": 788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007442545551812494,
+      "loss": 0.3152,
+      "step": 789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007436135195731299,
+      "loss": 0.2993,
+      "step": 790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007429719584568259,
+      "loss": 0.3247,
+      "step": 791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007423298732162736,
+      "loss": 0.3064,
+      "step": 792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007416872652365401,
+      "loss": 0.2865,
+      "step": 793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007410441359038202,
+      "loss": 0.2971,
+      "step": 794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007404004866054327,
+      "loss": 0.3271,
+      "step": 795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007397563187298188,
+      "loss": 0.2971,
+      "step": 796
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007391116336665384,
+      "loss": 0.3091,
+      "step": 797
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000738466432806266,
+      "loss": 0.3115,
+      "step": 798
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0007378207175407898,
+      "loss": 0.2764,
+      "step": 799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007371744892630073,
+      "loss": 0.2949,
+      "step": 800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007365277493669224,
+      "loss": 0.304,
+      "step": 801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007358804992476432,
+      "loss": 0.3196,
+      "step": 802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007352327403013779,
+      "loss": 0.3086,
+      "step": 803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007345844739254324,
+      "loss": 0.3157,
+      "step": 804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007339357015182071,
+      "loss": 0.3274,
+      "step": 805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007332864244791946,
+      "loss": 0.3203,
+      "step": 806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007326366442089754,
+      "loss": 0.3362,
+      "step": 807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007319863621092156,
+      "loss": 0.2947,
+      "step": 808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007313355795826643,
+      "loss": 0.3098,
+      "step": 809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0007306842980331496,
+      "loss": 0.313,
+      "step": 810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007300325188655761,
+      "loss": 0.3198,
+      "step": 811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007293802434859221,
+      "loss": 0.303,
+      "step": 812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007287274733012362,
+      "loss": 0.2971,
+      "step": 813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007280742097196342,
+      "loss": 0.3022,
+      "step": 814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007274204541502964,
+      "loss": 0.3127,
+      "step": 815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007267662080034645,
+      "loss": 0.322,
+      "step": 816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007261114726904379,
+      "loss": 0.2983,
+      "step": 817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007254562496235717,
+      "loss": 0.3235,
+      "step": 818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007248005402162731,
+      "loss": 0.3162,
+      "step": 819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007241443458829985,
+      "loss": 0.322,
+      "step": 820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0007234876680392498,
+      "loss": 0.3096,
+      "step": 821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007228305081015728,
+      "loss": 0.3213,
+      "step": 822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007221728674875522,
+      "loss": 0.3164,
+      "step": 823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007215147476158102,
+      "loss": 0.3175,
+      "step": 824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007208561499060032,
+      "loss": 0.3108,
+      "step": 825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007201970757788173,
+      "loss": 0.2908,
+      "step": 826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007195375266559669,
+      "loss": 0.3108,
+      "step": 827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007188775039601918,
+      "loss": 0.3113,
+      "step": 828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007182170091152518,
+      "loss": 0.2943,
+      "step": 829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007175560435459262,
+      "loss": 0.312,
+      "step": 830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007168946086780098,
+      "loss": 0.3018,
+      "step": 831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0007162327059383089,
+      "loss": 0.3184,
+      "step": 832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007155703367546401,
+      "loss": 0.3088,
+      "step": 833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007149075025558255,
+      "loss": 0.3054,
+      "step": 834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007142442047716905,
+      "loss": 0.313,
+      "step": 835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007135804448330604,
+      "loss": 0.2952,
+      "step": 836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007129162241717577,
+      "loss": 0.3254,
+      "step": 837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007122515442205985,
+      "loss": 0.292,
+      "step": 838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007115864064133897,
+      "loss": 0.3088,
+      "step": 839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007109208121849261,
+      "loss": 0.2942,
+      "step": 840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007102547629709866,
+      "loss": 0.3054,
+      "step": 841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007095882602083321,
+      "loss": 0.3047,
+      "step": 842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0007089213053347015,
+      "loss": 0.2988,
+      "step": 843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007082538997888087,
+      "loss": 0.3008,
+      "step": 844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007075860450103407,
+      "loss": 0.3088,
+      "step": 845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007069177424399526,
+      "loss": 0.3154,
+      "step": 846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000706248993519266,
+      "loss": 0.3113,
+      "step": 847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007055797996908651,
+      "loss": 0.3264,
+      "step": 848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007049101623982937,
+      "loss": 0.3232,
+      "step": 849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007042400830860529,
+      "loss": 0.3005,
+      "step": 850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007035695631995965,
+      "loss": 0.3093,
+      "step": 851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007028986041853289,
+      "loss": 0.2837,
+      "step": 852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007022272074906021,
+      "loss": 0.2972,
+      "step": 853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0007015553745637117,
+      "loss": 0.2926,
+      "step": 854
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007008831068538948,
+      "loss": 0.3074,
+      "step": 855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0007002104058113263,
+      "loss": 0.3008,
+      "step": 856
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006995372728871154,
+      "loss": 0.314,
+      "step": 857
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006988637095333036,
+      "loss": 0.3027,
+      "step": 858
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006981897172028604,
+      "loss": 0.3042,
+      "step": 859
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006975152973496808,
+      "loss": 0.2855,
+      "step": 860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006968404514285821,
+      "loss": 0.3177,
+      "step": 861
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006961651808953008,
+      "loss": 0.3088,
+      "step": 862
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006954894872064889,
+      "loss": 0.29,
+      "step": 863
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006948133718197118,
+      "loss": 0.3069,
+      "step": 864
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0006941368361934442,
+      "loss": 0.3152,
+      "step": 865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000693459881787067,
+      "loss": 0.2819,
+      "step": 866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006927825100608654,
+      "loss": 0.3171,
+      "step": 867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006921047224760238,
+      "loss": 0.2996,
+      "step": 868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006914265204946244,
+      "loss": 0.3171,
+      "step": 869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006907479055796431,
+      "loss": 0.2966,
+      "step": 870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006900688791949463,
+      "loss": 0.3025,
+      "step": 871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000689389442805288,
+      "loss": 0.2931,
+      "step": 872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006887095978763072,
+      "loss": 0.3044,
+      "step": 873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006880293458745237,
+      "loss": 0.3142,
+      "step": 874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006873486882673354,
+      "loss": 0.2948,
+      "step": 875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0006866676265230154,
+      "loss": 0.2856,
+      "step": 876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006859861621107084,
+      "loss": 0.3015,
+      "step": 877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006853042965004277,
+      "loss": 0.313,
+      "step": 878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000684622031163052,
+      "loss": 0.2874,
+      "step": 879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006839393675703225,
+      "loss": 0.2844,
+      "step": 880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006832563071948394,
+      "loss": 0.3,
+      "step": 881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006825728515100586,
+      "loss": 0.3026,
+      "step": 882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000681889001990289,
+      "loss": 0.2959,
+      "step": 883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006812047601106889,
+      "loss": 0.3096,
+      "step": 884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006805201273472632,
+      "loss": 0.3083,
+      "step": 885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006798351051768597,
+      "loss": 0.3097,
+      "step": 886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0006791496950771662,
+      "loss": 0.2961,
+      "step": 887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006784638985267076,
+      "loss": 0.2947,
+      "step": 888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006777777170048423,
+      "loss": 0.2988,
+      "step": 889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006770911519917591,
+      "loss": 0.2954,
+      "step": 890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006764042049684739,
+      "loss": 0.2869,
+      "step": 891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006757168774168269,
+      "loss": 0.3035,
+      "step": 892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006750291708194793,
+      "loss": 0.3188,
+      "step": 893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006743410866599094,
+      "loss": 0.29,
+      "step": 894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006736526264224101,
+      "loss": 0.3254,
+      "step": 895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006729637915920863,
+      "loss": 0.2957,
+      "step": 896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006722745836548499,
+      "loss": 0.3118,
+      "step": 897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0006715850040974181,
+      "loss": 0.3153,
+      "step": 898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006708950544073101,
+      "loss": 0.2963,
+      "step": 899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006702047360728429,
+      "loss": 0.2979,
+      "step": 900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000669514050583129,
+      "loss": 0.2971,
+      "step": 901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000668822999428073,
+      "loss": 0.313,
+      "step": 902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000668131584098368,
+      "loss": 0.2961,
+      "step": 903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006674398060854931,
+      "loss": 0.3162,
+      "step": 904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006667476668817094,
+      "loss": 0.2952,
+      "step": 905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006660551679800573,
+      "loss": 0.2935,
+      "step": 906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006653623108743529,
+      "loss": 0.3,
+      "step": 907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006646690970591856,
+      "loss": 0.2922,
+      "step": 908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0006639755280299133,
+      "loss": 0.2954,
+      "step": 909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006632816052826611,
+      "loss": 0.2957,
+      "step": 910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006625873303143166,
+      "loss": 0.2861,
+      "step": 911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006618927046225272,
+      "loss": 0.2886,
+      "step": 912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006611977297056968,
+      "loss": 0.2847,
+      "step": 913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006605024070629832,
+      "loss": 0.3073,
+      "step": 914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006598067381942938,
+      "loss": 0.2948,
+      "step": 915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006591107246002825,
+      "loss": 0.2949,
+      "step": 916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006584143677823477,
+      "loss": 0.301,
+      "step": 917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006577176692426278,
+      "loss": 0.3203,
+      "step": 918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006570206304839978,
+      "loss": 0.2998,
+      "step": 919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006563232530100676,
+      "loss": 0.3044,
+      "step": 920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0006556255383251769,
+      "loss": 0.2881,
+      "step": 921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006549274879343931,
+      "loss": 0.3149,
+      "step": 922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006542291033435082,
+      "loss": 0.3147,
+      "step": 923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006535303860590345,
+      "loss": 0.3098,
+      "step": 924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006528313375882021,
+      "loss": 0.2957,
+      "step": 925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006521319594389554,
+      "loss": 0.2893,
+      "step": 926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006514322531199509,
+      "loss": 0.3007,
+      "step": 927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006507322201405514,
+      "loss": 0.304,
+      "step": 928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006500318620108259,
+      "loss": 0.2823,
+      "step": 929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006493311802415438,
+      "loss": 0.312,
+      "step": 930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006486301763441731,
+      "loss": 0.3003,
+      "step": 931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0006479288518308763,
+      "loss": 0.2991,
+      "step": 932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006472272082145081,
+      "loss": 0.2917,
+      "step": 933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006465252470086108,
+      "loss": 0.3062,
+      "step": 934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006458229697274125,
+      "loss": 0.2866,
+      "step": 935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006451203778858228,
+      "loss": 0.3101,
+      "step": 936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006444174729994294,
+      "loss": 0.2881,
+      "step": 937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006437142565844963,
+      "loss": 0.303,
+      "step": 938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006430107301579587,
+      "loss": 0.2856,
+      "step": 939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006423068952374207,
+      "loss": 0.3079,
+      "step": 940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000641602753341152,
+      "loss": 0.2893,
+      "step": 941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0006408983059880846,
+      "loss": 0.3022,
+      "step": 942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000640193554697809,
+      "loss": 0.3,
+      "step": 943
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006394885009905716,
+      "loss": 0.2828,
+      "step": 944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006387831463872714,
+      "loss": 0.2749,
+      "step": 945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006380774924094559,
+      "loss": 0.297,
+      "step": 946
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006373715405793188,
+      "loss": 0.3069,
+      "step": 947
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006366652924196964,
+      "loss": 0.2935,
+      "step": 948
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006359587494540637,
+      "loss": 0.2955,
+      "step": 949
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006352519132065321,
+      "loss": 0.2886,
+      "step": 950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006345447852018457,
+      "loss": 0.3003,
+      "step": 951
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006338373669653776,
+      "loss": 0.3032,
+      "step": 952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006331296600231272,
+      "loss": 0.2762,
+      "step": 953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0006324216659017168,
+      "loss": 0.2866,
+      "step": 954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006317133861283876,
+      "loss": 0.3262,
+      "step": 955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006310048222309976,
+      "loss": 0.2877,
+      "step": 956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006302959757380177,
+      "loss": 0.3057,
+      "step": 957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006295868481785281,
+      "loss": 0.2864,
+      "step": 958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006288774410822153,
+      "loss": 0.2894,
+      "step": 959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000628167755979369,
+      "loss": 0.3016,
+      "step": 960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006274577944008785,
+      "loss": 0.293,
+      "step": 961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006267475578782293,
+      "loss": 0.2946,
+      "step": 962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006260370479435005,
+      "loss": 0.2726,
+      "step": 963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006253262661293602,
+      "loss": 0.2722,
+      "step": 964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0006246152139690641,
+      "loss": 0.3059,
+      "step": 965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00062390389299645,
+      "loss": 0.2891,
+      "step": 966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006231923047459361,
+      "loss": 0.2991,
+      "step": 967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000622480450752517,
+      "loss": 0.3074,
+      "step": 968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006217683325517608,
+      "loss": 0.2911,
