diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,55830 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 9300,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5842293906810036e-07,
+      "loss": 1.6372,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.168458781362007e-07,
+      "loss": 1.6834,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0752688172043011e-06,
+      "loss": 1.7268,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4336917562724014e-06,
+      "loss": 1.5999,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7921146953405017e-06,
+      "loss": 1.6425,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1505376344086023e-06,
+      "loss": 1.543,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5089605734767026e-06,
+      "loss": 1.6935,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.867383512544803e-06,
+      "loss": 1.4967,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 1.3894,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5842293906810035e-06,
+      "loss": 1.3568,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.942652329749104e-06,
+      "loss": 1.2956,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3010752688172045e-06,
+      "loss": 1.2162,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.659498207885305e-06,
+      "loss": 1.1681,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.017921146953405e-06,
+      "loss": 1.2038,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.376344086021506e-06,
+      "loss": 1.0844,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.734767025089606e-06,
+      "loss": 0.9878,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.0931899641577065e-06,
+      "loss": 0.9273,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.9636,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.810035842293908e-06,
+      "loss": 0.8639,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.168458781362007e-06,
+      "loss": 0.8276,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.526881720430108e-06,
+      "loss": 0.8497,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.885304659498208e-06,
+      "loss": 0.7471,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.24372759856631e-06,
+      "loss": 0.7664,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.602150537634409e-06,
+      "loss": 0.6883,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.960573476702509e-06,
+      "loss": 0.695,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.31899641577061e-06,
+      "loss": 0.7511,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.6557,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.003584229390681e-05,
+      "loss": 0.7604,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.039426523297491e-05,
+      "loss": 0.714,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 0.7108,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.7415,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1469534050179212e-05,
+      "loss": 0.7676,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1827956989247313e-05,
+      "loss": 0.7794,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2186379928315413e-05,
+      "loss": 0.7716,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2544802867383513e-05,
+      "loss": 0.7238,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.7021,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3261648745519714e-05,
+      "loss": 0.6921,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3620071684587816e-05,
+      "loss": 0.6811,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3978494623655914e-05,
+      "loss": 0.6629,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4336917562724014e-05,
+      "loss": 0.6608,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4695340501792116e-05,
+      "loss": 0.6054,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5053763440860215e-05,
+      "loss": 0.6638,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5412186379928317e-05,
+      "loss": 0.7462,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5770609318996415e-05,
+      "loss": 0.7126,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.7592,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.648745519713262e-05,
+      "loss": 0.7,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.684587813620072e-05,
+      "loss": 0.8176,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7204301075268818e-05,
+      "loss": 0.7269,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7562724014336916e-05,
+      "loss": 0.7077,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7921146953405018e-05,
+      "loss": 0.6224,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.827956989247312e-05,
+      "loss": 0.7029,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.863799283154122e-05,
+      "loss": 0.7415,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.899641577060932e-05,
+      "loss": 0.7144,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6877,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9713261648745522e-05,
+      "loss": 0.6703,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.007168458781362e-05,
+      "loss": 0.7914,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0430107526881722e-05,
+      "loss": 0.7528,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.078853046594982e-05,
+      "loss": 0.6979,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1146953405017922e-05,
+      "loss": 0.6841,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1505376344086024e-05,
+      "loss": 0.7214,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1863799283154125e-05,
+      "loss": 0.7936,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.7616,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.258064516129032e-05,
+      "loss": 0.7395,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2939068100358423e-05,
+      "loss": 0.6857,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3297491039426525e-05,
+      "loss": 0.7759,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3655913978494626e-05,
+      "loss": 0.6936,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4014336917562724e-05,
+      "loss": 0.7419,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4372759856630826e-05,
+      "loss": 0.6405,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4731182795698928e-05,
+      "loss": 0.6521,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5089605734767026e-05,
+      "loss": 0.6528,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5448028673835127e-05,
+      "loss": 0.6694,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.7184,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.616487455197133e-05,
+      "loss": 0.6048,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.652329749103943e-05,
+      "loss": 0.7889,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6881720430107527e-05,
+      "loss": 0.8032,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7240143369175632e-05,
+      "loss": 0.6836,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.759856630824373e-05,
+      "loss": 0.6927,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7956989247311828e-05,
+      "loss": 0.7611,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.831541218637993e-05,
+      "loss": 0.7119,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8673835125448028e-05,
+      "loss": 0.7331,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.6099,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.939068100358423e-05,
+      "loss": 0.6146,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.974910394265233e-05,
+      "loss": 0.6944,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.010752688172043e-05,
+      "loss": 0.7374,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.046594982078853e-05,
+      "loss": 0.6321,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0824372759856634e-05,
+      "loss": 0.6508,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.118279569892473e-05,
+      "loss": 0.6905,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.154121863799283e-05,
+      "loss": 0.6843,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1899641577060935e-05,
+      "loss": 0.6934,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.7689,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.261648745519714e-05,
+      "loss": 0.6776,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.297491039426524e-05,
+      "loss": 0.7915,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.7652,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.369175627240144e-05,
+      "loss": 0.6991,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.405017921146954e-05,
+      "loss": 0.6245,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4408602150537636e-05,
+      "loss": 0.645,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4767025089605734e-05,
+      "loss": 0.7001,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.512544802867383e-05,
+      "loss": 0.7195,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.6861,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5842293906810036e-05,
+      "loss": 0.7173,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6200716845878134e-05,
+      "loss": 0.6657,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.655913978494624e-05,
+      "loss": 0.6427,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.691756272401434e-05,
+      "loss": 0.67,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.727598566308244e-05,
+      "loss": 0.6867,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.763440860215054e-05,
+      "loss": 0.7213,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.799283154121864e-05,
+      "loss": 0.6767,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8351254480286743e-05,
+      "loss": 0.6912,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.6444,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.906810035842295e-05,
+      "loss": 0.6227,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9426523297491045e-05,
+      "loss": 0.7387,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.978494623655914e-05,
+      "loss": 0.6812,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.014336917562724e-05,
+      "loss": 0.7178,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.050179211469534e-05,
+      "loss": 0.6989,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0860215053763444e-05,
+      "loss": 0.7365,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.121863799283154e-05,
+      "loss": 0.6452,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.157706093189964e-05,
+      "loss": 0.7009,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1935483870967746e-05,
+      "loss": 0.6572,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2293906810035844e-05,
+      "loss": 0.7652,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.265232974910394e-05,
+      "loss": 0.7414,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.301075268817205e-05,
+      "loss": 0.757,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3369175627240145e-05,
+      "loss": 0.6876,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.372759856630825e-05,
+      "loss": 0.7155,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.408602150537635e-05,
+      "loss": 0.848,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.7111,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4802867383512545e-05,
+      "loss": 0.6501,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.516129032258064e-05,
+      "loss": 0.6966,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.551971326164875e-05,
+      "loss": 0.625,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5878136200716846e-05,
+      "loss": 0.7192,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6236559139784944e-05,
+      "loss": 0.7267,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.659498207885305e-05,
+      "loss": 0.6447,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.695340501792115e-05,
+      "loss": 0.7146,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.731182795698925e-05,
+      "loss": 0.6876,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.767025089605735e-05,
+      "loss": 0.6917,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.802867383512545e-05,
+      "loss": 0.6582,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8387096774193554e-05,
+      "loss": 0.7196,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.874551971326165e-05,
+      "loss": 0.7072,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.910394265232976e-05,
+      "loss": 0.6636,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9462365591397855e-05,
+      "loss": 0.7321,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982078853046595e-05,
+      "loss": 0.6879,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.017921146953405e-05,
+      "loss": 0.6703,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.053763440860215e-05,
+      "loss": 0.6817,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0896057347670255e-05,
+      "loss": 0.6853,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1254480286738346e-05,
+      "loss": 0.6893,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.161290322580645e-05,
+      "loss": 0.7376,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1971326164874556e-05,
+      "loss": 0.6863,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.232974910394266e-05,
+      "loss": 0.7358,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.268817204301075e-05,
+      "loss": 0.6428,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.304659498207886e-05,
+      "loss": 0.7345,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.340501792114696e-05,
+      "loss": 0.7162,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3763440860215054e-05,