+      "step": 969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000621055951679805,
+      "loss": 0.2811,
+      "step": 970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006203433096733542,
+      "loss": 0.2952,
+      "step": 971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006196304080696763,
+      "loss": 0.2823,
+      "step": 972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006189172484065989,
+      "loss": 0.3137,
+      "step": 973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006182038322225063,
+      "loss": 0.3104,
+      "step": 974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006174901610563364,
+      "loss": 0.2797,
+      "step": 975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0006167762364475771,
+      "loss": 0.2842,
+      "step": 976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006160620599362627,
+      "loss": 0.2662,
+      "step": 977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006153476330629711,
+      "loss": 0.2831,
+      "step": 978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006146329573688204,
+      "loss": 0.2727,
+      "step": 979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006139180343954651,
+      "loss": 0.2943,
+      "step": 980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006132028656850935,
+      "loss": 0.3042,
+      "step": 981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006124874527804237,
+      "loss": 0.2986,
+      "step": 982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006117717972247005,
+      "loss": 0.2914,
+      "step": 983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006110559005616928,
+      "loss": 0.2937,
+      "step": 984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006103397643356887,
+      "loss": 0.3049,
+      "step": 985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006096233900914934,
+      "loss": 0.282,
+      "step": 986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0006089067793744258,
+      "loss": 0.3047,
+      "step": 987
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006081899337303147,
+      "loss": 0.2805,
+      "step": 988
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006074728547054955,
+      "loss": 0.2922,
+      "step": 989
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006067555438468075,
+      "loss": 0.3013,
+      "step": 990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006060380027015898,
+      "loss": 0.2803,
+      "step": 991
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006053202328176778,
+      "loss": 0.29,
+      "step": 992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006046022357434012,
+      "loss": 0.3054,
+      "step": 993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006038840130275794,
+      "loss": 0.2864,
+      "step": 994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006031655662195183,
+      "loss": 0.2898,
+      "step": 995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006024468968690077,
+      "loss": 0.2926,
+      "step": 996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000601728006526317,
+      "loss": 0.2877,
+      "step": 997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0006010088967421923,
+      "loss": 0.3026,
+      "step": 998
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0006002895690678534,
+      "loss": 0.3147,
+      "step": 999
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005995700250549902,
+      "loss": 0.2866,
+      "step": 1000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005988502662557587,
+      "loss": 0.3058,
+      "step": 1001
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005981302942227787,
+      "loss": 0.3008,
+      "step": 1002
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00059741011050913,
+      "loss": 0.2803,
+      "step": 1003
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005966897166683483,
+      "loss": 0.2906,
+      "step": 1004
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005959691142544236,
+      "loss": 0.2983,
+      "step": 1005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005952483048217956,
+      "loss": 0.2811,
+      "step": 1006
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005945272899253496,
+      "loss": 0.272,
+      "step": 1007
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005938060711204154,
+      "loss": 0.2913,
+      "step": 1008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0005930846499627619,
+      "loss": 0.3042,
+      "step": 1009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005923630280085947,
+      "loss": 0.3018,
+      "step": 1010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005916412068145524,
+      "loss": 0.3059,
+      "step": 1011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000590919187937704,
+      "loss": 0.2765,
+      "step": 1012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000590196972935544,
+      "loss": 0.2893,
+      "step": 1013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005894745633659906,
+      "loss": 0.2834,
+      "step": 1014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005887519607873815,
+      "loss": 0.2874,
+      "step": 1015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005880291667584708,
+      "loss": 0.3169,
+      "step": 1016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005873061828384255,
+      "loss": 0.2859,
+      "step": 1017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005865830105868225,
+      "loss": 0.2931,
+      "step": 1018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005858596515636445,
+      "loss": 0.271,
+      "step": 1019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0005851361073292776,
+      "loss": 0.2925,
+      "step": 1020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005844123794445069,
+      "loss": 0.2721,
+      "step": 1021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005836884694705143,
+      "loss": 0.2926,
+      "step": 1022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005829643789688737,
+      "loss": 0.2729,
+      "step": 1023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005822401095015489,
+      "loss": 0.2765,
+      "step": 1024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00058151566263089,
+      "loss": 0.29,
+      "step": 1025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005807910399196294,
+      "loss": 0.3176,
+      "step": 1026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005800662429308786,
+      "loss": 0.283,
+      "step": 1027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005793412732281257,
+      "loss": 0.2866,
+      "step": 1028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005786161323752307,
+      "loss": 0.3038,
+      "step": 1029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005778908219364233,
+      "loss": 0.2838,
+      "step": 1030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0005771653434762988,
+      "loss": 0.2784,
+      "step": 1031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005764396985598149,
+      "loss": 0.2812,
+      "step": 1032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005757138887522884,
+      "loss": 0.2866,
+      "step": 1033
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005749879156193919,
+      "loss": 0.2991,
+      "step": 1034
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005742617807271502,
+      "loss": 0.2809,
+      "step": 1035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005735354856419371,
+      "loss": 0.297,
+      "step": 1036
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005728090319304718,
+      "loss": 0.2993,
+      "step": 1037
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005720824211598161,
+      "loss": 0.2896,
+      "step": 1038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005713556548973701,
+      "loss": 0.2861,
+      "step": 1039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005706287347108696,
+      "loss": 0.2833,
+      "step": 1040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005699016621683823,
+      "loss": 0.2969,
+      "step": 1041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0005691744388383047,
+      "loss": 0.2817,
+      "step": 1042
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005684470662893586,
+      "loss": 0.2948,
+      "step": 1043
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005677195460905873,
+      "loss": 0.2825,
+      "step": 1044
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000566991879811353,
+      "loss": 0.2932,
+      "step": 1045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005662640690213329,
+      "loss": 0.2817,
+      "step": 1046
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005655361152905161,
+      "loss": 0.2866,
+      "step": 1047
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005648080201891995,
+      "loss": 0.2728,
+      "step": 1048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005640797852879856,
+      "loss": 0.2888,
+      "step": 1049
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005633514121577781,
+      "loss": 0.2971,
+      "step": 1050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005626229023697789,
+      "loss": 0.2816,
+      "step": 1051
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005618942574954849,
+      "loss": 0.2913,
+      "step": 1052
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0005611654791066841,
+      "loss": 0.291,
+      "step": 1053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005604365687754528,
+      "loss": 0.2695,
+      "step": 1054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005597075280741518,
+      "loss": 0.2866,
+      "step": 1055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005589783585754232,
+      "loss": 0.2863,
+      "step": 1056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005582490618521864,
+      "loss": 0.2976,
+      "step": 1057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005575196394776359,
+      "loss": 0.2711,
+      "step": 1058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005567900930252374,
+      "loss": 0.2695,
+      "step": 1059
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005560604240687234,
+      "loss": 0.2966,
+      "step": 1060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005553306341820916,
+      "loss": 0.3032,
+      "step": 1061
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005546007249395998,
+      "loss": 0.2831,
+      "step": 1062
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005538706979157635,
+      "loss": 0.2959,
+      "step": 1063
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0005531405546853525,
+      "loss": 0.2826,
+      "step": 1064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005524102968233872,
+      "loss": 0.2761,
+      "step": 1065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005516799259051349,
+      "loss": 0.2666,
+      "step": 1066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005509494435061073,
+      "loss": 0.2885,
+      "step": 1067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005502188512020566,
+      "loss": 0.2827,
+      "step": 1068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005494881505689713,
+      "loss": 0.2789,
+      "step": 1069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005487573431830745,
+      "loss": 0.2803,
+      "step": 1070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000548026430620819,
+      "loss": 0.3008,
+      "step": 1071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005472954144588847,
+      "loss": 0.2828,
+      "step": 1072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005465642962741749,
+      "loss": 0.2786,
+      "step": 1073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005458330776438135,
+      "loss": 0.2819,
+      "step": 1074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0005451017601451399,
+      "loss": 0.2816,
+      "step": 1075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005443703453557079,
+      "loss": 0.2827,
+      "step": 1076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005436388348532807,
+      "loss": 0.2867,
+      "step": 1077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005429072302158279,
+      "loss": 0.2766,
+      "step": 1078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005421755330215223,
+      "loss": 0.2904,
+      "step": 1079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005414437448487368,
+      "loss": 0.3,
+      "step": 1080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005407118672760393,
+      "loss": 0.2773,
+      "step": 1081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005399799018821917,
+      "loss": 0.3105,
+      "step": 1082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005392478502461452,
+      "loss": 0.2902,
+      "step": 1083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005385157139470365,
+      "loss": 0.2891,
+      "step": 1084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005377834945641853,
+      "loss": 0.2844,
+      "step": 1085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0005370511936770906,
+      "loss": 0.2913,
+      "step": 1086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005363188128654271,
+      "loss": 0.2804,
+      "step": 1087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005355863537090418,
+      "loss": 0.2876,
+      "step": 1088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005348538177879506,
+      "loss": 0.2794,
+      "step": 1089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005341212066823356,
+      "loss": 0.2898,
+      "step": 1090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005333885219725402,
+      "loss": 0.2904,
+      "step": 1091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005326557652390676,
+      "loss": 0.309,
+      "step": 1092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005319229380625754,
+      "loss": 0.2845,
+      "step": 1093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005311900420238736,
+      "loss": 0.2944,
+      "step": 1094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005304570787039208,
+      "loss": 0.2778,
+      "step": 1095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005297240496838206,
+      "loss": 0.3181,
+      "step": 1096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0005289909565448184,
+      "loss": 0.3057,
+      "step": 1097
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000528257800868298,
+      "loss": 0.2867,
+      "step": 1098
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005275245842357777,
+      "loss": 0.2688,
+      "step": 1099
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005267913082289079,
+      "loss": 0.2874,
+      "step": 1100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005260579744294665,
+      "loss": 0.2922,
+      "step": 1101
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005253245844193564,
+      "loss": 0.2671,
+      "step": 1102
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005245911397806017,
+      "loss": 0.2964,
+      "step": 1103
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005238576420953442,
+      "loss": 0.2533,
+      "step": 1104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005231240929458406,
+      "loss": 0.2858,
+      "step": 1105
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005223904939144578,
+      "loss": 0.2618,
+      "step": 1106
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000521656846583671,
+      "loss": 0.2626,
+      "step": 1107
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0005209231525360594,
+      "loss": 0.2739,
+      "step": 1108
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005201894133543027,
+      "loss": 0.2727,
+      "step": 1109
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005194556306211784,
+      "loss": 0.2783,
+      "step": 1110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005187218059195577,
+      "loss": 0.2578,
+      "step": 1111
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005179879408324024,
+      "loss": 0.2882,
+      "step": 1112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005172540369427611,
+      "loss": 0.2903,
+      "step": 1113
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005165200958337667,
+      "loss": 0.278,
+      "step": 1114
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005157861190886319,
+      "loss": 0.2727,
+      "step": 1115
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005150521082906466,
+      "loss": 0.2932,
+      "step": 1116
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005143180650231741,
+      "loss": 0.2969,
+      "step": 1117
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005135839908696471,
+      "loss": 0.2726,
+      "step": 1118
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0005128498874135658,
+      "loss": 0.2914,
+      "step": 1119
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005121157562384936,
+      "loss": 0.2742,
+      "step": 1120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005113815989280528,
+      "loss": 0.2687,
+      "step": 1121
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005106474170659231,
+      "loss": 0.2761,
+      "step": 1122
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005099132122358365,
+      "loss": 0.2657,
+      "step": 1123
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005091789860215746,
+      "loss": 0.2731,
+      "step": 1124
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005084447400069655,
+      "loss": 0.2805,
+      "step": 1125
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005077104757758799,
+      "loss": 0.2668,
+      "step": 1126
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005069761949122274,
+      "loss": 0.2606,
+      "step": 1127
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000506241898999954,
+      "loss": 0.2542,
+      "step": 1128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005055075896230379,
+      "loss": 0.266,
+      "step": 1129
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0005047732683654862,
+      "loss": 0.2688,
+      "step": 1130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005040389368113318,
+      "loss": 0.2756,
+      "step": 1131
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005033045965446302,
+      "loss": 0.2899,
+      "step": 1132
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005025702491494549,
+      "loss": 0.2832,
+      "step": 1133
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005018358962098954,
+      "loss": 0.2706,
+      "step": 1134
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005011015393100529,
+      "loss": 0.2687,
+      "step": 1135
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0005003671800340371,
+      "loss": 0.29,
+      "step": 1136
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000499632819965963,
+      "loss": 0.2888,
+      "step": 1137
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004988984606899472,
+      "loss": 0.2823,
+      "step": 1138
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004981641037901046,
+      "loss": 0.3062,
+      "step": 1139
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004974297508505451,
+      "loss": 0.2765,
+      "step": 1140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0004966954034553698,
+      "loss": 0.2686,
+      "step": 1141
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004959610631886681,
+      "loss": 0.256,
+      "step": 1142
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000495226731634514,
+      "loss": 0.3005,
+      "step": 1143
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004944924103769623,
+      "loss": 0.2869,
+      "step": 1144
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004937581010000462,
+      "loss": 0.2737,
+      "step": 1145
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004930238050877727,
+      "loss": 0.2881,
+      "step": 1146
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004922895242241202,
+      "loss": 0.2693,
+      "step": 1147
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004915552599930346,
+      "loss": 0.2675,
+      "step": 1148
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004908210139784254,
+      "loss": 0.2897,
+      "step": 1149
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004900867877641636,
+      "loss": 0.2915,
+      "step": 1150
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000489352582934077,
+      "loss": 0.246,
+      "step": 1151
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0004886184010719472,
+      "loss": 0.2653,
+      "step": 1152
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004878842437615065,
+      "loss": 0.2911,
+      "step": 1153
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048715011258643404,
+      "loss": 0.2819,
+      "step": 1154
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004864160091303531,
+      "loss": 0.2667,
+      "step": 1155
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004856819349768262,
+      "loss": 0.2827,
+      "step": 1156
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004849478917093535,
+      "loss": 0.2935,
+      "step": 1157
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048421388091136815,
+      "loss": 0.2822,
+      "step": 1158
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048347990416623335,
+      "loss": 0.276,
+      "step": 1159
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000482745963057239,
+      "loss": 0.2728,
+      "step": 1160
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004820120591675978,
+      "loss": 0.2637,
+      "step": 1161
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00048127819408044233,
+      "loss": 0.2817,
+      "step": 1162
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0004805443693788216,
+      "loss": 0.2838,
+      "step": 1163
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004798105866456973,
+      "loss": 0.2828,
+      "step": 1164
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047907684746394065,
+      "loss": 0.2953,
+      "step": 1165
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000478343153416329,
+      "loss": 0.2983,
+      "step": 1166
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004776095060855424,
+      "loss": 0.2573,
+      "step": 1167
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004768759070541596,
+      "loss": 0.2804,
+      "step": 1168
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047614235790465576,
+      "loss": 0.2803,
+      "step": 1169
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047540886021939845,
+      "loss": 0.2621,
+      "step": 1170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004746754155806437,
+      "loss": 0.266,
+      "step": 1171
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004739420255705337,
+      "loss": 0.2756,
+      "step": 1172
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0004732086917710922,
+      "loss": 0.271,
+      "step": 1173
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00047247541576422224,
+      "loss": 0.283,
+      "step": 1174
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004717421991317021,
+      "loss": 0.2675,
+      "step": 1175
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004710090434551816,
+      "loss": 0.283,
+      "step": 1176
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00047027595031617935,
+      "loss": 0.2899,
+      "step": 1177
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004695429212960793,
+      "loss": 0.2783,
+      "step": 1178
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046880995797612664,
+      "loss": 0.2816,
+      "step": 1179
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004680770619374248,
+      "loss": 0.2711,
+      "step": 1180
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046734423476093255,
+      "loss": 0.2687,
+      "step": 1181
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046661147802745987,
+      "loss": 0.276,
+      "step": 1182
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00046587879331766457,
+      "loss": 0.2629,
+      "step": 1183
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004651461822120494,
+      "loss": 0.2603,
+      "step": 1184
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0004644136462909584,
+      "loss": 0.2811,
+      "step": 1185
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046368118713457294,
+      "loss": 0.2802,
+      "step": 1186
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004629488063229094,
+      "loss": 0.2755,
+      "step": 1187
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046221650543581467,
+      "loss": 0.2863,
+      "step": 1188
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004614842860529635,
+      "loss": 0.2759,
+      "step": 1189
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00046075214975385486,
+      "loss": 0.2808,
+      "step": 1190
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004600200981178084,
+      "loss": 0.275,
+      "step": 1191
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004592881327239609,
+      "loss": 0.28,
+      "step": 1192
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045855625515126344,
+      "loss": 0.2651,
+      "step": 1193
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0004578244669784777,
+      "loss": 0.2657,
+      "step": 1194
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045709276978417215,
+      "loss": 0.2808,
+      "step": 1195
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00045636116514671934,
+      "loss": 0.2826,
+      "step": 1196
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045562965464429216,
+      "loss": 0.2706,
+      "step": 1197
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004548982398548601,
+      "loss": 0.2744,
+      "step": 1198
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045416692235618664,
+      "loss": 0.2694,
+      "step": 1199
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000453435703725825,
+      "loss": 0.3003,
+      "step": 1200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045270458554111533,
+      "loss": 0.2772,
+      "step": 1201
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004519735693791811,
+      "loss": 0.2797,
+      "step": 1202
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004512426568169257,
+      "loss": 0.2935,
+      "step": 1203
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00045051184943102884,
+      "loss": 0.2625,
+      "step": 1204
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004497811487979436,
+      "loss": 0.2745,
+      "step": 1205
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0004490505564938927,
+      "loss": 0.2904,
+      "step": 1206
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00044832007409486516,
+      "loss": 0.2764,
+      "step": 1207
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004475897031766129,
+      "loss": 0.2672,
+      "step": 1208
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004468594453146476,
+      "loss": 0.2935,
+      "step": 1209
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044612930208423656,
+      "loss": 0.2772,
+      "step": 1210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004453992750604003,
+      "loss": 0.266,
+      "step": 1211
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044466936581790843,
+      "loss": 0.2891,
+      "step": 1212
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044393957593127643,
+      "loss": 0.2689,
+      "step": 1213
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004432099069747625,
+      "loss": 0.2599,
+      "step": 1214
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044248036052236406,
+      "loss": 0.2739,
+      "step": 1215
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004417509381478138,
+      "loss": 0.291,
+      "step": 1216
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00044102164142457707,
+      "loss": 0.2628,
+      "step": 1217
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0004402924719258483,
+      "loss": 0.293,
+      "step": 1218
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004395634312245473,
+      "loss": 0.2598,
+      "step": 1219
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043883452089331593,
+      "loss": 0.2699,
+      "step": 1220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004381057425045152,
+      "loss": 0.2709,
+      "step": 1221
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043737709763022115,
+      "loss": 0.2897,
+      "step": 1222
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000436648587842222,
+      "loss": 0.2701,
+      "step": 1223
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043592021471201457,
+      "loss": 0.2856,
+      "step": 1224
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043519197981080056,
+      "loss": 0.2815,
+      "step": 1225
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00043446388470948395,
+      "loss": 0.2931,
+      "step": 1226
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004337359309786672,
+      "loss": 0.2803,
+      "step": 1227
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004330081201886472,
+      "loss": 0.2894,
+      "step": 1228
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0004322804539094128,
+      "loss": 0.2667,
+      "step": 1229
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00043155293371064155,
+      "loss": 0.2841,
+      "step": 1230
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00043082556116169537,
+      "loss": 0.2755,
+      "step": 1231
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00043009833783161773,
+      "loss": 0.2869,
+      "step": 1232
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042937126528913046,
+      "loss": 0.2727,
+      "step": 1233
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042864434510262996,
+      "loss": 0.2887,
+      "step": 1234
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042791757884018394,
+      "loss": 0.2928,
+      "step": 1235
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004271909680695281,
+      "loss": 0.2701,
+      "step": 1236
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000426464514358063,
+      "loss": 0.2828,
+      "step": 1237
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004257382192728498,
+      "loss": 0.2822,
+      "step": 1238
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0004250120843806083,
+      "loss": 0.2643,
+      "step": 1239
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00042428611124771184,
+      "loss": 0.2766,
+      "step": 1240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042356030144018533,
+      "loss": 0.2675,
+      "step": 1241
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004228346565237013,
+      "loss": 0.2881,
+      "step": 1242
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004221091780635768,
+      "loss": 0.2657,
+      "step": 1243
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00042138386762476934,
+      "loss": 0.2607,
+      "step": 1244
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004206587267718743,
+      "loss": 0.2587,
+      "step": 1245
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004199337570691214,
+      "loss": 0.2623,
+      "step": 1246
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004192089600803707,
+      "loss": 0.27,
+      "step": 1247
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004184843373691099,
+      "loss": 0.2567,
+      "step": 1248
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00041775989049845104,
+      "loss": 0.2609,
+      "step": 1249
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0004170356210311264,
+      "loss": 0.2662,
+      "step": 1250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00041631153052948595,
+      "loss": 0.2656,
+      "step": 1251
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041558762055549314,
+      "loss": 0.2842,
+      "step": 1252
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041486389267072257,
+      "loss": 0.2609,
+      "step": 1253
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004141403484363556,
+      "loss": 0.2631,
+      "step": 1254
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004134169894131776,
+      "loss": 0.2867,
+      "step": 1255
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004126938171615746,
+      "loss": 0.2639,
+      "step": 1256
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00041197083324152927,
+      "loss": 0.2925,
+      "step": 1257
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004112480392126187,
+      "loss": 0.2637,
+      "step": 1258
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004105254366340095,
+      "loss": 0.269,
+      "step": 1259
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00040980302706445607,
+      "loss": 0.2465,
+      "step": 1260
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0004090808120622961,
+      "loss": 0.2571,
+      "step": 1261
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00040835879318544744,
+      "loss": 0.2698,
+      "step": 1262
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004076369719914055,
+      "loss": 0.2731,
+      "step": 1263
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004069153500372382,
+      "loss": 0.2783,
+      "step": 1264
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040619392887958475,
+      "loss": 0.2506,
+      "step": 1265
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004054727100746505,
+      "loss": 0.2766,
+      "step": 1266
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004047516951782045,
+      "loss": 0.2891,
+      "step": 1267
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004040308857455763,
+      "loss": 0.2648,
+      "step": 1268
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004033102833316517,
+      "loss": 0.2607,
+      "step": 1269
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004025898894908702,
+      "loss": 0.2621,
+      "step": 1270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004018697057772213,
+      "loss": 0.2867,
+      "step": 1271
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00040114973374424126,
+      "loss": 0.2585,
+      "step": 1272
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0004004299749450098,
+      "loss": 0.2659,
+      "step": 1273
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003997104309321465,
+      "loss": 0.2655,
+      "step": 1274
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039899110325780786,
+      "loss": 0.2827,
+      "step": 1275
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003982719934736832,
+      "loss": 0.2736,
+      "step": 1276
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039755310313099244,
+      "loss": 0.2842,
+      "step": 1277
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039683443378048174,
+      "loss": 0.2731,
+      "step": 1278
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039611598697242067,
+      "loss": 0.2778,
+      "step": 1279
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003953977642565989,
+      "loss": 0.2783,
+      "step": 1280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003946797671823223,
+      "loss": 0.2595,
+      "step": 1281
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039396199729841043,
+      "loss": 0.3037,
+      "step": 1282
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039324445615319253,
+      "loss": 0.2759,
+      "step": 1283
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00039252714529450445,
+      "loss": 0.2532,
+      "step": 1284
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003918100662696853,
+      "loss": 0.2523,
+      "step": 1285
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039109322062557425,
+      "loss": 0.2682,
+      "step": 1286
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003903766099085068,
+      "loss": 0.2693,
+      "step": 1287
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038966023566431153,
+      "loss": 0.2847,
+      "step": 1288
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038894409943830726,
+      "loss": 0.2673,
+      "step": 1289
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003882282027752995,
+      "loss": 0.2686,
+      "step": 1290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003875125472195764,
+      "loss": 0.266,
+      "step": 1291
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038679713431490667,
+      "loss": 0.2803,
+      "step": 1292
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038608196560453495,
+      "loss": 0.2859,
+      "step": 1293
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003853670426311797,
+      "loss": 0.2644,
+      "step": 1294
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038465236693702897,
+      "loss": 0.2721,
+      "step": 1295
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00038393794006373735,
+      "loss": 0.2605,
+      "step": 1296
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038322376355242287,
+      "loss": 0.2639,
+      "step": 1297
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003825098389436635,
+      "loss": 0.2612,
+      "step": 1298
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038179616777749383,
+      "loss": 0.2678,
+      "step": 1299
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00038108275159340127,
+      "loss": 0.2734,
+      "step": 1300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003803695919303238,
+      "loss": 0.2759,
+      "step": 1301
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0003796566903266459,
+      "loss": 0.2719,
+      "step": 1302
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00037894404832019516,
+      "loss": 0.2711,
+      "step": 1303
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00037823166744823934,
+      "loss": 0.2784,
+      "step": 1304
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00037751954924748304,
+      "loss": 0.2938,
+      "step": 1305
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00037680769525406395,
+      "loss": 0.2644,
+      "step": 1306
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00037609610700355014,
+      "loss": 0.2578,
+      "step": 1307
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037538478603093596,
+      "loss": 0.2822,
+      "step": 1308
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037467373387063967,
+      "loss": 0.2606,
+      "step": 1309
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037396295205649964,
+      "loss": 0.2684,
+      "step": 1310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003732524421217709,
+      "loss": 0.2634,
+      "step": 1311
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003725422055991218,
+      "loss": 0.2686,
+      "step": 1312
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037183224402063107,
+      "loss": 0.2711,
+      "step": 1313
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00037112255891778487,
+      "loss": 0.2736,
+      "step": 1314
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000370413151821472,
+      "loss": 0.2498,
+      "step": 1315
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003697040242619823,
+      "loss": 0.2682,
+      "step": 1316
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0003689951777690024,
+      "loss": 0.2637,
+      "step": 1317
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00036828661387161256,
+      "loss": 0.2742,
+      "step": 1318
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036757833409828335,
+      "loss": 0.2534,
+      "step": 1319
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036687033997687287,
+      "loss": 0.2695,
+      "step": 1320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003661626330346224,
+      "loss": 0.2581,
+      "step": 1321
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036545521479815435,
+      "loss": 0.2521,
+      "step": 1322
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036474808679346795,
+      "loss": 0.2874,
+      "step": 1323
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003640412505459365,
+      "loss": 0.2544,
+      "step": 1324
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036333470758030385,
+      "loss": 0.2698,
+      "step": 1325
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036262845942068133,
+      "loss": 0.2692,