+      "loss": 0.7311,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.412186379928316e-05,
+      "loss": 0.6706,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4480286738351264e-05,
+      "loss": 0.743,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4838709677419355e-05,
+      "loss": 0.5863,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.519713261648746e-05,
+      "loss": 0.732,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.555555555555556e-05,
+      "loss": 0.6939,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5913978494623656e-05,
+      "loss": 0.6755,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6272401433691755e-05,
+      "loss": 0.7382,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.663082437275986e-05,
+      "loss": 0.7222,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6989247311827965e-05,
+      "loss": 0.6909,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7347670250896056e-05,
+      "loss": 0.7859,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.770609318996416e-05,
+      "loss": 0.6667,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8064516129032266e-05,
+      "loss": 0.6877,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.842293906810036e-05,
+      "loss": 0.7789,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.878136200716846e-05,
+      "loss": 0.7751,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.913978494623657e-05,
+      "loss": 0.6481,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.949820788530466e-05,
+      "loss": 0.7093,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9856630824372764e-05,
+      "loss": 0.66,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.021505376344086e-05,
+      "loss": 0.6466,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.057347670250897e-05,
+      "loss": 0.7368,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.093189964157706e-05,
+      "loss": 0.7663,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.129032258064517e-05,
+      "loss": 0.692,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.164874551971327e-05,
+      "loss": 0.6453,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.200716845878137e-05,
+      "loss": 0.6901,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.236559139784946e-05,
+      "loss": 0.6788,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.272401433691756e-05,
+      "loss": 0.678,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.308243727598566e-05,
+      "loss": 0.8074,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.344086021505376e-05,
+      "loss": 0.5856,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.379928315412187e-05,
+      "loss": 0.7114,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.415770609318996e-05,
+      "loss": 0.706,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.451612903225807e-05,
+      "loss": 0.7223,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.487455197132617e-05,
+      "loss": 0.7512,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.523297491039428e-05,
+      "loss": 0.615,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.559139784946236e-05,
+      "loss": 0.728,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.594982078853047e-05,
+      "loss": 0.6099,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.630824372759857e-05,
+      "loss": 0.6598,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.6355,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.702508960573477e-05,
+      "loss": 0.7034,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.738351254480288e-05,
+      "loss": 0.6627,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.774193548387096e-05,
+      "loss": 0.5975,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.810035842293908e-05,
+      "loss": 0.6775,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.845878136200717e-05,
+      "loss": 0.6076,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.881720430107527e-05,
+      "loss": 0.7467,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.917562724014337e-05,
+      "loss": 0.7113,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.953405017921147e-05,
+      "loss": 0.7279,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.989247311827958e-05,
+      "loss": 0.6856,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.025089605734767e-05,
+      "loss": 0.7018,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.060931899641578e-05,
+      "loss": 0.7005,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.096774193548388e-05,
+      "loss": 0.7095,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.132616487455197e-05,
+      "loss": 0.8213,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.168458781362007e-05,
+      "loss": 0.5771,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.204301075268818e-05,
+      "loss": 0.711,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.240143369175627e-05,
+      "loss": 0.7112,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.275985663082438e-05,
+      "loss": 0.7205,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.311827956989248e-05,
+      "loss": 0.7259,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.347670250896058e-05,
+      "loss": 0.7693,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.383512544802867e-05,
+      "loss": 0.6742,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.419354838709677e-05,
+      "loss": 0.7164,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.455197132616488e-05,
+      "loss": 0.666,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.491039426523297e-05,
+      "loss": 0.7496,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.526881720430108e-05,
+      "loss": 0.6501,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.562724014336918e-05,
+      "loss": 0.6247,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.598566308243728e-05,
+      "loss": 0.6615,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.634408602150538e-05,
+      "loss": 0.6224,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.670250896057349e-05,
+      "loss": 0.7558,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.706093189964157e-05,
+      "loss": 0.6831,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.741935483870968e-05,
+      "loss": 0.7389,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.777777777777778e-05,
+      "loss": 0.5964,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.81362007168459e-05,
+      "loss": 0.6557,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.849462365591398e-05,
+      "loss": 0.6607,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.885304659498209e-05,
+      "loss": 0.6587,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.921146953405019e-05,
+      "loss": 0.6702,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.956989247311829e-05,
+      "loss": 0.6738,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.992831541218638e-05,
+      "loss": 0.6582,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.028673835125448e-05,
+      "loss": 0.6771,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.064516129032258e-05,
+      "loss": 0.6668,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.100358422939068e-05,
+      "loss": 0.6547,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.136200716845879e-05,
+      "loss": 0.749,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.172043010752689e-05,
+      "loss": 0.7169,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.207885304659499e-05,
+      "loss": 0.7037,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.243727598566309e-05,
+      "loss": 0.7571,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.27956989247312e-05,
+      "loss": 0.6475,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.315412186379928e-05,
+      "loss": 0.8196,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.351254480286739e-05,
+      "loss": 0.6781,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.387096774193549e-05,
+      "loss": 0.7955,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.422939068100359e-05,
+      "loss": 0.722,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.458781362007169e-05,
+      "loss": 0.6781,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.494623655913979e-05,
+      "loss": 0.6678,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.530465949820788e-05,
+      "loss": 0.7282,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.566308243727598e-05,
+      "loss": 0.6465,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.60215053763441e-05,
+      "loss": 0.7001,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.637992831541219e-05,
+      "loss": 0.7487,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.673835125448029e-05,
+      "loss": 0.7395,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.709677419354839e-05,
+      "loss": 0.7916,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.74551971326165e-05,
+      "loss": 0.6551,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.781362007168459e-05,
+      "loss": 0.7582,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.81720430107527e-05,
+      "loss": 0.6631,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.85304659498208e-05,
+      "loss": 0.742,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 0.717,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.924731182795699e-05,
+      "loss": 0.7154,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.960573476702509e-05,
+      "loss": 0.6922,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.99641577060932e-05,
+      "loss": 0.71,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.032258064516129e-05,
+      "loss": 0.6963,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.06810035842294e-05,
+      "loss": 0.6313,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.10394265232975e-05,
+      "loss": 0.7351,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.13978494623656e-05,
+      "loss": 0.6719,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.175627240143369e-05,
+      "loss": 0.7222,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.21146953405018e-05,
+      "loss": 0.6591,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.247311827956989e-05,
+      "loss": 0.6325,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.2831541218638e-05,
+      "loss": 0.5594,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.31899641577061e-05,
+      "loss": 0.7197,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.35483870967742e-05,
+      "loss": 0.6672,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.39068100358423e-05,
+      "loss": 0.6389,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.42652329749104e-05,
+      "loss": 0.7165,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.46236559139785e-05,
+      "loss": 0.6837,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.49820788530466e-05,
+      "loss": 0.6388,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.53405017921147e-05,
+      "loss": 0.7712,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.56989247311828e-05,
+      "loss": 0.6857,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.60573476702509e-05,
+      "loss": 0.6716,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.6415770609319e-05,
+      "loss": 0.7072,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.677419354838711e-05,
+      "loss": 0.7173,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.713261648745519e-05,
+      "loss": 0.6908,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.74910394265233e-05,
+      "loss": 0.6876,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.78494623655914e-05,
+      "loss": 0.6556,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.820788530465951e-05,
+      "loss": 0.7828,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.85663082437276e-05,
+      "loss": 0.6163,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.892473118279571e-05,
+      "loss": 0.6332,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.928315412186381e-05,
+      "loss": 0.6929,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.96415770609319e-05,
+      "loss": 0.702,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 0.6216,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999696799171e-05,
+      "loss": 0.675,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999998787196723e-05,
+      "loss": 0.6334,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999997271192762e-05,
+      "loss": 0.6125,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999995148787475e-05,
+      "loss": 0.6798,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999241998112e-05,
+      "loss": 0.6529,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999989084774026e-05,
+      "loss": 0.7482,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999985143166599e-05,
+      "loss": 0.6373,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999980595159316e-05,
+      "loss": 0.7442,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99997544075273e-05,
+      "loss": 0.7046,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999969679947464e-05,
+      "loss": 0.6073,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999963312744216e-05,
+      "loss": 0.6827,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999956339143762e-05,
+      "loss": 0.7549,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999948759146945e-05,
+      "loss": 0.7072,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999940572754686e-05,
+      "loss": 0.7006,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999931779967976e-05,
+      "loss": 0.7122,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999922380787882e-05,
+      "loss": 0.6615,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999912375215546e-05,
+      "loss": 0.6477,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999901763252177e-05,
+      "loss": 0.5999,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999890544899066e-05,
+      "loss": 0.774,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999878720157572e-05,
+      "loss": 0.6401,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99986628902913e-05,
+      "loss": 0.7147,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999853251515247e-05,
+      "loss": 0.676,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999839607617503e-05,
+      "loss": 0.6625,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999825357337555e-05,
+      "loss": 0.6571,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99981050067713e-05,
+      "loss": 0.7949,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99979503763803e-05,
+      "loss": 0.5901,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99977896822213e-05,
+      "loss": 0.7043,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99976229243138e-05,
+      "loss": 0.7198,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999745010267803e-05,
+      "loss": 0.631,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999727121733491e-05,
+      "loss": 0.6461,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999708626830618e-05,
+      "loss": 0.6229,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999689525561424e-05,
+      "loss": 0.6964,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999669817928227e-05,
+      "loss": 0.6892,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999649503933419e-05,
+      "loss": 0.7017,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999628583579462e-05,
+      "loss": 0.6741,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999607056868891e-05,
+      "loss": 0.5839,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999584923804319e-05,
+      "loss": 0.6121,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999562184388431e-05,
+      "loss": 0.7359,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999538838623983e-05,
+      "loss": 0.6874,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999514886513808e-05,
+      "loss": 0.6132,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99949032806081e-05,
+      "loss": 0.7104,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999465163267968e-05,
+      "loss": 0.6672,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999439392138334e-05,
+      "loss": 0.6668,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999413014675032e-05,
+      "loss": 0.7134,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999386030881264e-05,
+      "loss": 0.6985,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999358440760297e-05,
+      "loss": 0.6422,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999330244315486e-05,
+      "loss": 0.5997,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999301441550243e-05,
+      "loss": 0.7526,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999272032468064e-05,
+      "loss": 0.6733,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999242017072516e-05,
+      "loss": 0.7548,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99921139536724e-05,
+      "loss": 0.6879,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999180167355947e-05,
+      "loss": 0.6996,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999148333042428e-05,
+      "loss": 0.632,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99911589243054e-05,
+      "loss": 0.6495,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99908284552422e-05,
+      "loss": 0.5707,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999049192327475e-05,
+      "loss": 0.6267,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999014932844388e-05,
+      "loss": 0.7419,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998980067079111e-05,
+      "loss": 0.6354,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998944595035876e-05,
+      "loss": 0.5917,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998908516718984e-05,
+      "loss": 0.6941,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998871832132808e-05,
+      "loss": 0.6055,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998834541281798e-05,
+      "loss": 0.6185,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998796644170481e-05,
+      "loss": 0.5655,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998758140803449e-05,
+      "loss": 0.6449,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998719031185371e-05,
+      "loss": 0.6633,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998679315320993e-05,
+      "loss": 0.5606,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998638993215131e-05,
+      "loss": 0.6517,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998598064872673e-05,
+      "loss": 0.7765,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998556530298586e-05,
+      "loss": 0.6454,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998514389497906e-05,
+      "loss": 0.5508,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998471642475745e-05,
+      "loss": 0.686,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998428289237283e-05,
+      "loss": 0.5931,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998384329787784e-05,
+      "loss": 0.5997,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998339764132575e-05,
+      "loss": 0.5459,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998294592277063e-05,
+      "loss": 0.6237,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998248814226725e-05,
+      "loss": 0.6316,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998202429987116e-05,
+      "loss": 0.6806,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998155439563857e-05,
+      "loss": 0.7197,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998107842962651e-05,
+      "loss": 0.6383,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998059640189269e-05,
+      "loss": 0.5768,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998010831249557e-05,
+      "loss": 0.6468,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997961416149434e-05,
+      "loss": 0.6251,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997911394894894e-05,
+      "loss": 0.5756,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997860767492002e-05,
+      "loss": 0.5938,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9978095339469e-05,
+      "loss": 0.6941,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997757694265802e-05,
+      "loss": 0.6823,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997705248454993e-05,
+      "loss": 0.6506,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997652196520835e-05,
+      "loss": 0.7306,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997598538469763e-05,
+      "loss": 0.646,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997544274308281e-05,
+      "loss": 0.6156,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997489404042974e-05,
+      "loss": 0.5896,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997433927680498e-05,
+      "loss": 0.52,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997377845227576e-05,
+      "loss": 0.6953,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997321156691011e-05,
+      "loss": 0.7346,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997263862077681e-05,
+      "loss": 0.6048,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997205961394533e-05,
+      "loss": 0.682,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99714745464859e-05,
+      "loss": 0.6303,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997088341846947e-05,
+      "loss": 0.6854,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997028622996774e-05,
+      "loss": 0.5802,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996968298105312e-05,
+      "loss": 0.5819,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99690736717988e-05,
+      "loss": 0.6346,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996845830227863e-05,
+      "loss": 0.6773,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99678368725673e-05,
+      "loss": 0.6526,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996720938274014e-05,
+      "loss": 0.6309,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996657583287326e-05,
+      "loss": 0.7196,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99659362230435e-05,
+      "loss": 0.6895,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996529055332842e-05,
+      "loss": 0.6314,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996463882380636e-05,
+      "loss": 0.6563,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996398103455633e-05,
+      "loss": 0.5865,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996331718565813e-05,
+      "loss": 0.5669,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996264727719223e-05,
+      "loss": 0.7543,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996197130923992e-05,
+      "loss": 0.6112,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996128928188317e-05,
+      "loss": 0.6861,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99606011952047e-05,
+      "loss": 0.6572,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995990704928794e-05,
+      "loss": 0.7158,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995920684421711e-05,
+      "loss": 0.7098,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99585005800771e-05,
+      "loss": 0.6782,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995778825695357e-05,
+      "loss": 0.6044,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995706987493294e-05,
+      "loss": 0.6071,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99563454341023e-05,
+      "loss": 0.6149,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995561493454953e-05,
+      "loss": 0.5915,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995487837636322e-05,
+      "loss": 0.6478,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99541357596327e-05,
+      "loss": 0.6694,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995338708444804e-05,
+      "loss": 0.671,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995263235090003e-05,
+      "loss": 0.5984,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995187155908022e-05,
+      "loss": 0.7182,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995110470908085e-05,
+      "loss": 0.6492,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995033180099495e-05,
+      "loss": 0.5256,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994955283491625e-05,
+      "loss": 0.6774,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994876781093923e-05,
+      "loss": 0.6302,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994797672915908e-05,
+      "loss": 0.6317,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994717958967177e-05,
+      "loss": 0.6555,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994637639257396e-05,
+      "loss": 0.6849,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994556713796306e-05,
+      "loss": 0.6263,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994475182593722e-05,
+      "loss": 0.6857,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994393045659532e-05,
+      "loss": 0.7027,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994310303003698e-05,
+      "loss": 0.6017,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994226954636254e-05,
+      "loss": 0.5816,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99414300056731e-05,
+      "loss": 0.6429,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994058440807047e-05,
+      "loss": 0.5803,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993973275365721e-05,
+      "loss": 0.602,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99388750425366e-05,
+      "loss": 0.7261,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99380112748127e-05,
+      "loss": 0.5974,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993714145059022e-05,
+      "loss": 0.5617,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993626556997466e-05,
+      "loss": 0.6055,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993538363307226e-05,
+      "loss": 0.6015,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993449563998998e-05,
+      "loss": 0.7007,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993360159083552e-05,
+      "loss": 0.7458,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993270148571731e-05,
+      "loss": 0.6493,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99317953247445e-05,
+      "loss": 0.6984,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993088310802701e-05,
+      "loss": 0.6491,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992996483567548e-05,
+      "loss": 0.6353,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992904050780123e-05,
+      "loss": 0.6134,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992811012451642e-05,
+      "loss": 0.5912,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992717368593385e-05,
+      "loss": 0.5733,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992623119216711e-05,
+      "loss": 0.6985,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99252826433305e-05,
+      "loss": 0.6399,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992432803953906e-05,
+      "loss": 0.6259,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992336738090857e-05,
+      "loss": 0.6611,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992240066755554e-05,
+      "loss": 0.661,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992142789959719e-05,
+      "loss": 0.5597,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992044907715153e-05,
+      "loss": 0.5928,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991946420033724e-05,