+      "step": 1326
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00036192250759054423,
+      "loss": 0.2654,
+      "step": 1327
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003612168536127287,
+      "loss": 0.2808,
+      "step": 1328
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003605114990094285,
+      "loss": 0.2751,
+      "step": 1329
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035980644530219107,
+      "loss": 0.271,
+      "step": 1330
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035910169401191544,
+      "loss": 0.271,
+      "step": 1331
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035839724665884796,
+      "loss": 0.2836,
+      "step": 1332
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035769310476257933,
+      "loss": 0.2625,
+      "step": 1333
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003569892698420413,
+      "loss": 0.2755,
+      "step": 1334
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003562857434155038,
+      "loss": 0.2823,
+      "step": 1335
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003555825270005707,
+      "loss": 0.2644,
+      "step": 1336
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035487962211417745,
+      "loss": 0.2635,
+      "step": 1337
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003541770302725875,
+      "loss": 0.2516,
+      "step": 1338
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0003534747529913893,
+      "loss": 0.2771,
+      "step": 1339
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00035277279178549205,
+      "loss": 0.2689,
+      "step": 1340
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003520711481691238,
+      "loss": 0.2675,
+      "step": 1341
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00035136982365582706,
+      "loss": 0.2952,
+      "step": 1342
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003506688197584562,
+      "loss": 0.2443,
+      "step": 1343
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034996813798917416,
+      "loss": 0.2854,
+      "step": 1344
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034926777985944856,
+      "loss": 0.2446,
+      "step": 1345
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034856774688004917,
+      "loss": 0.2762,
+      "step": 1346
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003478680405610446,
+      "loss": 0.2789,
+      "step": 1347
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003471686624117982,
+      "loss": 0.2888,
+      "step": 1348
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034646961394096563,
+      "loss": 0.2479,
+      "step": 1349
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0003457708966564918,
+      "loss": 0.2615,
+      "step": 1350
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00034507251206560686,
+      "loss": 0.2728,
+      "step": 1351
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003443744616748231,
+      "loss": 0.2601,
+      "step": 1352
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003436767469899324,
+      "loss": 0.2766,
+      "step": 1353
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034297936951600214,
+      "loss": 0.2572,
+      "step": 1354
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034228233075737224,
+      "loss": 0.2717,
+      "step": 1355
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00034158563221765224,
+      "loss": 0.2667,
+      "step": 1356
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003408892753997175,
+      "loss": 0.2607,
+      "step": 1357
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003401932618057063,
+      "loss": 0.2463,
+      "step": 1358
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00033949759293701694,
+      "loss": 0.2566,
+      "step": 1359
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00033880227029430335,
+      "loss": 0.254,
+      "step": 1360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0003381072953774731,
+      "loss": 0.292,
+      "step": 1361
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00033741266968568363,
+      "loss": 0.2549,
+      "step": 1362
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033671839471733907,
+      "loss": 0.2778,
+      "step": 1363
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003360244719700868,
+      "loss": 0.2532,
+      "step": 1364
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033533090294081446,
+      "loss": 0.2769,
+      "step": 1365
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033463768912564705,
+      "loss": 0.266,
+      "step": 1366
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033394483201994277,
+      "loss": 0.2634,
+      "step": 1367
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003332523331182906,
+      "loss": 0.2659,
+      "step": 1368
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033256019391450693,
+      "loss": 0.2595,
+      "step": 1369
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00033186841590163196,
+      "loss": 0.2623,
+      "step": 1370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003311770005719272,
+      "loss": 0.2888,
+      "step": 1371
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0003304859494168712,
+      "loss": 0.2598,
+      "step": 1372
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00032979526392715733,
+      "loss": 0.2725,
+      "step": 1373
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032910494559269007,
+      "loss": 0.2642,
+      "step": 1374
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000328414995902582,
+      "loss": 0.2512,
+      "step": 1375
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003277254163451503,
+      "loss": 0.26,
+      "step": 1376
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003270362084079138,
+      "loss": 0.2782,
+      "step": 1377
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003263473735775899,
+      "loss": 0.2559,
+      "step": 1378
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003256589133400908,
+      "loss": 0.2518,
+      "step": 1379
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032497082918052074,
+      "loss": 0.2506,
+      "step": 1380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003242831225831731,
+      "loss": 0.2677,
+      "step": 1381
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032359579503152605,
+      "loss": 0.235,
+      "step": 1382
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00032290884800824105,
+      "loss": 0.2731,
+      "step": 1383
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0003222222829951578,
+      "loss": 0.2596,
+      "step": 1384
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003215361014732925,
+      "loss": 0.2612,
+      "step": 1385
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003208503049228339,
+      "loss": 0.2557,
+      "step": 1386
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003201648948231404,
+      "loss": 0.2607,
+      "step": 1387
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003194798726527369,
+      "loss": 0.2557,
+      "step": 1388
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003187952398893111,
+      "loss": 0.2653,
+      "step": 1389
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003181109980097111,
+      "loss": 0.2567,
+      "step": 1390
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0003174271484899415,
+      "loss": 0.2343,
+      "step": 1391
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031674369280516065,
+      "loss": 0.2729,
+      "step": 1392
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031606063242967754,
+      "loss": 0.2709,
+      "step": 1393
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031537796883694805,
+      "loss": 0.2577,
+      "step": 1394
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00031469570349957254,
+      "loss": 0.2712,
+      "step": 1395
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003140138378892917,
+      "loss": 0.241,
+      "step": 1396
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031333237347698473,
+      "loss": 0.252,
+      "step": 1397
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031265131173266464,
+      "loss": 0.2554,
+      "step": 1398
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00031197065412547634,
+      "loss": 0.2733,
+      "step": 1399
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003112904021236929,
+      "loss": 0.2631,
+      "step": 1400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000310610557194712,
+      "loss": 0.2495,
+      "step": 1401
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00030993112080505384,
+      "loss": 0.2725,
+      "step": 1402
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000309252094420357,
+      "loss": 0.2454,
+      "step": 1403
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003085734795053755,
+      "loss": 0.2709,
+      "step": 1404
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0003078952775239762,
+      "loss": 0.2812,
+      "step": 1405
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00030721748993913466,
+      "loss": 0.2467,
+      "step": 1406
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030654011821293315,
+      "loss": 0.2605,
+      "step": 1407
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030586316380655607,
+      "loss": 0.2595,
+      "step": 1408
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030518662818028843,
+      "loss": 0.2722,
+      "step": 1409
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003045105127935112,
+      "loss": 0.2363,
+      "step": 1410
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030383481910469933,
+      "loss": 0.265,
+      "step": 1411
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030315954857141806,
+      "loss": 0.2639,
+      "step": 1412
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003024847026503193,
+      "loss": 0.2756,
+      "step": 1413
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00030181028279713967,
+      "loss": 0.2777,
+      "step": 1414
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003011362904666965,
+      "loss": 0.2407,
+      "step": 1415
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0003004627271128846,
+      "loss": 0.2671,
+      "step": 1416
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002997895941886737,
+      "loss": 0.283,
+      "step": 1417
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029911689314610513,
+      "loss": 0.2621,
+      "step": 1418
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029844462543628846,
+      "loss": 0.2885,
+      "step": 1419
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002977727925093981,
+      "loss": 0.2728,
+      "step": 1420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029710139581467116,
+      "loss": 0.2723,
+      "step": 1421
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002964304368004037,
+      "loss": 0.267,
+      "step": 1422
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002957599169139472,
+      "loss": 0.2727,
+      "step": 1423
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00029508983760170637,
+      "loss": 0.2571,
+      "step": 1424
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002944202003091351,
+      "loss": 0.2532,
+      "step": 1425
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002937510064807341,
+      "loss": 0.2534,
+      "step": 1426
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002930822575600475,
+      "loss": 0.2667,
+      "step": 1427
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002924139549896594,
+      "loss": 0.2559,
+      "step": 1428
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00029174610021119135,
+      "loss": 0.2401,
+      "step": 1429
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002910786946652987,
+      "loss": 0.2399,
+      "step": 1430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002904117397916681,
+      "loss": 0.2482,
+      "step": 1431
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028974523702901344,
+      "loss": 0.2523,
+      "step": 1432
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028907918781507396,
+      "loss": 0.2562,
+      "step": 1433
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002884135935866103,
+      "loss": 0.2675,
+      "step": 1434
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002877484557794017,
+      "loss": 0.2562,
+      "step": 1435
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028708377582824246,
+      "loss": 0.2677,
+      "step": 1436
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028641955516693976,
+      "loss": 0.2682,
+      "step": 1437
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028575579522830964,
+      "loss": 0.2593,
+      "step": 1438
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002850924974441745,
+      "loss": 0.2688,
+      "step": 1439
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002844296632453598,
+      "loss": 0.2546,
+      "step": 1440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002837672940616911,
+      "loss": 0.2952,
+      "step": 1441
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002831053913219903,
+      "loss": 0.2668,
+      "step": 1442
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002824439564540739,
+      "loss": 0.2651,
+      "step": 1443
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028178299088474834,
+      "loss": 0.246,
+      "step": 1444
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002811224960398083,
+      "loss": 0.2456,
+      "step": 1445
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028046247334403295,
+      "loss": 0.27,
+      "step": 1446
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002798029242211828,
+      "loss": 0.2551,
+      "step": 1447
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00027914385009399713,
+      "loss": 0.2803,
+      "step": 1448
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00027848525238418986,
+      "loss": 0.2802,
+      "step": 1449
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000277827132512448,
+      "loss": 0.2635,
+      "step": 1450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027716949189842733,
+      "loss": 0.2651,
+      "step": 1451
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027651233196075004,
+      "loss": 0.2437,
+      "step": 1452
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027585565411700165,
+      "loss": 0.2479,
+      "step": 1453
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027519945978372683,
+      "loss": 0.2814,
+      "step": 1454
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027454375037642856,
+      "loss": 0.2412,
+      "step": 1455
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027388852730956235,
+      "loss": 0.2798,
+      "step": 1456
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027323379199653574,
+      "loss": 0.2467,
+      "step": 1457
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002725795458497036,
+      "loss": 0.2684,
+      "step": 1458
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002719257902803658,
+      "loss": 0.2712,
+      "step": 1459
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00027127252669876395,
+      "loss": 0.2594,
+      "step": 1460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000270619756514078,
+      "loss": 0.2778,
+      "step": 1461
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026996748113442394,
+      "loss": 0.2655,
+      "step": 1462
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026931570196685054,
+      "loss": 0.2656,
+      "step": 1463
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026866442041733566,
+      "loss": 0.2672,
+      "step": 1464
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026801363789078446,
+      "loss": 0.2659,
+      "step": 1465
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026736335579102474,
+      "loss": 0.2675,
+      "step": 1466
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002667135755208057,
+      "loss": 0.2594,
+      "step": 1467
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026606429848179304,
+      "loss": 0.2396,
+      "step": 1468
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002654155260745679,
+      "loss": 0.2528,
+      "step": 1469
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026476725969862226,
+      "loss": 0.2587,
+      "step": 1470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002641195007523568,
+      "loss": 0.2592,
+      "step": 1471
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00026347225063307766,
+      "loss": 0.2448,
+      "step": 1472
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00026282551073699285,
+      "loss": 0.265,
+      "step": 1473
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002621792824592103,
+      "loss": 0.2639,
+      "step": 1474
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002615335671937341,
+      "loss": 0.2484,
+      "step": 1475
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002608883663334617,
+      "loss": 0.2581,
+      "step": 1476
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000260243681270181,
+      "loss": 0.257,
+      "step": 1477
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025959951339456737,
+      "loss": 0.2531,
+      "step": 1478
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025895586409618014,
+      "loss": 0.2699,
+      "step": 1479
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025831273476346005,
+      "loss": 0.2778,
+      "step": 1480
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002576701267837265,
+      "loss": 0.2611,
+      "step": 1481
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025702804154317414,
+      "loss": 0.293,
+      "step": 1482
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025638648042687005,
+      "loss": 0.2593,
+      "step": 1483
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002557454448187507,
+      "loss": 0.2494,
+      "step": 1484
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025510493610161867,
+      "loss": 0.245,
+      "step": 1485
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002544649556571402,
+      "loss": 0.2598,
+      "step": 1486
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002538255048658421,
+      "loss": 0.2456,
+      "step": 1487
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025318658510710846,
+      "loss": 0.2638,
+      "step": 1488
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002525481977591779,
+      "loss": 0.256,
+      "step": 1489
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025191034419914075,
+      "loss": 0.2433,
+      "step": 1490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002512730258029355,
+      "loss": 0.2479,
+      "step": 1491
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002506362439453463,
+      "loss": 0.2466,
+      "step": 1492
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 0.2721,
+      "step": 1493
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002493642953393635,
+      "loss": 0.2712,
+      "step": 1494
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002487291313347397,
+      "loss": 0.2705,
+      "step": 1495
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002480945093562656,
+      "loss": 0.2615,
+      "step": 1496
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024746043077290917,
+      "loss": 0.2631,
+      "step": 1497
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002468268969524656,
+      "loss": 0.2565,
+      "step": 1498
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024619390926155534,
+      "loss": 0.2527,
+      "step": 1499
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024556146906562086,
+      "loss": 0.2461,
+      "step": 1500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002449295777289234,
+      "loss": 0.2554,
+      "step": 1501
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024429823661454057,
+      "loss": 0.2632,
+      "step": 1502
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024366744708436233,
+      "loss": 0.244,
+      "step": 1503
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00024303721049908971,
+      "loss": 0.2549,
+      "step": 1504