+      "loss": 0.7429,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991847326927379e-05,
+      "loss": 0.7193,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991747628408137e-05,
+      "loss": 0.6825,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991647324488087e-05,
+      "loss": 0.6052,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991546415179394e-05,
+      "loss": 0.6897,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991444900494297e-05,
+      "loss": 0.702,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99134278044511e-05,
+      "loss": 0.6252,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991240055044213e-05,
+      "loss": 0.6686,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991136724304069e-05,
+      "loss": 0.6676,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991032788237209e-05,
+      "loss": 0.5742,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990928246856237e-05,
+      "loss": 0.6272,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990823100173833e-05,
+      "loss": 0.6203,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99071734820275e-05,
+      "loss": 0.6204,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990610990955811e-05,
+      "loss": 0.5247,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990504028445917e-05,
+      "loss": 0.6699,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99039646068604e-05,
+      "loss": 0.5811,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990288287689227e-05,
+      "loss": 0.5543,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990179509468596e-05,
+      "loss": 0.7113,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990070126037338e-05,
+      "loss": 0.62,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989960137408722e-05,
+      "loss": 0.6461,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989849543596088e-05,
+      "loss": 0.6476,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989738344612844e-05,
+      "loss": 0.643,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989626540472483e-05,
+      "loss": 0.5191,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989514131188559e-05,
+      "loss": 0.6438,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989401116774708e-05,
+      "loss": 0.6063,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989287497244634e-05,
+      "loss": 0.6429,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98917327261212e-05,
+      "loss": 0.5908,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989058442891017e-05,
+      "loss": 0.6015,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988943008095253e-05,
+      "loss": 0.6143,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988826968238825e-05,
+      "loss": 0.7026,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98871032333581e-05,
+      "loss": 0.5141,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988593073400354e-05,
+      "loss": 0.6499,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988475218446675e-05,
+      "loss": 0.603,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988356758489069e-05,
+      "loss": 0.5484,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988237693541901e-05,
+      "loss": 0.6706,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988118023619613e-05,
+      "loss": 0.555,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987997748736716e-05,
+      "loss": 0.5924,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987876868907799e-05,
+      "loss": 0.6733,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987755384147521e-05,
+      "loss": 0.6395,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987633294470618e-05,
+      "loss": 0.6716,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987510599891895e-05,
+      "loss": 0.6778,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987387300426234e-05,
+      "loss": 0.6307,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987263396088586e-05,
+      "loss": 0.6163,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987138886893982e-05,
+      "loss": 0.6901,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98701377285752e-05,
+      "loss": 0.6374,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986888053994375e-05,
+      "loss": 0.6511,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986761730319791e-05,
+      "loss": 0.658,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986634801849093e-05,
+      "loss": 0.6172,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986507268597674e-05,
+      "loss": 0.6945,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986379130581e-05,
+      "loss": 0.5779,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98625038781461e-05,
+      "loss": 0.6069,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986121040314122e-05,
+      "loss": 0.6249,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985991088095223e-05,
+      "loss": 0.5994,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98586053117367e-05,
+      "loss": 0.5921,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985729369565299e-05,
+      "loss": 0.5427,
+      "step": 496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985597603286016e-05,
+      "loss": 0.5023,
+      "step": 497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985465232351805e-05,
+      "loss": 0.5081,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985332256778717e-05,
+      "loss": 0.5814,
+      "step": 499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985198676582881e-05,
+      "loss": 0.6143,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985064491780497e-05,
+      "loss": 0.5884,
+      "step": 501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984929702387838e-05,
+      "loss": 0.6023,
+      "step": 502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984794308421252e-05,
+      "loss": 0.6354,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984658309897162e-05,
+      "loss": 0.6467,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984521706832057e-05,
+      "loss": 0.5161,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984384499242507e-05,
+      "loss": 0.6004,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984246687145156e-05,
+      "loss": 0.6306,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984108270556711e-05,
+      "loss": 0.5931,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983969249493964e-05,
+      "loss": 0.6234,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983829623973775e-05,
+      "loss": 0.8257,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983689394013077e-05,
+      "loss": 0.5731,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983548559628875e-05,
+      "loss": 0.5566,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983407120838253e-05,
+      "loss": 0.4841,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983265077658364e-05,
+      "loss": 0.501,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983122430106434e-05,
+      "loss": 0.5724,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982979178199764e-05,
+      "loss": 0.4829,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982835321955728e-05,
+      "loss": 0.5928,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98269086139177e-05,
+      "loss": 0.6287,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982545796525415e-05,
+      "loss": 0.5807,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982400127374253e-05,
+      "loss": 0.6402,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982253853955951e-05,
+      "loss": 0.623,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982106976288253e-05,
+      "loss": 0.6896,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981959494388966e-05,
+      "loss": 0.6092,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981811408275983e-05,
+      "loss": 0.6678,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981662717967258e-05,
+      "loss": 0.6838,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981513423480828e-05,
+      "loss": 0.6473,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9813635248348e-05,
+      "loss": 0.6198,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981213022047351e-05,
+      "loss": 0.6037,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981061915136736e-05,
+      "loss": 0.5769,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98091020412128e-05,
+      "loss": 0.6788,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980757889019385e-05,
+      "loss": 0.6319,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980604969849521e-05,
+      "loss": 0.6122,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980451446630236e-05,
+      "loss": 0.5964,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980297319380147e-05,
+      "loss": 0.5953,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980142588117949e-05,
+      "loss": 0.646,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979987252862408e-05,
+      "loss": 0.5228,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97983131363236e-05,
+      "loss": 0.6635,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979674770446721e-05,
+      "loss": 0.6418,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979517623324474e-05,
+      "loss": 0.5256,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97935987228468e-05,
+      "loss": 0.5777,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979201517346471e-05,
+      "loss": 0.618,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979042558529048e-05,
+      "loss": 0.6456,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978882995851697e-05,
+      "loss": 0.6775,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978722829333762e-05,
+      "loss": 0.5749,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978562058994674e-05,
+      "loss": 0.6499,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978400684853928e-05,
+      "loss": 0.4958,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978238706931098e-05,
+      "loss": 0.6021,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978076125245827e-05,
+      "loss": 0.6056,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977912939817832e-05,
+      "loss": 0.6068,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977749150666907e-05,
+      "loss": 0.5631,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977584757812914e-05,
+      "loss": 0.5142,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977419761275793e-05,
+      "loss": 0.6434,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97725416107555e-05,
+      "loss": 0.5978,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977087957232275e-05,
+      "loss": 0.5932,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976921149766124e-05,
+      "loss": 0.6369,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976753738697321e-05,
+      "loss": 0.6365,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97658572404618e-05,
+      "loss": 0.5309,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97641710583307e-05,
+      "loss": 0.6649,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976247884078445e-05,
+      "loss": 0.5663,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976078058802826e-05,
+      "loss": 0.6726,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975907630026811e-05,
+      "loss": 0.6388,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975736597771069e-05,
+      "loss": 0.6721,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975564962056343e-05,
+      "loss": 0.6114,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97539272290345e-05,
+      "loss": 0.6099,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975219880333277e-05,
+      "loss": 0.6008,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975046434366787e-05,
+      "loss": 0.6403,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974872385025017e-05,
+      "loss": 0.5749,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974697732329074e-05,
+      "loss": 0.759,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974522476300144e-05,
+      "loss": 0.6359,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974346616959476e-05,
+      "loss": 0.6411,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974170154328403e-05,
+      "loss": 0.5745,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973993088428323e-05,
+      "loss": 0.5048,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973815419280715e-05,
+      "loss": 0.5941,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973637146907122e-05,
+      "loss": 0.6207,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973458271329167e-05,
+      "loss": 0.5454,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973278792568546e-05,
+      "loss": 0.5576,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973098710647022e-05,
+      "loss": 0.5907,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97291802558644e-05,
+      "loss": 0.6059,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97273673740871e-05,
+      "loss": 0.5133,
+      "step": 579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97255484613582e-05,
+      "loss": 0.5589,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97237235178983e-05,
+      "loss": 0.5488,
+      "step": 581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972189254392873e-05,
+      "loss": 0.6956,
+      "step": 582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972005553967154e-05,
+      "loss": 0.6277,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971821250534955e-05,
+      "loss": 0.5192,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971636344118625e-05,
+      "loss": 0.5307,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971450834740592e-05,
+      "loss": 0.6356,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971264722423354e-05,
+      "loss": 0.6581,
+      "step": 587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971078007189482e-05,
+      "loss": 0.658,
+      "step": 588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970890689061622e-05,
+      "loss": 0.6051,
+      "step": 589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970702768062491e-05,
+      "loss": 0.5378,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97051424421488e-05,
+      "loss": 0.6593,
+      "step": 591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970325117541654e-05,
+      "loss": 0.5494,
+      "step": 592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970135388065751e-05,
+      "loss": 0.6116,
+      "step": 593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96994505581018e-05,
+      "loss": 0.5344,
+      "step": 594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969754120798024e-05,
+      "loss": 0.4965,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969562583052441e-05,
+      "loss": 0.5856,
+      "step": 596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969370442596662e-05,
+      "loss": 0.6397,
+      "step": 597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969177699453987e-05,
+      "loss": 0.6526,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968984353647795e-05,
+      "loss": 0.4632,
+      "step": 599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96879040520153e-05,
+      "loss": 0.6334,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96859585413872e-05,
+      "loss": 0.5772,
+      "step": 601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968400700482958e-05,
+      "loss": 0.6998,
+      "step": 602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968204944257911e-05,
+      "loss": 0.5768,
+      "step": 603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968008585487323e-05,
+      "loss": 0.5863,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967811624195004e-05,
+      "loss": 0.6732,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967614060404848e-05,
+      "loss": 0.5366,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96741589414081e-05,
+      "loss": 0.539,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967217125426925e-05,
+      "loss": 0.6317,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967017754287302e-05,
+      "loss": 0.6492,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966817780746119e-05,
+      "loss": 0.5557,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966617204827628e-05,
+      "loss": 0.6395,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966416026556155e-05,
+      "loss": 0.5882,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966214245956103e-05,
+      "loss": 0.6043,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966011863051938e-05,
+      "loss": 0.644,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96580887786821e-05,
+      "loss": 0.6024,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965605290429533e-05,
+      "loss": 0.6003,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965401100760603e-05,
+      "loss": 0.5707,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96519630888618e-05,
+      "loss": 0.6216,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964990914831103e-05,
+      "loss": 0.4643,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964784918620282e-05,
+      "loss": 0.5996,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9645783202787e-05,
+      "loss": 0.5732,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964371119831414e-05,
+      "loss": 0.5348,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964163317303552e-05,
+      "loss": 0.6062,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96395491272032e-05,
+      "loss": 0.6975,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963745906106987e-05,
+      "loss": 0.6754,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963536297488907e-05,
+      "loss": 0.5518,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963326086891498e-05,
+      "loss": 0.5623,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963115274340258e-05,
+      "loss": 0.6502,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962903859860751e-05,
+      "loss": 0.6012,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962691843478617e-05,
+      "loss": 0.6188,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962479225219574e-05,
+      "loss": 0.6367,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962266005109404e-05,
+      "loss": 0.595,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962052183173968e-05,
+      "loss": 0.6126,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9618377594392e-05,
+      "loss": 0.5953,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961622733931102e-05,
+      "loss": 0.599,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961407106675754e-05,
+      "loss": 0.514,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961190877699308e-05,
+      "loss": 0.5725,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960974047027986e-05,
+      "loss": 0.5957,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960756614688089e-05,
+      "loss": 0.5289,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960538580705986e-05,
+      "loss": 0.525,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960319945108118e-05,
+      "loss": 0.6883,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960100707921003e-05,
+      "loss": 0.6606,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95988086917123e-05,
+      "loss": 0.5861,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95966042888546e-05,
+      "loss": 0.6358,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95943938709043e-05,
+      "loss": 0.608,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959217743812947e-05,
+      "loss": 0.591,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958995499079892e-05,
+      "loss": 0.6143,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95877265291822e-05,
+      "loss": 0.6385,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958549205354956e-05,
+      "loss": 0.6347,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958325156417201e-05,
+      "loss": 0.6323,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958100506132127e-05,
+      "loss": 0.6203,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957875254526978e-05,
+      "loss": 0.5681,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957649401629076e-05,
+      "loss": 0.5856,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957422947465813e-05,
+      "loss": 0.633,
+      "step": 654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95719589206465e-05,
+      "loss": 0.5464,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956968235453126e-05,
+      "loss": 0.5754,
+      "step": 656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95673997765885e-05,
+      "loss": 0.5353,
+      "step": 657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956511118709508e-05,
+      "loss": 0.5929,
+      "step": 658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956281658632855e-05,
+      "loss": 0.6177,
+      "step": 659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956051597456719e-05,
+      "loss": 0.5772,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955820935209001e-05,
+      "loss": 0.5933,
+      "step": 661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955589671917678e-05,
+      "loss": 0.5568,
+      "step": 662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955357807610797e-05,
+      "loss": 0.605,
+      "step": 663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955125342316479e-05,
+      "loss": 0.5999,
+      "step": 664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954892276062915e-05,
+      "loss": 0.5607,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954658608878375e-05,
+      "loss": 0.5914,
+      "step": 666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954424340791196e-05,
+      "loss": 0.6091,
+      "step": 667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954189471829789e-05,
+      "loss": 0.6439,
+      "step": 668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953954002022642e-05,
+      "loss": 0.6752,
+      "step": 669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953717931398312e-05,
+      "loss": 0.5619,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953481259985427e-05,
+      "loss": 0.6823,
+      "step": 671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953243987812695e-05,
+      "loss": 0.5781,
+      "step": 672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953006114908889e-05,
+      "loss": 0.5686,
+      "step": 673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952767641302859e-05,
+      "loss": 0.5549,
+      "step": 674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952528567023529e-05,
+      "loss": 0.5712,
+      "step": 675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952288892099892e-05,
+      "loss": 0.6361,
+      "step": 676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952048616561017e-05,
+      "loss": 0.5085,
+      "step": 677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951807740436042e-05,
+      "loss": 0.6445,
+      "step": 678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951566263754184e-05,
+      "loss": 0.5185,
+      "step": 679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951324186544729e-05,
+      "loss": 0.6654,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951081508837035e-05,
+      "loss": 0.5641,
+      "step": 681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950838230660534e-05,
+      "loss": 0.5426,
+      "step": 682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950594352044731e-05,
+      "loss": 0.5363,
+      "step": 683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950349873019204e-05,
+      "loss": 0.5517,
+      "step": 684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950104793613604e-05,
+      "loss": 0.5477,
+      "step": 685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949859113857653e-05,
+      "loss": 0.5367,
+      "step": 686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949612833781147e-05,
+      "loss": 0.6475,
+      "step": 687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949365953413957e-05,
+      "loss": 0.6425,
+      "step": 688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949118472786023e-05,
+      "loss": 0.5977,
+      "step": 689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948870391927361e-05,
+      "loss": 0.5422,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948621710868055e-05,
+      "loss": 0.6129,
+      "step": 691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94837242963827e-05,
+      "loss": 0.5692,
+      "step": 692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948122548268235e-05,
+      "loss": 0.584,
+      "step": 693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947872066788257e-05,
+      "loss": 0.5194,
+      "step": 694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947620985228715e-05,
+      "loss": 0.5358,
+      "step": 695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94736930362006e-05,
+      "loss": 0.6166,
+      "step": 696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947117021992816e-05,
+      "loss": 0.5636,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946864140377579e-05,
+      "loss": 0.5472,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946610658805018e-05,
+      "loss": 0.641,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946356577305878e-05,
+      "loss": 0.6477,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946101895910973e-05,
+      "loss": 0.5444,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945846614651189e-05,
+      "loss": 0.5014,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945590733557488e-05,
+      "loss": 0.5534,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945334252660904e-05,
+      "loss": 0.4991,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945077171992542e-05,
+      "loss": 0.574,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944819491583583e-05,
+      "loss": 0.5658,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944561211465276e-05,
+      "loss": 0.5046,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944302331668947e-05,
+      "loss": 0.527,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94404285222599e-05,
+      "loss": 0.6389,
+      "step": 709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943782773167878e-05,
+      "loss": 0.6387,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943522094526152e-05,
+      "loss": 0.6137,
+      "step": 711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943260816332429e-05,
+      "loss": 0.685,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942998938618394e-05,
+      "loss": 0.63,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942736461415809e-05,
+      "loss": 0.6107,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942473384756509e-05,
+      "loss": 0.5486,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942209708672398e-05,
+      "loss": 0.6025,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941945433195454e-05,
+      "loss": 0.5731,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941680558357732e-05,
+      "loss": 0.5702,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94141508419135e-05,
+      "loss": 0.6911,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941149010728512e-05,
+      "loss": 0.5793,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940882338001483e-05,
+      "loss": 0.4753,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940615066042605e-05,
+      "loss": 0.6133,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940347194884295e-05,