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002424075282182302,
+      "loss": 0.244,
+      "step": 1505
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024177840160009595,
+      "loss": 0.2737,
+      "step": 1506
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002411498320018005,
+      "loss": 0.2639,
+      "step": 1507
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00024052182077925584,
+      "loss": 0.2474,
+      "step": 1508
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002398943692871694,
+      "loss": 0.2435,
+      "step": 1509
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023926747887904082,
+      "loss": 0.2631,
+      "step": 1510
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023864115090716,
+      "loss": 0.2587,
+      "step": 1511
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002380153867226033,
+      "loss": 0.2609,
+      "step": 1512
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023739018767523096,
+      "loss": 0.2645,
+      "step": 1513
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023676555511368424,
+      "loss": 0.2473,
+      "step": 1514
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00023614149038538208,
+      "loss": 0.2396,
+      "step": 1515
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002355179948365189,
+      "loss": 0.26,
+      "step": 1516
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000234895069812061,
+      "loss": 0.2438,
+      "step": 1517
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023427271665574424,
+      "loss": 0.2532,
+      "step": 1518
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023365093671007077,
+      "loss": 0.2675,
+      "step": 1519
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023302973131630623,
+      "loss": 0.2321,
+      "step": 1520
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023240910181447717,
+      "loss": 0.2405,
+      "step": 1521
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023178904954336717,
+      "loss": 0.2688,
+      "step": 1522
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023116957584051528,
+      "loss": 0.2629,
+      "step": 1523
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00023055068204221225,
+      "loss": 0.2627,
+      "step": 1524
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000229932369483498,
+      "loss": 0.2684,
+      "step": 1525
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00022931463949815844,
+      "loss": 0.2772,
+      "step": 1526
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002286974934187232,
+      "loss": 0.2491,
+      "step": 1527
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022808093257646185,
+      "loss": 0.2335,
+      "step": 1528
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022746495830138153,
+      "loss": 0.2516,
+      "step": 1529
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022684957192222445,
+      "loss": 0.2493,
+      "step": 1530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002262347747664645,
+      "loss": 0.2537,
+      "step": 1531
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022562056816030447,
+      "loss": 0.2723,
+      "step": 1532
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002250069534286735,
+      "loss": 0.2457,
+      "step": 1533
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002243939318952234,
+      "loss": 0.251,
+      "step": 1534
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022378150488232685,
+      "loss": 0.2693,
+      "step": 1535
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000223169673711074,
+      "loss": 0.2471,
+      "step": 1536
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022255843970126954,
+      "loss": 0.2399,
+      "step": 1537
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00022194780417143025,
+      "loss": 0.2644,
+      "step": 1538
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022133776843878184,
+      "loss": 0.2588,
+      "step": 1539
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022072833381925588,
+      "loss": 0.2565,
+      "step": 1540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00022011950162748734,
+      "loss": 0.2773,
+      "step": 1541
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021951127317681185,
+      "loss": 0.2455,
+      "step": 1542
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002189036497792628,
+      "loss": 0.2522,
+      "step": 1543
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021829663274556806,
+      "loss": 0.2418,
+      "step": 1544
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002176902233851477,
+      "loss": 0.2715,
+      "step": 1545
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021708442300611113,
+      "loss": 0.2776,
+      "step": 1546
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00021647923291525352,
+      "loss": 0.2665,
+      "step": 1547
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000215874654418054,
+      "loss": 0.2611,
+      "step": 1548
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002152706888186724,
+      "loss": 0.2606,
+      "step": 1549
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021466733741994638,
+      "loss": 0.2648,
+      "step": 1550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002140646015233888,
+      "loss": 0.252,
+      "step": 1551
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021346248242918458,
+      "loss": 0.2566,
+      "step": 1552
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021286098143618793,
+      "loss": 0.2516,
+      "step": 1553
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021226009984192024,
+      "loss": 0.2875,
+      "step": 1554
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021165983894256646,
+      "loss": 0.2604,
+      "step": 1555
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021106020003297283,
+      "loss": 0.2616,
+      "step": 1556
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00021046118440664358,
+      "loss": 0.2723,
+      "step": 1557
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020986279335573888,
+      "loss": 0.2511,
+      "step": 1558
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020926502817107085,
+      "loss": 0.2505,
+      "step": 1559
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00020866789014210225,
+      "loss": 0.2618,
+      "step": 1560
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002080713805569427,
+      "loss": 0.2594,
+      "step": 1561
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002074755007023461,
+      "loss": 0.2451,
+      "step": 1562
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020688025186370814,
+      "loss": 0.2401,
+      "step": 1563
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020628563532506335,
+      "loss": 0.2545,
+      "step": 1564
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020569165236908194,
+      "loss": 0.2666,
+      "step": 1565
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020509830427706748,
+      "loss": 0.2668,
+      "step": 1566
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002045055923289544,
+      "loss": 0.2408,
+      "step": 1567
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020391351780330459,
+      "loss": 0.2638,
+      "step": 1568
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020332208197730502,
+      "loss": 0.2684,
+      "step": 1569
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020273128612676507,
+      "loss": 0.2671,
+      "step": 1570
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00020214113152611303,
+      "loss": 0.265,
+      "step": 1571
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020155161944839455,
+      "loss": 0.2334,
+      "step": 1572
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020096275116526896,
+      "loss": 0.2614,
+      "step": 1573
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00020037452794700696,
+      "loss": 0.2341,
+      "step": 1574
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019978695106248766,
+      "loss": 0.2566,
+      "step": 1575
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019920002177919622,
+      "loss": 0.2631,
+      "step": 1576
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019861374136322037,
+      "loss": 0.2482,
+      "step": 1577
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019802811107924812,
+      "loss": 0.2755,
+      "step": 1578
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001974431321905656,
+      "loss": 0.2482,
+      "step": 1579
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019685880595905343,
+      "loss": 0.2408,
+      "step": 1580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001962751336451844,
+      "loss": 0.2402,
+      "step": 1581
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00019569211650802076,
+      "loss": 0.2551,
+      "step": 1582
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019510975580521155,
+      "loss": 0.2478,
+      "step": 1583
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001945280527929894,
+      "loss": 0.2739,
+      "step": 1584
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019394700872616856,
+      "loss": 0.2603,
+      "step": 1585
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019336662485814178,
+      "loss": 0.2626,
+      "step": 1586
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019278690244087766,
+      "loss": 0.2667,
+      "step": 1587
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001922078427249181,
+      "loss": 0.2563,
+      "step": 1588
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019162944695937513,
+      "loss": 0.2485,
+      "step": 1589
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00019105171639192852,
+      "loss": 0.2473,
+      "step": 1590
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001904746522688236,
+      "loss": 0.2469,
+      "step": 1591
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00018989825583486764,
+      "loss": 0.2405,
+      "step": 1592
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00018932252833342783,
+      "loss": 0.2512,
+      "step": 1593
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018874747100642843,
+      "loss": 0.2626,
+      "step": 1594
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018817308509434794,
+      "loss": 0.2485,
+      "step": 1595
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018759937183621635,
+      "loss": 0.2756,
+      "step": 1596
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018702633246961282,
+      "loss": 0.2239,
+      "step": 1597
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001864539682306629,
+      "loss": 0.2773,
+      "step": 1598
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001858822803540357,
+      "loss": 0.2467,
+      "step": 1599
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018531127007294156,
+      "loss": 0.2534,
+      "step": 1600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018474093861912854,
+      "loss": 0.2494,
+      "step": 1601
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018417128722288112,
+      "loss": 0.2539,
+      "step": 1602
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018360231711301618,
+      "loss": 0.2482,
+      "step": 1603
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00018303402951688126,
+      "loss": 0.2732,
+      "step": 1604
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018246642566035177,
+      "loss": 0.2634,
+      "step": 1605
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001818995067678279,
+      "loss": 0.2787,
+      "step": 1606
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001813332740622327,
+      "loss": 0.2671,
+      "step": 1607
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001807677287650083,
+      "loss": 0.2614,
+      "step": 1608
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018020287209611464,
+      "loss": 0.2611,
+      "step": 1609
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017963870527402598,
+      "loss": 0.2427,
+      "step": 1610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017907522951572847,
+      "loss": 0.2588,
+      "step": 1611
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017851244603671768,
+      "loss": 0.2538,
+      "step": 1612
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017795035605099548,
+      "loss": 0.2595,
+      "step": 1613
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001773889607710682,
+      "loss": 0.2501,
+      "step": 1614
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017682826140794317,
+      "loss": 0.2592,
+      "step": 1615
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017626825917112693,
+      "loss": 0.2567,
+      "step": 1616
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017570895526862202,
+      "loss": 0.271,
+      "step": 1617
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017515035090692467,
+      "loss": 0.2366,
+      "step": 1618
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001745924472910221,
+      "loss": 0.246,
+      "step": 1619
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001740352456243896,
+      "loss": 0.2738,
+      "step": 1620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001734787471089887,
+      "loss": 0.2582,
+      "step": 1621
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000172922952945264,
+      "loss": 0.2566,
+      "step": 1622
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017236786433214057,
+      "loss": 0.2625,
+      "step": 1623
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017181348246702184,
+      "loss": 0.2351,
+      "step": 1624
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017125980854578615,
+      "loss": 0.245,
+      "step": 1625
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017070684376278532,
+      "loss": 0.2738,
+      "step": 1626
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017015458931084082,
+      "loss": 0.2667,
+      "step": 1627
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001696030463812424,
+      "loss": 0.2692,
+      "step": 1628
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016905221616374467,
+      "loss": 0.2639,
+      "step": 1629
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016850209984656494,
+      "loss": 0.2527,
+      "step": 1630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001679526986163804,
+      "loss": 0.2584,
+      "step": 1631
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016740401365832608,
+      "loss": 0.2577,
+      "step": 1632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016685604615599114,
+      "loss": 0.2609,
+      "step": 1633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001663087972914178,
+      "loss": 0.2637,
+      "step": 1634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001657622682450979,
+      "loss": 0.2539,
+      "step": 1635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016521646019597048,
+      "loss": 0.2684,
+      "step": 1636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00016467137432141904,
+      "loss": 0.271,
+      "step": 1637
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016412701179726986,
+      "loss": 0.2574,
+      "step": 1638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001635833737977881,
+      "loss": 0.2639,
+      "step": 1639
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001630404614956766,
+      "loss": 0.2753,
+      "step": 1640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016249827606207252,
+      "loss": 0.2582,
+      "step": 1641
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016195681866654516,
+      "loss": 0.2584,
+      "step": 1642
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016141609047709327,
+      "loss": 0.2539,
+      "step": 1643
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016087609266014286,
+      "loss": 0.2422,
+      "step": 1644
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00016033682638054377,
+      "loss": 0.2482,
+      "step": 1645
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015979829280156842,
+      "loss": 0.281,
+      "step": 1646
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001592604930849086,
+      "loss": 0.2561,
+      "step": 1647
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015872342839067305,
+      "loss": 0.2374,
+      "step": 1648
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001581870998773845,
+      "loss": 0.2341,
+      "step": 1649
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001576515087019783,
+      "loss": 0.2393,
+      "step": 1650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015711665601979906,
+      "loss": 0.2522,
+      "step": 1651
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000156582542984598,
+      "loss": 0.2614,
+      "step": 1652
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015604917074853126,
+      "loss": 0.2576,
+      "step": 1653
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015551654046215668,
+      "loss": 0.2531,
+      "step": 1654
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001549846532744319,
+      "loss": 0.2421,
+      "step": 1655
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015445351033271148,
+      "loss": 0.254,
+      "step": 1656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015392311278274428,
+      "loss": 0.2462,
+      "step": 1657
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015339346176867146,
+      "loss": 0.2621,
+      "step": 1658
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015286455843302395,
+      "loss": 0.2601,
+      "step": 1659
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015233640391671972,
+      "loss": 0.233,
+      "step": 1660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015180899935906113,
+      "loss": 0.2632,
+      "step": 1661
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015128234589773315,
+      "loss": 0.2565,
+      "step": 1662
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001507564446688006,
+      "loss": 0.2534,
+      "step": 1663
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015023129680670515,
+      "loss": 0.2528,
+      "step": 1664
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001497069034442638,
+      "loss": 0.2555,
+      "step": 1665
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014918326571266582,
+      "loss": 0.2544,
+      "step": 1666
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014866038474147054,
+      "loss": 0.2296,
+      "step": 1667
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014813826165860488,
+      "loss": 0.2366,
+      "step": 1668
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014761689759036056,
+      "loss": 0.2554,
+      "step": 1669
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014709629366139242,
+      "loss": 0.2565,
+      "step": 1670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014657645099471544,
+      "loss": 0.2421,
+      "step": 1671
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014605737071170255,
+      "loss": 0.245,
+      "step": 1672
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014553905393208173,
+      "loss": 0.2609,
+      "step": 1673
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014502150177393426,
+      "loss": 0.2711,
+      "step": 1674
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014450471535369224,
+      "loss": 0.2473,
+      "step": 1675
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001439886957861355,
+      "loss": 0.2457,
+      "step": 1676
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001434734441843899,
+      "loss": 0.241,
+      "step": 1677
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014295896165992473,
+      "loss": 0.2617,
+      "step": 1678
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014244524932255026,
+      "loss": 0.2435,
+      "step": 1679
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014193230828041532,
+      "loss": 0.2609,
+      "step": 1680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001414201396400051,
+      "loss": 0.2578,
+      "step": 1681
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001409087445061381,
+      "loss": 0.2706,
+      "step": 1682