+      "loss": 0.5951,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940078724559039e-05,
+      "loss": 0.6097,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939809655099397e-05,
+      "loss": 0.5933,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939539986538004e-05,
+      "loss": 0.5717,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939269718907564e-05,
+      "loss": 0.6166,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938998852240854e-05,
+      "loss": 0.6076,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938727386570727e-05,
+      "loss": 0.5127,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938455321930104e-05,
+      "loss": 0.5598,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938182658351983e-05,
+      "loss": 0.61,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937909395869434e-05,
+      "loss": 0.6172,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937635534515595e-05,
+      "loss": 0.5188,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937361074323681e-05,
+      "loss": 0.5987,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93708601532698e-05,
+      "loss": 0.5156,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93681035755885e-05,
+      "loss": 0.5122,
+      "step": 736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936534101052724e-05,
+      "loss": 0.6461,
+      "step": 737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936257245842104e-05,
+      "loss": 0.5526,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93597979196057e-05,
+      "loss": 0.5594,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935701739441771e-05,
+      "loss": 0.5854,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935423088319429e-05,
+      "loss": 0.554,
+      "step": 741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935143838627337e-05,
+      "loss": 0.5655,
+      "step": 742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934863990399366e-05,
+      "loss": 0.562,
+      "step": 743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934583543669453e-05,
+      "loss": 0.6747,
+      "step": 744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934302498471613e-05,
+      "loss": 0.4893,
+      "step": 745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934020854839931e-05,
+      "loss": 0.6417,
+      "step": 746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933738612808563e-05,
+      "loss": 0.5519,
+      "step": 747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933455772411741e-05,
+      "loss": 0.5929,
+      "step": 748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933172333683767e-05,
+      "loss": 0.5638,
+      "step": 749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932888296659018e-05,
+      "loss": 0.5972,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93260366137194e-05,
+      "loss": 0.6455,
+      "step": 751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932318427857055e-05,
+      "loss": 0.5312,
+      "step": 752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932032596148958e-05,
+      "loss": 0.524,
+      "step": 753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93174616628231e-05,
+      "loss": 0.5974,
+      "step": 754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931459138291852e-05,
+      "loss": 0.5661,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931171512212396e-05,
+      "loss": 0.6083,
+      "step": 756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930883288078825e-05,
+      "loss": 0.5454,
+      "step": 757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930594465926093e-05,
+      "loss": 0.5088,
+      "step": 758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93030504578923e-05,
+      "loss": 0.6117,
+      "step": 759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930015027703337e-05,
+      "loss": 0.6265,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929724411703587e-05,
+      "loss": 0.6096,
+      "step": 761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929433197825226e-05,
+      "loss": 0.6237,
+      "step": 762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929141386103573e-05,
+      "loss": 0.6136,
+      "step": 763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928848976574019e-05,
+      "loss": 0.6087,
+      "step": 764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928555969272027e-05,
+      "loss": 0.6115,
+      "step": 765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928262364233133e-05,
+      "loss": 0.5667,
+      "step": 766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927968161492947e-05,
+      "loss": 0.5812,
+      "step": 767
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927673361087146e-05,
+      "loss": 0.5414,
+      "step": 768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927377963051487e-05,
+      "loss": 0.5019,
+      "step": 769
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927081967421798e-05,
+      "loss": 0.6053,
+      "step": 770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92678537423397e-05,
+      "loss": 0.4646,
+      "step": 771
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926488183523983e-05,
+      "loss": 0.5472,
+      "step": 772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926190395327875e-05,
+      "loss": 0.6383,
+      "step": 773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925892009681761e-05,
+      "loss": 0.6047,
+      "step": 774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925593026621833e-05,
+      "loss": 0.4391,
+      "step": 775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92529344618435e-05,
+      "loss": 0.6466,
+      "step": 776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924993268405642e-05,
+      "loss": 0.6185,
+      "step": 777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92469249332212e-05,
+      "loss": 0.5536,
+      "step": 778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924391120970261e-05,
+      "loss": 0.594,
+      "step": 779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924089151386613e-05,
+      "loss": 0.5847,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923786584607802e-05,
+      "loss": 0.5866,
+      "step": 781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92348342067052e-05,
+      "loss": 0.5974,
+      "step": 782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923179659611536e-05,
+      "loss": 0.5696,
+      "step": 783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922875301467692e-05,
+      "loss": 0.5867,
+      "step": 784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9225703462759e-05,
+      "loss": 0.6137,
+      "step": 785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922264794073144e-05,
+      "loss": 0.5919,
+      "step": 786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921958644896482e-05,
+      "loss": 0.4949,
+      "step": 787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921651898783045e-05,
+      "loss": 0.6322,
+      "step": 788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921344555770033e-05,
+      "loss": 0.5076,
+      "step": 789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92103661589472e-05,
+      "loss": 0.6023,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920728079194459e-05,
+      "loss": 0.5283,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920418945706662e-05,
+      "loss": 0.6147,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920109215468826e-05,
+      "loss": 0.5084,
+      "step": 793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919798888518513e-05,
+      "loss": 0.5084,
+      "step": 794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91948796489336e-05,
+      "loss": 0.6212,
+      "step": 795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919176444631076e-05,
+      "loss": 0.5498,
+      "step": 796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91886432776944e-05,
+      "loss": 0.4852,
+      "step": 797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91855161434631e-05,
+      "loss": 0.6414,
+      "step": 798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91823830439961e-05,
+      "loss": 0.6138,
+      "step": 799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917924397967336e-05,
+      "loss": 0.5411,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917609895087562e-05,
+      "loss": 0.5668,
+      "step": 801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917294795798429e-05,
+      "loss": 0.5229,
+      "step": 802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916979100138154e-05,
+      "loss": 0.5667,
+      "step": 803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916662808145023e-05,
+      "loss": 0.5348,
+      "step": 804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916345919857397e-05,
+      "loss": 0.5503,
+      "step": 805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916028435313708e-05,
+      "loss": 0.5619,
+      "step": 806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915710354552461e-05,
+      "loss": 0.4461,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915391677612233e-05,
+      "loss": 0.645,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915072404531673e-05,
+      "loss": 0.5196,
+      "step": 809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914752535349502e-05,
+      "loss": 0.5469,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914432070104513e-05,
+      "loss": 0.5051,
+      "step": 811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914111008835575e-05,
+      "loss": 0.554,
+      "step": 812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913789351581625e-05,
+      "loss": 0.6305,
+      "step": 813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913467098381673e-05,
+      "loss": 0.5832,
+      "step": 814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913144249274803e-05,
+      "loss": 0.5531,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91282080430017e-05,
+      "loss": 0.501,
+      "step": 816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912496763496999e-05,
+      "loss": 0.502,
+      "step": 817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912172126904593e-05,
+      "loss": 0.5636,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911846894562324e-05,
+      "loss": 0.5777,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911521066509635e-05,
+      "loss": 0.4884,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911194642786043e-05,
+      "loss": 0.5142,
+      "step": 821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910867623431137e-05,
+      "loss": 0.515,
+      "step": 822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910540008484577e-05,
+      "loss": 0.5518,
+      "step": 823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910211797986098e-05,
+      "loss": 0.5203,
+      "step": 824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909882991975504e-05,
+      "loss": 0.5376,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90955359049267e-05,
+      "loss": 0.5052,
+      "step": 826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909223593577552e-05,
+      "loss": 0.4862,
+      "step": 827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90889300127017e-05,
+      "loss": 0.6335,
+      "step": 828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908561813610615e-05,
+      "loss": 0.5775,
+      "step": 829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908230030639058e-05,
+      "loss": 0.5572,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907897652395736e-05,
+      "loss": 0.5936,
+      "step": 831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907564678920959e-05,
+      "loss": 0.5651,
+      "step": 832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907231110255111e-05,
+      "loss": 0.5689,
+      "step": 833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906896946438646e-05,
+      "loss": 0.5559,
+      "step": 834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906562187512094e-05,
+      "loss": 0.5293,
+      "step": 835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906226833516054e-05,
+      "loss": 0.4962,
+      "step": 836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905890884491195e-05,
+      "loss": 0.5374,
+      "step": 837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905554340478264e-05,
+      "loss": 0.625,
+      "step": 838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905217201518077e-05,
+      "loss": 0.5727,
+      "step": 839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904879467651523e-05,
+      "loss": 0.6058,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90454113891956e-05,
+      "loss": 0.5968,
+      "step": 841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904202215363222e-05,
+      "loss": 0.4933,
+      "step": 842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903862697023614e-05,
+      "loss": 0.5799,
+      "step": 843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903522583941911e-05,
+      "loss": 0.5504,
+      "step": 844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903181876159365e-05,
+      "loss": 0.5522,
+      "step": 845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902840573717293e-05,
+      "loss": 0.4903,
+      "step": 846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902498676657095e-05,
+      "loss": 0.5639,
+      "step": 847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90215618502023e-05,
+      "loss": 0.4317,
+      "step": 848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901813098848238e-05,
+      "loss": 0.5083,
+      "step": 849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901469418182728e-05,
+      "loss": 0.6416,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901125143065384e-05,
+      "loss": 0.7063,
+      "step": 851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900780273537955e-05,
+      "loss": 0.5746,
+      "step": 852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900434809642272e-05,
+      "loss": 0.5079,
+      "step": 853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90008875142023e-05,
+      "loss": 0.577,
+      "step": 854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8997420989138e-05,
+      "loss": 0.532,
+      "step": 855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899394852165024e-05,
+      "loss": 0.4918,
+      "step": 856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899047011216016e-05,
+      "loss": 0.5418,
+      "step": 857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898698576108962e-05,
+      "loss": 0.4662,
+      "step": 858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898349546886122e-05,
+      "loss": 0.6199,
+      "step": 859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897999923589825e-05,
+      "loss": 0.508,
+      "step": 860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897649706262473e-05,
+      "loss": 0.6429,
+      "step": 861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897298894946541e-05,
+      "loss": 0.5721,
+      "step": 862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896947489684576e-05,
+      "loss": 0.4951,
+      "step": 863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896595490519196e-05,
+      "loss": 0.4754,
+      "step": 864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896242897493091e-05,
+      "loss": 0.576,
+      "step": 865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895889710649026e-05,
+      "loss": 0.6399,
+      "step": 866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895535930029833e-05,
+      "loss": 0.5342,
+      "step": 867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895181555678418e-05,
+      "loss": 0.5432,
+      "step": 868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894826587637764e-05,
+      "loss": 0.5344,
+      "step": 869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894471025950916e-05,
+      "loss": 0.609,
+      "step": 870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894114870661001e-05,
+      "loss": 0.6607,
+      "step": 871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893758121811212e-05,
+      "loss": 0.6165,
+      "step": 872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893400779444816e-05,
+      "loss": 0.5211,
+      "step": 873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893042843605152e-05,
+      "loss": 0.6036,
+      "step": 874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892684314335627e-05,
+      "loss": 0.5084,
+      "step": 875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892325191679729e-05,
+      "loss": 0.4894,
+      "step": 876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891965475681008e-05,
+      "loss": 0.5286,
+      "step": 877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891605166383095e-05,
+      "loss": 0.5327,
+      "step": 878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891244263829684e-05,
+      "loss": 0.5923,
+      "step": 879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890882768064549e-05,
+      "loss": 0.4874,
+      "step": 880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890520679131529e-05,
+      "loss": 0.579,
+      "step": 881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89015799707454e-05,
+      "loss": 0.4712,
+      "step": 882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.889794721937567e-05,
+      "loss": 0.5768,
+      "step": 883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889430853764672e-05,
+      "loss": 0.5881,
+      "step": 884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88906639259998e-05,
+      "loss": 0.493,
+      "step": 885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888701338487696e-05,
+      "loss": 0.6058,
+      "step": 886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888335691472093e-05,
+      "loss": 0.464,
+      "step": 887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887969451597517e-05,
+      "loss": 0.4985,
+      "step": 888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887602618908385e-05,
+      "loss": 0.5417,
+      "step": 889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887235193449187e-05,
+      "loss": 0.5604,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886867175264485e-05,
+      "loss": 0.5455,
+      "step": 891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886498564398912e-05,
+      "loss": 0.5383,
+      "step": 892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886129360897173e-05,
+      "loss": 0.5968,
+      "step": 893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885759564804045e-05,
+      "loss": 0.5128,
+      "step": 894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885389176164379e-05,
+      "loss": 0.653,
+      "step": 895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885018195023093e-05,
+      "loss": 0.5603,
+      "step": 896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88464662142518e-05,
+      "loss": 0.5205,
+      "step": 897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884274455415707e-05,
+      "loss": 0.4961,
+      "step": 898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883901697039808e-05,
+      "loss": 0.4871,
+      "step": 899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883528346342694e-05,
+      "loss": 0.5136,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88315440336964e-05,
+      "loss": 0.6074,
+      "step": 901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882779868166004e-05,
+      "loss": 0.5842,
+      "step": 902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882404740777207e-05,
+      "loss": 0.4805,
+      "step": 903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882029021248743e-05,
+      "loss": 0.4499,
+      "step": 904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881652709626182e-05,
+      "loss": 0.4626,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881275805955162e-05,
+      "loss": 0.6104,
+      "step": 906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880898310281394e-05,
+      "loss": 0.5423,
+      "step": 907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88052022265066e-05,
+      "loss": 0.5202,
+      "step": 908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880141543108816e-05,
+      "loss": 0.5781,
+      "step": 909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87976227170179e-05,
+      "loss": 0.568,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879382408475574e-05,
+      "loss": 0.6314,
+      "step": 911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879001953476245e-05,
+      "loss": 0.5459,
+      "step": 912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87862090674994e-05,
+      "loss": 0.5066,
+      "step": 913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878239268342875e-05,
+      "loss": 0.4757,
+      "step": 914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877857038301334e-05,
+      "loss": 0.4424,
+      "step": 915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877474216671675e-05,
+      "loss": 0.4796,
+      "step": 916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877090803500326e-05,
+      "loss": 0.6441,
+      "step": 917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876706798833789e-05,
+      "loss": 0.5154,
+      "step": 918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876322202718633e-05,
+      "loss": 0.5414,
+      "step": 919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875937015201504e-05,
+      "loss": 0.4957,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875551236329117e-05,
+      "loss": 0.6296,
+      "step": 921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87516486614826e-05,
+      "loss": 0.4894,
+      "step": 922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874777904705794e-05,
+      "loss": 0.5185,
+      "step": 923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874390352048644e-05,
+      "loss": 0.6223,
+      "step": 924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87400220822382e-05,
+      "loss": 0.5314,
+      "step": 925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873613473278389e-05,
+      "loss": 0.6134,
+      "step": 926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873224147259503e-05,
+      "loss": 0.5551,
+      "step": 927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872834230214376e-05,
+      "loss": 0.5237,
+      "step": 928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872443722190298e-05,
+      "loss": 0.5573,
+      "step": 929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872052623234632e-05,
+      "loss": 0.5546,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871660933394807e-05,
+      "loss": 0.5214,
+      "step": 931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871268652718329e-05,
+      "loss": 0.4901,
+      "step": 932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870875781252773e-05,
+      "loss": 0.5722,
+      "step": 933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870482319045788e-05,
+      "loss": 0.5362,
+      "step": 934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870088266145094e-05,
+      "loss": 0.4497,
+      "step": 935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869693622598482e-05,
+      "loss": 0.6083,
+      "step": 936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86929838845381e-05,
+      "loss": 0.4806,
+      "step": 937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868902563759017e-05,
+      "loss": 0.5091,
+      "step": 938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868506148562107e-05,
+      "loss": 0.5763,
+      "step": 939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868109142911157e-05,
+      "loss": 0.5434,
+      "step": 940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867711546854316e-05,
+      "loss": 0.4476,
+      "step": 941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867313360439805e-05,
+      "loss": 0.5395,
+      "step": 942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866914583715918e-05,
+      "loss": 0.4608,
+      "step": 943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866515216731014e-05,
+      "loss": 0.5461,
+      "step": 944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866115259533533e-05,
+      "loss": 0.5847,
+      "step": 945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865714712171981e-05,
+      "loss": 0.4797,
+      "step": 946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865313574694935e-05,
+      "loss": 0.4807,
+      "step": 947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864911847151044e-05,
+      "loss": 0.5076,
+      "step": 948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864509529589033e-05,
+      "loss": 0.5741,
+      "step": 949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864106622057693e-05,
+      "loss": 0.4879,
+      "step": 950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863703124605892e-05,
+      "loss": 0.4702,
+      "step": 951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86329903728256e-05,
+      "loss": 0.5216,
+      "step": 952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862894360136712e-05,
+      "loss": 0.4478,
+      "step": 953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862489093217422e-05,
+      "loss": 0.5394,
+      "step": 954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862083236573843e-05,
+      "loss": 0.5469,
+      "step": 955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861676790255198e-05,
+      "loss": 0.6067,
+      "step": 956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861269754310779e-05,
+      "loss": 0.552,
+      "step": 957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860862128789953e-05,
+      "loss": 0.5067,
+      "step": 958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860453913742158e-05,
+      "loss": 0.3949,
+      "step": 959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8600451092169e-05,
+      "loss": 0.4371,
+      "step": 960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85963571526376e-05,
+      "loss": 0.4981,
+      "step": 961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85922573193239e-05,
+      "loss": 0.473,
+      "step": 962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858815159272512e-05,
+      "loss": 0.4927,
+      "step": 963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85840399733392e-05,
+      "loss": 0.4911,
+      "step": 964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857992246166483e-05,
+      "loss": 0.6038,
+      "step": 965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857579905820133e-05,
+      "loss": 0.4923,
+      "step": 966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857166976344885e-05,
+      "loss": 0.4925,
+      "step": 967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856753457790812e-05,
+      "loss": 0.5296,
+      "step": 968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856339350208073e-05,
+      "loss": 0.5407,
+      "step": 969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855924653646887e-05,
+      "loss": 0.563,
+      "step": 970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85550936815755e-05,
+      "loss": 0.5053,
+      "step": 971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855093493790425e-05,
+      "loss": 0.4497,
+      "step": 972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854677030595954e-05,
+      "loss": 0.5428,
+      "step": 973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85425997862464e-05,