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014039812398196488,
+      "loss": 0.2426,
+      "step": 1683
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001398882791689649,
+      "loss": 0.25,
+      "step": 1684
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001393792111669439,
+      "loss": 0.2456,
+      "step": 1685
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013887092107403242,
+      "loss": 0.2759,
+      "step": 1686
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013836340998668283,
+      "loss": 0.2485,
+      "step": 1687
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013785667899966685,
+      "loss": 0.2643,
+      "step": 1688
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013735072920607362,
+      "loss": 0.2443,
+      "step": 1689
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013684556169730704,
+      "loss": 0.2516,
+      "step": 1690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013634117756308368,
+      "loss": 0.241,
+      "step": 1691
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013583757789143014,
+      "loss": 0.2371,
+      "step": 1692
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013533476376868087,
+      "loss": 0.2552,
+      "step": 1693
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013483273627947555,
+      "loss": 0.2483,
+      "step": 1694
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001343314965067573,
+      "loss": 0.2622,
+      "step": 1695
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013383104553177,
+      "loss": 0.2504,
+      "step": 1696
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013333138443405562,
+      "loss": 0.2405,
+      "step": 1697
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001328325142914527,
+      "loss": 0.2639,
+      "step": 1698
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013233443618009338,
+      "loss": 0.2432,
+      "step": 1699
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013183715117440144,
+      "loss": 0.2677,
+      "step": 1700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013134066034708935,
+      "loss": 0.2551,
+      "step": 1701
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013084496476915698,
+      "loss": 0.2534,
+      "step": 1702
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013035006550988842,
+      "loss": 0.2405,
+      "step": 1703
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001298559636368502,
+      "loss": 0.2478,
+      "step": 1704
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001293626602158887,
+      "loss": 0.2502,
+      "step": 1705
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012887015631112765,
+      "loss": 0.2653,
+      "step": 1706
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012837845298496647,
+      "loss": 0.2439,
+      "step": 1707
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012788755129807767,
+      "loss": 0.2252,
+      "step": 1708
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012739745230940403,
+      "loss": 0.2561,
+      "step": 1709
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012690815707615727,
+      "loss": 0.2444,
+      "step": 1710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012641966665381517,
+      "loss": 0.2545,
+      "step": 1711
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012593198209611951,
+      "loss": 0.2461,
+      "step": 1712
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001254451044550733,
+      "loss": 0.2467,
+      "step": 1713
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012495903478093929,
+      "loss": 0.2506,
+      "step": 1714
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00012447377412223737,
+      "loss": 0.2542,
+      "step": 1715
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001239893235257421,
+      "loss": 0.2588,
+      "step": 1716
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012350568403648088,
+      "loss": 0.26,
+      "step": 1717
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012302285669773082,
+      "loss": 0.2592,
+      "step": 1718
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012254084255101773,
+      "loss": 0.2499,
+      "step": 1719
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012205964263611325,
+      "loss": 0.2507,
+      "step": 1720
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012157925799103203,
+      "loss": 0.2607,
+      "step": 1721
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012109968965203066,
+      "loss": 0.2677,
+      "step": 1722
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012062093865360457,
+      "loss": 0.2347,
+      "step": 1723
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00012014300602848627,
+      "loss": 0.2522,
+      "step": 1724
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00011966589280764257,
+      "loss": 0.25,
+      "step": 1725
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00011918960002027306,
+      "loss": 0.2358,
+      "step": 1726
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011871412869380732,
+      "loss": 0.2384,
+      "step": 1727
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011823947985390316,
+      "loss": 0.2688,
+      "step": 1728
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011776565452444388,
+      "loss": 0.2401,
+      "step": 1729
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011729265372753672,
+      "loss": 0.2388,
+      "step": 1730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001168204784835098,
+      "loss": 0.257,
+      "step": 1731
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011634912981091094,
+      "loss": 0.2582,
+      "step": 1732
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011587860872650435,
+      "loss": 0.2629,
+      "step": 1733
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011540891624526956,
+      "loss": 0.2489,
+      "step": 1734
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011494005338039836,
+      "loss": 0.2371,
+      "step": 1735
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011447202114329325,
+      "loss": 0.25,
+      "step": 1736
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00011400482054356448,
+      "loss": 0.2491,
+      "step": 1737
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011353845258902867,
+      "loss": 0.245,
+      "step": 1738
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011307291828570638,
+      "loss": 0.2443,
+      "step": 1739
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011260821863781956,
+      "loss": 0.2473,
+      "step": 1740
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011214435464779005,
+      "loss": 0.2396,
+      "step": 1741
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00011168132731623688,
+      "loss": 0.2754,
+      "step": 1742
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001112191376419741,
+      "loss": 0.2482,
+      "step": 1743
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001107577866220092,
+      "loss": 0.2522,
+      "step": 1744
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001102972752515402,
+      "loss": 0.2494,
+      "step": 1745
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010983760452395414,
+      "loss": 0.255,
+      "step": 1746
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010937877543082464,
+      "loss": 0.2511,
+      "step": 1747
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010892078896190971,
+      "loss": 0.2549,
+      "step": 1748
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010846364610514991,
+      "loss": 0.2574,
+      "step": 1749
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010800734784666555,
+      "loss": 0.2528,
+      "step": 1750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010755189517075553,
+      "loss": 0.2545,
+      "step": 1751
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010709728905989446,
+      "loss": 0.2382,
+      "step": 1752
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010664353049473086,
+      "loss": 0.2366,
+      "step": 1753
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010619062045408501,
+      "loss": 0.2454,
+      "step": 1754
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010573855991494657,
+      "loss": 0.2397,
+      "step": 1755
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010528734985247317,
+      "loss": 0.2782,
+      "step": 1756
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010483699123998718,
+      "loss": 0.2336,
+      "step": 1757
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010438748504897495,
+      "loss": 0.2397,
+      "step": 1758
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010393883224908357,
+      "loss": 0.2545,
+      "step": 1759
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010349103380811953,
+      "loss": 0.2357,
+      "step": 1760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010304409069204629,
+      "loss": 0.2623,
+      "step": 1761
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010259800386498202,
+      "loss": 0.2491,
+      "step": 1762
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010215277428919806,
+      "loss": 0.2415,
+      "step": 1763
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001017084029251164,
+      "loss": 0.252,
+      "step": 1764
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010126489073130779,
+      "loss": 0.2523,
+      "step": 1765
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010082223866448953,
+      "loss": 0.2423,
+      "step": 1766
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010038044767952381,
+      "loss": 0.2477,
+      "step": 1767
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.993951872941492e-05,
+      "loss": 0.262,
+      "step": 1768
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.949945276530781e-05,
+      "loss": 0.2454,
+      "step": 1769
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.906025073648583e-05,
+      "loss": 0.2423,
+      "step": 1770
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.862191359036882e-05,
+      "loss": 0.2502,
+      "step": 1771
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.818444227251089e-05,
+      "loss": 0.2424,
+      "step": 1772
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.774783772659851e-05,
+      "loss": 0.2368,
+      "step": 1773
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.731210089444804e-05,
+      "loss": 0.2273,
+      "step": 1774
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.687723271600445e-05,
+      "loss": 0.2386,
+      "step": 1775
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.644323412933876e-05,
+      "loss": 0.2365,
+      "step": 1776
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.60101060706462e-05,
+      "loss": 0.2458,
+      "step": 1777
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.557784947424414e-05,
+      "loss": 0.2423,
+      "step": 1778
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.51464652725702e-05,
+      "loss": 0.2614,
+      "step": 1779
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.471595439617986e-05,
+      "loss": 0.2371,
+      "step": 1780
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.42863177737447e-05,
+      "loss": 0.2363,
+      "step": 1781
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.385755633205073e-05,
+      "loss": 0.2462,
+      "step": 1782
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.342967099599587e-05,
+      "loss": 0.2383,
+      "step": 1783
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.30026626885882e-05,
+      "loss": 0.2378,
+      "step": 1784
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.257653233094399e-05,
+      "loss": 0.2454,
+      "step": 1785
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.215128084228563e-05,
+      "loss": 0.2347,
+      "step": 1786
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.172690913993937e-05,
+      "loss": 0.2509,
+      "step": 1787
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.130341813933407e-05,
+      "loss": 0.241,
+      "step": 1788
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.088080875399862e-05,
+      "loss": 0.2673,
+      "step": 1789
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.045908189556007e-05,
+      "loss": 0.2568,
+      "step": 1790
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.003823847374204e-05,
+      "loss": 0.2637,
+      "step": 1791
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.961827939636197e-05,
+      "loss": 0.2562,
+      "step": 1792
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.919920556932987e-05,
+      "loss": 0.2588,
+      "step": 1793
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.878101789664622e-05,
+      "loss": 0.2606,
+      "step": 1794
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.836371728039988e-05,
+      "loss": 0.2312,
+      "step": 1795
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.794730462076633e-05,
+      "loss": 0.2622,
+      "step": 1796
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.753178081600538e-05,
+      "loss": 0.2434,
+      "step": 1797
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.711714676245974e-05,
+      "loss": 0.2448,
+      "step": 1798
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.67034033545524e-05,
+      "loss": 0.2435,
+      "step": 1799
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.629055148478543e-05,
+      "loss": 0.2454,
+      "step": 1800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.58785920437376e-05,
+      "loss": 0.2438,
+      "step": 1801
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.546752592006268e-05,
+      "loss": 0.2466,
+      "step": 1802
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.505735400048747e-05,
+      "loss": 0.2576,
+      "step": 1803
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.46480771698096e-05,
+      "loss": 0.245,
+      "step": 1804
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.423969631089595e-05,
+      "loss": 0.2477,
+      "step": 1805
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.38322123046808e-05,
+      "loss": 0.2521,
+      "step": 1806
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.342562603016379e-05,
+      "loss": 0.233,
+      "step": 1807
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.301993836440785e-05,
+      "loss": 0.2401,
+      "step": 1808
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.261515018253773e-05,
+      "loss": 0.2455,
+      "step": 1809
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.221126235773779e-05,
+      "loss": 0.2448,
+      "step": 1810
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.180827576124988e-05,
+      "loss": 0.2625,
+      "step": 1811
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.140619126237226e-05,
+      "loss": 0.2451,
+      "step": 1812
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.100500972845687e-05,
+      "loss": 0.2223,
+      "step": 1813
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.060473202490809e-05,
+      "loss": 0.2365,
+      "step": 1814
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.02053590151805e-05,
+      "loss": 0.259,
+      "step": 1815
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.980689156077719e-05,
+      "loss": 0.2506,
+      "step": 1816
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.940933052124771e-05,
+      "loss": 0.2401,
+      "step": 1817
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.901267675418627e-05,
+      "loss": 0.2454,
+      "step": 1818
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.861693111523021e-05,
+      "loss": 0.2538,
+      "step": 1819
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.822209445805795e-05,
+      "loss": 0.2512,
+      "step": 1820
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.782816763438677e-05,
+      "loss": 0.2454,
+      "step": 1821
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.743515149397185e-05,
+      "loss": 0.2491,
+      "step": 1822
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.704304688460317e-05,
+      "loss": 0.2568,
+      "step": 1823
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.6651854652105e-05,
+      "loss": 0.244,
+      "step": 1824
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.626157564033331e-05,
+      "loss": 0.2286,
+      "step": 1825
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.587221069117411e-05,
+      "loss": 0.2401,
+      "step": 1826
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.548376064454165e-05,
+      "loss": 0.248,
+      "step": 1827
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.509622633837671e-05,
+      "loss": 0.225,
+      "step": 1828
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.470960860864446e-05,
+      "loss": 0.2334,
+      "step": 1829
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.432390828933295e-05,
+      "loss": 0.2334,
+      "step": 1830
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.393912621245142e-05,
+      "loss": 0.233,
+      "step": 1831
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.355526320802824e-05,
+      "loss": 0.233,
+      "step": 1832
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.317232010410918e-05,
+      "loss": 0.212,
+      "step": 1833
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.279029772675571e-05,
+      "loss": 0.251,
+      "step": 1834
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.240919690004322e-05,
+      "loss": 0.2501,
+      "step": 1835
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.202901844605891e-05,
+      "loss": 0.2573,
+      "step": 1836
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.164976318490058e-05,
+      "loss": 0.2419,
+      "step": 1837
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.127143193467444e-05,
+      "loss": 0.266,
+      "step": 1838
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.089402551149365e-05,
+      "loss": 0.2281,
+      "step": 1839
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.051754472947625e-05,
+      "loss": 0.2329,
+      "step": 1840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.014199040074353e-05,
+      "loss": 0.2605,
+      "step": 1841
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.976736333541816e-05,
+      "loss": 0.2433,
+      "step": 1842
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.939366434162287e-05,
+      "loss": 0.2394,
+      "step": 1843
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.90208942254782e-05,
+      "loss": 0.2593,
+      "step": 1844
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.864905379110114e-05,
+      "loss": 0.2408,
+      "step": 1845
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.82781438406031e-05,
+      "loss": 0.2546,
+      "step": 1846
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.790816517408849e-05,
+      "loss": 0.2438,
+      "step": 1847
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.753911858965233e-05,
+      "loss": 0.2439,
+      "step": 1848
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.717100488337952e-05,
+      "loss": 0.2633,
+      "step": 1849
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.680382484934244e-05,
+      "loss": 0.2438,
+      "step": 1850
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.643757927959943e-05,
+      "loss": 0.2452,
+      "step": 1851
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.607226896419305e-05,
+      "loss": 0.244,
+      "step": 1852
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.570789469114823e-05,
+      "loss": 0.225,
+      "step": 1853
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.534445724647109e-05,
+      "loss": 0.2433,
+      "step": 1854
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.498195741414637e-05,
+      "loss": 0.2391,
+      "step": 1855
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.462039597613673e-05,
+      "loss": 0.239,
+      "step": 1856
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.425977371238029e-05,
+      "loss": 0.2329,
+      "step": 1857
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.39000914007894e-05,
+      "loss": 0.2528,
+      "step": 1858
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.35413498172489e-05,
+      "loss": 0.2269,
+      "step": 1859
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.318354973561385e-05,
+      "loss": 0.2539,
+      "step": 1860
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.282669192770896e-05,
+      "loss": 0.2474,
+      "step": 1861
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.247077716332594e-05,
+      "loss": 0.2477,
+      "step": 1862
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.211580621022245e-05,
+      "loss": 0.2433,
+      "step": 1863
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.176177983412012e-05,
+      "loss": 0.2386,
+      "step": 1864
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.140869879870287e-05,
+      "loss": 0.2397,
+      "step": 1865
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.105656386561576e-05,
+      "loss": 0.2588,
+      "step": 1866
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.070537579446239e-05,
+      "loss": 0.2434,
+      "step": 1867
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.035513534280435e-05,
+      "loss": 0.2551,
+      "step": 1868
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.000584326615899e-05,
+      "loss": 0.256,
+      "step": 1869
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.965750031799771e-05,
+      "loss": 0.2592,
+      "step": 1870
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.9310107249744784e-05,
+      "loss": 0.2472,
+      "step": 1871
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.896366481077503e-05,
+      "loss": 0.2349,
+      "step": 1872
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.8618173748413105e-05,
+      "loss": 0.2571,
+      "step": 1873
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.827363480793108e-05,
+      "loss": 0.2522,
+      "step": 1874
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.793004873254731e-05,
+      "loss": 0.2418,
+      "step": 1875
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.758741626342479e-05,
+      "loss": 0.2416,
+      "step": 1876
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.72457381396691e-05,
+      "loss": 0.2358,
+      "step": 1877
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6905015098327575e-05,
+      "loss": 0.2555,
+      "step": 1878
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.6565247874386996e-05,
+      "loss": 0.2461,
+      "step": 1879
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.622643720077247e-05,
+      "loss": 0.2472,
+      "step": 1880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.588858380834572e-05,
+      "loss": 0.2714,
+      "step": 1881
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.55516884259033e-05,
+      "loss": 0.2391,
+      "step": 1882
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.521575178017546e-05,
+      "loss": 0.2662,
+      "step": 1883
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.4880774595824246e-05,
+      "loss": 0.2374,
+      "step": 1884
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.454675759544175e-05,
+      "loss": 0.2441,
+      "step": 1885
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.421370149954907e-05,
+      "loss": 0.234,
+      "step": 1886
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.388160702659445e-05,
+      "loss": 0.2444,
+      "step": 1887
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.355047489295195e-05,
+      "loss": 0.2561,
+      "step": 1888
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.3220305812919315e-05,
+      "loss": 0.2515,
+      "step": 1889
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.289110049871726e-05,
+      "loss": 0.2418,
+      "step": 1890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.256285966048718e-05,
+      "loss": 0.2629,
+      "step": 1891
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.223558400629025e-05,
+      "loss": 0.2446,
+      "step": 1892
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.190927424210545e-05,
+      "loss": 0.2454,
+      "step": 1893
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.158393107182835e-05,
+      "loss": 0.2476,
+      "step": 1894
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.125955519726921e-05,
+      "loss": 0.2386,
+      "step": 1895
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.0936147318151996e-05,
+      "loss": 0.2301,
+      "step": 1896
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.061370813211219e-05,
+      "loss": 0.2509,
+      "step": 1897
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.0292238334695986e-05,
+      "loss": 0.2368,
+      "step": 1898
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.997173861935833e-05,
+      "loss": 0.2571,
+      "step": 1899
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.965220967746181e-05,
+      "loss": 0.2296,
+      "step": 1900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.93336521982744e-05,
+      "loss": 0.2555,
+      "step": 1901
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.901606686896898e-05,
+      "loss": 0.2554,
+      "step": 1902
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.869945437462126e-05,
+      "loss": 0.2744,
+      "step": 1903
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.838381539820818e-05,
+      "loss": 0.2606,
+      "step": 1904
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.8069150620606906e-05,
+      "loss": 0.2214,
+      "step": 1905
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.775546072059311e-05,
+      "loss": 0.248,
+      "step": 1906
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7442746374839365e-05,
+      "loss": 0.2549,
+      "step": 1907
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.713100825791411e-05,
+      "loss": 0.2415,
+      "step": 1908
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.68202470422795e-05,
+      "loss": 0.2284,
+      "step": 1909
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.651046339829079e-05,
+      "loss": 0.2532,
+      "step": 1910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.620165799419418e-05,
+      "loss": 0.2394,
+      "step": 1911
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.5893831496126024e-05,
+      "loss": 0.2493,
+      "step": 1912
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.558698456811056e-05,
+      "loss": 0.2402,
+      "step": 1913
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.528111787205935e-05,
+      "loss": 0.2469,
+      "step": 1914
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.497623206776935e-05,
+      "loss": 0.2551,
+      "step": 1915
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.467232781292146e-05,
+      "loss": 0.2274,
+      "step": 1916
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.436940576307941e-05,
+      "loss": 0.2443,
+      "step": 1917
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.406746657168809e-05,
+      "loss": 0.2261,
+      "step": 1918
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.376651089007227e-05,
+      "loss": 0.2384,
+      "step": 1919
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.346653936743533e-05,
+      "loss": 0.2428,
+      "step": 1920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.316755265085715e-05,
+      "loss": 0.2484,
+      "step": 1921
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.2869551385293807e-05,
+      "loss": 0.2584,
+      "step": 1922
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.257253621357532e-05,
+      "loss": 0.252,
+      "step": 1923
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.227650777640474e-05,
+      "loss": 0.2629,
+      "step": 1924
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.198146671235642e-05,
+      "loss": 0.2135,
+      "step": 1925
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.168741365787487e-05,
+      "loss": 0.2561,
+      "step": 1926
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.139434924727359e-05,
+      "loss": 0.2314,
+      "step": 1927
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.110227411273287e-05,
+      "loss": 0.2343,
+      "step": 1928
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.081118888429946e-05,
+      "loss": 0.264,
+      "step": 1929
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.05210941898847e-05,
+      "loss": 0.282,
+      "step": 1930
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.0231990655263004e-05,
+      "loss": 0.2386,
+      "step": 1931
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.994387890407092e-05,
+      "loss": 0.2308,
+      "step": 1932
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.965675955780551e-05,
+      "loss": 0.2411,
+      "step": 1933
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.937063323582296e-05,
+      "loss": 0.243,
+      "step": 1934
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.908550055533755e-05,
+      "loss": 0.2527,
+      "step": 1935
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.88013621314201e-05,
+      "loss": 0.2463,
+      "step": 1936
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.85182185769965e-05,
+      "loss": 0.2716,
+      "step": 1937
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.823607050284678e-05,
+      "loss": 0.2396,
+      "step": 1938
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7954918517603634e-05,
+      "loss": 0.2421,
+      "step": 1939
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7674763227750706e-05,
+      "loss": 0.2334,
+      "step": 1940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.739560523762192e-05,
+      "loss": 0.2523,
+      "step": 1941
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.711744514939991e-05,
+      "loss": 0.241,
+      "step": 1942
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.684028356311453e-05,
+      "loss": 0.2307,
+      "step": 1943
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.656412107664181e-05,
+      "loss": 0.2351,
+      "step": 1944
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6288958285702725e-05,
+      "loss": 0.2531,
+      "step": 1945
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.601479578386135e-05,
+      "loss": 0.2439,
+      "step": 1946
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.57416341625244e-05,
+      "loss": 0.24,
+      "step": 1947
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.546947401093953e-05,
+      "loss": 0.2234,
+      "step": 1948
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.519831591619377e-05,
+      "loss": 0.2401,
+      "step": 1949
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.492816046321284e-05,
+      "loss": 0.2651,
+      "step": 1950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.465900823475959e-05,
+      "loss": 0.2579,
+      "step": 1951
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.439085981143286e-05,
+      "loss": 0.2528,
+      "step": 1952
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4123715771665786e-05,
+      "loss": 0.2407,
+      "step": 1953
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3857576691725343e-05,
+      "loss": 0.252,
+      "step": 1954
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.359244314571036e-05,
+      "loss": 0.2404,
+      "step": 1955
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.332831570555073e-05,
+      "loss": 0.2463,
+      "step": 1956
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3065194941006184e-05,
+      "loss": 0.254,
+      "step": 1957
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2803081419664485e-05,
+      "loss": 0.2336,
+      "step": 1958
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.254197570694101e-05,
+      "loss": 0.2341,
+      "step": 1959
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.228187836607704e-05,
+      "loss": 0.2349,
+      "step": 1960
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.202278995813862e-05,
+      "loss": 0.2395,
+      "step": 1961
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.176471104201545e-05,
+      "loss": 0.2614,
+      "step": 1962
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.150764217441954e-05,
+      "loss": 0.2505,
+      "step": 1963
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.125158390988425e-05,
+      "loss": 0.2277,
+      "step": 1964
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0996536800762586e-05,
+      "loss": 0.2422,
+      "step": 1965
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.074250139722678e-05,
+      "loss": 0.2401,
+      "step": 1966
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0489478247266332e-05,
+      "loss": 0.2477,
+      "step": 1967
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0237467896687353e-05,
+      "loss": 0.2417,
+      "step": 1968
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9986470889111274e-05,
+      "loss": 0.2645,
+      "step": 1969
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9736487765973263e-05,
+      "loss": 0.2545,
+      "step": 1970
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.948751906652175e-05,
+      "loss": 0.23,
+      "step": 1971
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9239565327816907e-05,
+      "loss": 0.2521,
+      "step": 1972
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8992627084729107e-05,
+      "loss": 0.2457,
+      "step": 1973
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.874670486993858e-05,
+      "loss": 0.2164,
+      "step": 1974
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8501799213933644e-05,
+      "loss": 0.2419,
+      "step": 1975
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.825791064500993e-05,
+      "loss": 0.2462,
+      "step": 1976
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.8015039689268773e-05,
+      "loss": 0.2466,
+      "step": 1977
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7773186870616586e-05,
+      "loss": 0.2206,
+      "step": 1978
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7532352710763486e-05,
+      "loss": 0.2346,
+      "step": 1979
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7292537729222243e-05,
+      "loss": 0.2566,
+      "step": 1980
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.705374244330705e-05,
+      "loss": 0.2384,
+      "step": 1981
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6815967368132467e-05,
+      "loss": 0.2352,
+      "step": 1982
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6579213016612315e-05,
+      "loss": 0.2504,
+      "step": 1983
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6343479899458732e-05,
+      "loss": 0.234,
+      "step": 1984
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.6108768525180626e-05,
+      "loss": 0.2274,
+      "step": 1985
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5875079400083044e-05,
+      "loss": 0.2375,
+      "step": 1986
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5642413028265865e-05,
+      "loss": 0.2366,
+      "step": 1987
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.5410769911622942e-05,
+      "loss": 0.2427,
+      "step": 1988
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.518015054984041e-05,
+      "loss": 0.2373,
+      "step": 1989
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4950555440396317e-05,
+      "loss": 0.2373,
+      "step": 1990
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4721985078559216e-05,
+      "loss": 0.2319,
+      "step": 1991
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4494439957387094e-05,
+      "loss": 0.2485,
+      "step": 1992
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4267920567726364e-05,
+      "loss": 0.2527,
+      "step": 1993
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4042427398210943e-05,
+      "loss": 0.2606,
+      "step": 1994
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3817960935260673e-05,
+      "loss": 0.2471,
+      "step": 1995
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.359452166308107e-05,
+      "loss": 0.2413,
+      "step": 1996
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.337211006366147e-05,
+      "loss": 0.2747,
+      "step": 1997
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.3150726616774697e-05,
+      "loss": 0.2461,
+      "step": 1998
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2930371799975592e-05,
+      "loss": 0.2574,
+      "step": 1999
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2711046088600028e-05,
+      "loss": 0.2456,
+      "step": 2000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2492749955764168e-05,
+      "loss": 0.2416,
+      "step": 2001
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.2275483872362835e-05,
+      "loss": 0.2664,
+      "step": 2002
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.205924830706929e-05,
+      "loss": 0.2377,
+      "step": 2003
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.184404372633364e-05,
+      "loss": 0.2625,
+      "step": 2004
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.162987059438204e-05,
+      "loss": 0.2332,
+      "step": 2005
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.141672937321576e-05,
+      "loss": 0.2317,
+      "step": 2006
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1204620522609853e-05,
+      "loss": 0.2386,
+      "step": 2007
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0993544500112706e-05,
+      "loss": 0.2463,
+      "step": 2008
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0783501761044434e-05,
+      "loss": 0.2402,
+      "step": 2009
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.057449275849649e-05,
+      "loss": 0.2434,
+      "step": 2010
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0366517943330275e-05,
+      "loss": 0.2297,
+      "step": 2011
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0159577764176428e-05,
+      "loss": 0.2311,
+      "step": 2012
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9953672667433585e-05,
+      "loss": 0.2161,
+      "step": 2013
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.974880309726762e-05,
+      "loss": 0.2599,
+      "step": 2014
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.954496949561063e-05,
+      "loss": 0.2531,
+      "step": 2015
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.934217230216001e-05,
+      "loss": 0.2455,
+      "step": 2016
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9140411954377433e-05,
+      "loss": 0.2766,
+      "step": 2017
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.893968888748804e-05,
+      "loss": 0.2367,
+      "step": 2018
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8740003534479357e-05,
+      "loss": 0.262,
+      "step": 2019
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8541356326100432e-05,
+      "loss": 0.2396,
+      "step": 2020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8343747690860723e-05,
+      "loss": 0.2312,
+      "step": 2021
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8147178055029577e-05,
+      "loss": 0.2335,
+      "step": 2022
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.795164784263503e-05,
+      "loss": 0.2489,
+      "step": 2023
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7757157475462916e-05,
+      "loss": 0.2373,
+      "step": 2024
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7563707373055904e-05,
+      "loss": 0.2511,
+      "step": 2025
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.737129795271275e-05,
+      "loss": 0.2476,
+      "step": 2026
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.7179929629487223e-05,
+      "loss": 0.2369,
+      "step": 2027
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6989602816187442e-05,
+      "loss": 0.2484,
+      "step": 2028
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6800317923374718e-05,
+      "loss": 0.2419,
+      "step": 2029
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6612075359362877e-05,
+      "loss": 0.2349,
+      "step": 2030
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.642487553021732e-05,
+      "loss": 0.2499,
+      "step": 2031
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6238718839753973e-05,
+      "loss": 0.2297,
+      "step": 2032
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6053605689538675e-05,
+      "loss": 0.2212,
+      "step": 2033
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5869536478886228e-05,
+      "loss": 0.225,
+      "step": 2034
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5686511604859455e-05,
+      "loss": 0.2303,
+      "step": 2035
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.550453146226838e-05,
+      "loss": 0.2313,
+      "step": 2036
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5323596443669484e-05,
+      "loss": 0.2665,
+      "step": 2037
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5143706939364843e-05,
+      "loss": 0.2435,
+      "step": 2038
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.496486333740088e-05,
+      "loss": 0.2596,
+      "step": 2039
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4787066023568107e-05,
+      "loss": 0.2313,
+      "step": 2040
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4610315381400175e-05,
+      "loss": 0.2474,
+      "step": 2041
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.443461179217259e-05,
+      "loss": 0.2438,
+      "step": 2042
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.425995563490251e-05,
+      "loss": 0.2506,
+      "step": 2043
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.40863472863475e-05,
+      "loss": 0.2317,
+      "step": 2044
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.3913787121004718e-05,
+      "loss": 0.2402,
+      "step": 2045
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.3742275511110569e-05,
+      "loss": 0.236,
+      "step": 2046
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3571812826639329e-05,
+      "loss": 0.2522,
+      "step": 2047
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3402399435302693e-05,
+      "loss": 0.2357,
+      "step": 2048
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3234035702548886e-05,
+      "loss": 0.2347,
+      "step": 2049
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3066721991561892e-05,
+      "loss": 0.249,
+      "step": 2050
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2900458663260506e-05,
+      "loss": 0.2316,
+      "step": 2051
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2735246076297835e-05,
+      "loss": 0.2438,
+      "step": 2052
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2571084587060467e-05,
+      "loss": 0.2292,
+      "step": 2053
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.240797454966741e-05,
+      "loss": 0.2477,
+      "step": 2054
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2245916315969774e-05,
+      "loss": 0.2361,
+      "step": 2055
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2084910235549584e-05,
+      "loss": 0.2573,
+      "step": 2056
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1924956655719355e-05,
+      "loss": 0.2269,
+      "step": 2057
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.176605592152108e-05,
+      "loss": 0.2412,
+      "step": 2058
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1608208375725793e-05,
+      "loss": 0.2262,
+      "step": 2059
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1451414358832513e-05,
+      "loss": 0.2526,
+      "step": 2060
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1295674209067742e-05,
+      "loss": 0.2302,
+      "step": 2061
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1140988262384633e-05,
+      "loss": 0.2346,
+      "step": 2062
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.098735685246216e-05,
+      "loss": 0.2305,
+      "step": 2063
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0834780310704672e-05,
+      "loss": 0.2292,
+      "step": 2064
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0683258966240949e-05,
+      "loss": 0.2383,
+      "step": 2065
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0532793145923535e-05,
+      "loss": 0.2438,
+      "step": 2066
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0383383174328076e-05,
+      "loss": 0.2648,
+      "step": 2067
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.0235029373752758e-05,
+      "loss": 0.2435,
+      "step": 2068
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.0087732064217204e-05,
+      "loss": 0.2422,
+      "step": 2069
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.94149156346219e-06,
+      "loss": 0.2413,
+      "step": 2070
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.79630818694871e-06,
+      "loss": 0.2365,
+      "step": 2071
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.652182247857578e-06,
+      "loss": 0.2377,
+      "step": 2072
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.509114057088374e-06,
+      "loss": 0.2406,
+      "step": 2073
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.367103923259123e-06,
+      "loss": 0.252,
+      "step": 2074
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.226152152705336e-06,
+      "loss": 0.2257,
+      "step": 2075
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.086259049479573e-06,
+      "loss": 0.2255,
+      "step": 2076
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.94742491535072e-06,
+      "loss": 0.2382,
+      "step": 2077
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.80965004980322e-06,
+      "loss": 0.2618,
+      "step": 2078
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 8.672934750036554e-06,
+      "loss": 0.2614,
+      "step": 2079
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.537279310964763e-06,
+      "loss": 0.2432,
+      "step": 2080
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.402684025215323e-06,
+      "loss": 0.2361,
+      "step": 2081
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.269149183128988e-06,
+      "loss": 0.2345,
+      "step": 2082
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.136675072758947e-06,
+      "loss": 0.2257,
+      "step": 2083
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 8.005261979870227e-06,
+      "loss": 0.2474,
+      "step": 2084
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.874910187939178e-06,
+      "loss": 0.2474,
+      "step": 2085
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.745619978152651e-06,
+      "loss": 0.2209,
+      "step": 2086
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.617391629407611e-06,
+      "loss": 0.2417,
+      "step": 2087
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.490225418310403e-06,
+      "loss": 0.2501,
+      "step": 2088
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.364121619176212e-06,
+      "loss": 0.2561,
+      "step": 2089
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.239080504028439e-06,
+      "loss": 0.2299,
+      "step": 2090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.115102342598101e-06,
+      "loss": 0.2454,
+      "step": 2091
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.99218740232338e-06,
+      "loss": 0.2446,
+      "step": 2092
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.870335948348793e-06,
+      "loss": 0.2373,
+      "step": 2093
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.749548243524806e-06,
+      "loss": 0.2484,
+      "step": 2094
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.6298245484073815e-06,
+      "loss": 0.251,
+      "step": 2095
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.511165121257045e-06,
+      "loss": 0.2432,
+      "step": 2096
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.393570218038713e-06,
+      "loss": 0.2336,
+      "step": 2097
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.277040092420916e-06,
+      "loss": 0.2422,
+      "step": 2098
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.161574995775354e-06,
+      "loss": 0.2418,
+      "step": 2099
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.047175177176178e-06,
+      "loss": 0.2421,
+      "step": 2100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.933840883399766e-06,
+      "loss": 0.2325,
+      "step": 2101
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.821572358923832e-06,
+      "loss": 0.236,
+      "step": 2102
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.710369845927266e-06,
+      "loss": 0.2444,
+      "step": 2103
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.600233584289294e-06,
+      "loss": 0.2358,
+      "step": 2104
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.4911638115891505e-06,
+      "loss": 0.2539,
+      "step": 2105
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.38316076310541e-06,
+      "loss": 0.2273,
+      "step": 2106
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.276224671815654e-06,
+      "loss": 0.2289,
+      "step": 2107
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.170355768395918e-06,
+      "loss": 0.2494,
+      "step": 2108
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.065554281220075e-06,
+      "loss": 0.2493,
+      "step": 2109
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.961820436359565e-06,
+      "loss": 0.231,
+      "step": 2110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.859154457582671e-06,
+      "loss": 0.2245,
+      "step": 2111
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.757556566354071e-06,
+      "loss": 0.2189,
+      "step": 2112
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.657026981834622e-06,
+      "loss": 0.245,
+      "step": 2113
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.55756592088058e-06,
+      "loss": 0.2415,
+      "step": 2114
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.4591735980432094e-06,
+      "loss": 0.2482,
+      "step": 2115
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.361850225568453e-06,
+      "loss": 0.2357,
+      "step": 2116
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.265596013396378e-06,
+      "loss": 0.2422,
+      "step": 2117
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.170411169160615e-06,
+      "loss": 0.2343,
+      "step": 2118
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0762958981880875e-06,
+      "loss": 0.2333,
+      "step": 2119
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.983250403498506e-06,
+      "loss": 0.2312,
+      "step": 2120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.891274885803986e-06,
+      "loss": 0.225,
+      "step": 2121
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.8003695435084305e-06,
+      "loss": 0.2178,
+      "step": 2122
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7105345727073135e-06,
+      "loss": 0.2295,
+      "step": 2123
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.621770167187066e-06,
+      "loss": 0.267,
+      "step": 2124
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.534076518424967e-06,
+      "loss": 0.2483,
+      "step": 2125
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.447453815588253e-06,
+      "loss": 0.243,
+      "step": 2126
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.361902245534176e-06,
+      "loss": 0.2313,
+      "step": 2127
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.2774219928094483e-06,
+      "loss": 0.2294,
+      "step": 2128
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.194013239649518e-06,
+      "loss": 0.2526,
+      "step": 2129
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.111676165978794e-06,
+      "loss": 0.238,
+      "step": 2130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.030410949409701e-06,
+      "loss": 0.2424,
+      "step": 2131
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.9502177652426242e-06,
+      "loss": 0.2369,
+      "step": 2132
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.87109678646541e-06,
+      "loss": 0.2472,
+      "step": 2133
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.793048183752922e-06,
+      "loss": 0.2423,
+      "step": 2134
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.7160721254669283e-06,
+      "loss": 0.2423,
+      "step": 2135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.6401687776553827e-06,
+      "loss": 0.2394,
+      "step": 2136
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5653383040524227e-06,
+      "loss": 0.236,
+      "step": 2137
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4915808660777583e-06,
+      "loss": 0.2413,
+      "step": 2138
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.4188966228363418e-06,
+      "loss": 0.246,
+      "step": 2139
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3472857311183093e-06,
+      "loss": 0.2576,
+      "step": 2140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.2767483453981496e-06,
+      "loss": 0.2357,
+      "step": 2141
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.207284617834926e-06,
+      "loss": 0.23,
+      "step": 2142
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.1388946982714986e-06,
+      "loss": 0.2382,
+      "step": 2143
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.0715787342343583e-06,
+      "loss": 0.2467,
+      "step": 2144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.0053368709334606e-06,
+      "loss": 0.2332,
+      "step": 2145
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.940169251261725e-06,
+      "loss": 0.2543,
+      "step": 2146
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8760760157946476e-06,
+      "loss": 0.2428,
+      "step": 2147
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8130573027902997e-06,
+      "loss": 0.2349,
+      "step": 2148
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7511132481888293e-06,
+      "loss": 0.235,
+      "step": 2149
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.690243985612072e-06,
+      "loss": 0.2356,
+      "step": 2150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.630449646363552e-06,
+      "loss": 0.2444,
+      "step": 2151
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5717303594279808e-06,
+      "loss": 0.2375,
+      "step": 2152
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.5140862514709254e-06,
+      "loss": 0.2399,
+      "step": 2153
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.4575174468388074e-06,
+      "loss": 0.2343,
+      "step": 2154
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.4020240675583495e-06,
+      "loss": 0.2351,
+      "step": 2155
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3476062333364625e-06,
+      "loss": 0.2415,
+      "step": 2156
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2942640615600243e-06,
+      "loss": 0.2444,
+      "step": 2157
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2419976672954358e-06,
+      "loss": 0.2388,
+      "step": 2158
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1908071632886764e-06,
+      "loss": 0.2434,
+      "step": 2159
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1406926599646372e-06,
+      "loss": 0.2474,
+      "step": 2160
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0916542654273443e-06,
+      "loss": 0.2351,
+      "step": 2161
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0436920854594023e-06,
+      "loss": 0.2307,
+      "step": 2162
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.968062235218845e-07,
+      "loss": 0.2523,
+      "step": 2163
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.509967807541098e-07,
+      "loss": 0.238,
+      "step": 2164
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.062638559734216e-07,
+      "loss": 0.224,
+      "step": 2165
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.626075456749093e-07,
+      "loss": 0.2233,
+      "step": 2166
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.20027944031354e-07,
+      "loss": 0.249,
+      "step": 2167
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.785251428928386e-07,
+      "loss": 0.2347,
+      "step": 2168
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.380992317866375e-07,
+      "loss": 0.2092,
+      "step": 2169
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.987502979170502e-07,
+      "loss": 0.238,
+      "step": 2170
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.604784261652896e-07,
+      "loss": 0.2336,
+      "step": 2171
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.232836990889834e-07,
+      "loss": 0.2495,
+      "step": 2172
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.871661969223951e-07,
+      "loss": 0.2357,
+      "step": 2173
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.521259975759252e-07,
+      "loss": 0.2318,
+      "step": 2174
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.181631766362216e-07,
+      "loss": 0.2601,
+      "step": 2175
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.85277807365736e-07,
+      "loss": 0.2472,
+      "step": 2176
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.5346996070283476e-07,
+      "loss": 0.2333,
+      "step": 2177
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.2273970526141017e-07,
+      "loss": 0.2343,
+      "step": 2178
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.930871073309361e-07,
+      "loss": 0.2411,
+      "step": 2179
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.645122308761906e-07,
+      "loss": 0.236,
+      "step": 2180
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.3701513753703337e-07,
+      "loss": 0.2434,
+      "step": 2181
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.10595886628684e-07,
+      "loss": 0.2557,
+      "step": 2182
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.852545351409996e-07,
+      "loss": 0.2446,
+      "step": 2183
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6099113773880856e-07,
+      "loss": 0.2572,
+      "step": 2184
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3780574676174338e-07,
+      "loss": 0.2618,
+      "step": 2185
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1569841222374155e-07,
+      "loss": 0.2246,
+      "step": 2186
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9466918181354487e-07,
+      "loss": 0.2419,
+      "step": 2187
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7471810089403352e-07,
+      "loss": 0.2501,
+      "step": 2188
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5584521250250339e-07,
+      "loss": 0.2311,
+      "step": 2189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3805055735038874e-07,
+      "loss": 0.2434,
+      "step": 2190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2133417382320656e-07,
+      "loss": 0.2577,
+      "step": 2191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0569609798050106e-07,
+      "loss": 0.2561,
+      "step": 2192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.113636355578825e-08,
+      "loss": 0.2313,
+      "step": 2193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.765500195650033e-08,
+      "loss": 0.2329,
+      "step": 2194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.525204226370817e-08,
+      "loss": 0.2434,
+      "step": 2195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.392751123239892e-08,
+      "loss": 0.2488,
+      "step": 2196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.368143329114283e-08,
+      "loss": 0.2336,
+      "step": 2197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4513830542093337e-08,
+      "loss": 0.243,
+      "step": 2198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6424722761098087e-08,
+      "loss": 0.2424,
+      "step": 2199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9414127397476833e-08,
+      "loss": 0.2565,
+      "step": 2200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3482059574021488e-08,
+      "loss": 0.2565,
+      "step": 2201
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.628532087107122e-09,
+      "loss": 0.2335,
+      "step": 2202
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8535554063589e-09,
+      "loss": 0.2572,
+      "step": 2203
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1571376750406658e-09,
+      "loss": 0.2527,
+      "step": 2204
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.392847095553321e-10,
+      "loss": 0.243,
+      "step": 2205
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.2426,
+      "step": 2206
+    },
+    {
+      "epoch": 2.0,
+      "step": 2206,
+      "total_flos": 308175427338240.0,
+      "train_loss": 0.34422937221993993,
+      "train_runtime": 53418.5216,
+      "train_samples_per_second": 10.572,
+      "train_steps_per_second": 0.041
+    }
+  ],
+  "max_steps": 2206,
+  "num_train_epochs": 2,
+  "total_flos": 308175427338240.0,
+  "trial_name": null,
+  "trial_params": null
+}