+      "loss": 0.5342,
+      "step": 974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85384233792707e-05,
+      "loss": 0.5514,
+      "step": 975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85342410855389e-05,
+      "loss": 0.4729,
+      "step": 976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853005290555826e-05,
+      "loss": 0.5918,
+      "step": 977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852585883983673e-05,
+      "loss": 0.5121,
+      "step": 978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852165888888294e-05,
+      "loss": 0.5086,
+      "step": 979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851745305320627e-05,
+      "loss": 0.5497,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851324133331683e-05,
+      "loss": 0.5121,
+      "step": 981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850902372972537e-05,
+      "loss": 0.5367,
+      "step": 982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850480024294345e-05,
+      "loss": 0.6601,
+      "step": 983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850057087348327e-05,
+      "loss": 0.4895,
+      "step": 984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849633562185778e-05,
+      "loss": 0.5363,
+      "step": 985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849209448858063e-05,
+      "loss": 0.5402,
+      "step": 986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84878474741662e-05,
+      "loss": 0.5375,
+      "step": 987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848359457912953e-05,
+      "loss": 0.5048,
+      "step": 988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847933580398645e-05,
+      "loss": 0.4777,
+      "step": 989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847507114925345e-05,
+      "loss": 0.4901,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847080061544776e-05,
+      "loss": 0.5346,
+      "step": 991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846652420308728e-05,
+      "loss": 0.5852,
+      "step": 992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84622419126907e-05,
+      "loss": 0.5511,
+      "step": 993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845795374477735e-05,
+      "loss": 0.5106,
+      "step": 994
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845365969986731e-05,
+      "loss": 0.5025,
+      "step": 995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844935977848136e-05,
+      "loss": 0.488,
+      "step": 996
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844505398114098e-05,
+      "loss": 0.5109,
+      "step": 997
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844074230836839e-05,
+      "loss": 0.4037,
+      "step": 998
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843642476068653e-05,
+      "loss": 0.5018,
+      "step": 999
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843210133861902e-05,
+      "loss": 0.6134,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84277720426902e-05,
+      "loss": 0.5065,
+      "step": 1001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842343687342515e-05,
+      "loss": 0.572,
+      "step": 1002
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841909583134959e-05,
+      "loss": 0.418,
+      "step": 1003
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841474891699005e-05,
+      "loss": 0.49,
+      "step": 1004
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841039613087372e-05,
+      "loss": 0.5115,
+      "step": 1005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84060374735285e-05,
+      "loss": 0.4695,
+      "step": 1006
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8401672945483e-05,
+      "loss": 0.6614,
+      "step": 1007
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839730254726656e-05,
+      "loss": 0.595,
+      "step": 1008
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839292627940924e-05,
+      "loss": 0.5035,
+      "step": 1009
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838854414244177e-05,
+      "loss": 0.4846,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838415613689561e-05,
+      "loss": 0.6066,
+      "step": 1011
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837976226330296e-05,
+      "loss": 0.525,
+      "step": 1012
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837536252219673e-05,
+      "loss": 0.503,
+      "step": 1013
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837095691411047e-05,
+      "loss": 0.5093,
+      "step": 1014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836654543957853e-05,
+      "loss": 0.6088,
+      "step": 1015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836212809913592e-05,
+      "loss": 0.4443,
+      "step": 1016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835770489331839e-05,
+      "loss": 0.5471,
+      "step": 1017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835327582266237e-05,
+      "loss": 0.5536,
+      "step": 1018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834884088770504e-05,
+      "loss": 0.5409,
+      "step": 1019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834440008898425e-05,
+      "loss": 0.4698,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83399534270386e-05,
+      "loss": 0.5028,
+      "step": 1021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833550090240737e-05,
+      "loss": 0.5159,
+      "step": 1022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833104251563056e-05,
+      "loss": 0.4533,
+      "step": 1023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83265782672489e-05,
+      "loss": 0.472,
+      "step": 1024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832210815780381e-05,
+      "loss": 0.453,
+      "step": 1025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831763218783743e-05,
+      "loss": 0.5672,
+      "step": 1026
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831315035789259e-05,
+      "loss": 0.5416,
+      "step": 1027
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830866266851286e-05,
+      "loss": 0.5284,
+      "step": 1028
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83041691202425e-05,
+      "loss": 0.5211,
+      "step": 1029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82996697136265e-05,
+      "loss": 0.561,
+      "step": 1030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829516444921055e-05,
+      "loss": 0.5496,
+      "step": 1031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829065332754104e-05,
+      "loss": 0.493,
+      "step": 1032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82861363491651e-05,
+      "loss": 0.4871,
+      "step": 1033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828161351463053e-05,
+      "loss": 0.4674,
+      "step": 1034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827708482448586e-05,
+      "loss": 0.5091,
+      "step": 1035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827255027928035e-05,
+      "loss": 0.5292,
+      "step": 1036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826800987956394e-05,
+      "loss": 0.6958,
+      "step": 1037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826346362588728e-05,
+      "loss": 0.5565,
+      "step": 1038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825891151880176e-05,
+      "loss": 0.4612,
+      "step": 1039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825435355885946e-05,
+      "loss": 0.4513,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824978974661314e-05,
+      "loss": 0.5868,
+      "step": 1041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824522008261633e-05,
+      "loss": 0.4841,
+      "step": 1042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824064456742326e-05,
+      "loss": 0.57,
+      "step": 1043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823606320158882e-05,
+      "loss": 0.5668,
+      "step": 1044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823147598566863e-05,
+      "loss": 0.4918,
+      "step": 1045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822688292021904e-05,
+      "loss": 0.4979,
+      "step": 1046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822228400579712e-05,
+      "loss": 0.4874,
+      "step": 1047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82176792429606e-05,
+      "loss": 0.5533,
+      "step": 1048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821306863226796e-05,
+      "loss": 0.4334,
+      "step": 1049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820845217427836e-05,
+      "loss": 0.5384,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820382986955172e-05,
+      "loss": 0.5614,
+      "step": 1051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81992017186486e-05,
+      "loss": 0.5374,
+      "step": 1052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819456772213034e-05,
+      "loss": 0.57,
+      "step": 1053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818992788055889e-05,
+      "loss": 0.5239,
+      "step": 1054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818528219449705e-05,
+      "loss": 0.5554,
+      "step": 1055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81806306645082e-05,
+      "loss": 0.4646,
+      "step": 1056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817597329115649e-05,
+      "loss": 0.568,
+      "step": 1057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817131007500676e-05,
+      "loss": 0.5604,
+      "step": 1058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816664101662458e-05,
+      "loss": 0.4201,
+      "step": 1059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816196611657622e-05,
+      "loss": 0.4572,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815728537542863e-05,
+      "loss": 0.4748,
+      "step": 1061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815259879374952e-05,
+      "loss": 0.4598,
+      "step": 1062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814790637210724e-05,
+      "loss": 0.4875,
+      "step": 1063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814320811107094e-05,
+      "loss": 0.444,
+      "step": 1064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81385040112104e-05,
+      "loss": 0.4398,
+      "step": 1065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813379407309613e-05,
+      "loss": 0.4193,
+      "step": 1066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812907829729936e-05,
+      "loss": 0.523,
+      "step": 1067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812435668439202e-05,
+      "loss": 0.4669,
+      "step": 1068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.811962923494673e-05,
+      "loss": 0.4815,
+      "step": 1069
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811489594953687e-05,
+      "loss": 0.5077,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811015682873647e-05,
+      "loss": 0.5435,
+      "step": 1071
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810541187312032e-05,
+      "loss": 0.5439,
+      "step": 1072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810066108326387e-05,
+      "loss": 0.5828,
+      "step": 1073
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809590445974328e-05,
+      "loss": 0.5045,
+      "step": 1074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809114200313547e-05,
+      "loss": 0.5937,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8086373714018e-05,
+      "loss": 0.4723,
+      "step": 1076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808159959296921e-05,
+      "loss": 0.518,
+      "step": 1077
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807681964056805e-05,
+      "loss": 0.447,
+      "step": 1078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80720338573943e-05,
+      "loss": 0.4391,
+      "step": 1079
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806724224402833e-05,
+      "loss": 0.4942,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806244480105128e-05,
+      "loss": 0.5467,
+      "step": 1081
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8057641529045e-05,
+      "loss": 0.5739,
+      "step": 1082
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805283242859203e-05,
+      "loss": 0.4583,
+      "step": 1083
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80480175002756e-05,
+      "loss": 0.5829,
+      "step": 1084
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80431967446797e-05,
+      "loss": 0.4353,
+      "step": 1085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803837016238895e-05,
+      "loss": 0.575,
+      "step": 1086
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803353775398875e-05,
+      "loss": 0.4733,
+      "step": 1087
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802869952006517e-05,
+      "loss": 0.5395,
+      "step": 1088
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802385546120499e-05,
+      "loss": 0.4468,
+      "step": 1089
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80190055779957e-05,
+      "loss": 0.4898,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80141498710255e-05,
+      "loss": 0.5143,
+      "step": 1091
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800928834088328e-05,
+      "loss": 0.5443,
+      "step": 1092
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800442098815865e-05,
+      "loss": 0.4745,
+      "step": 1093
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799954781344194e-05,
+      "loss": 0.5768,
+      "step": 1094
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799466881732415e-05,
+      "loss": 0.4854,
+      "step": 1095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798978400039702e-05,
+      "loss": 0.5928,
+      "step": 1096
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798489336325298e-05,
+      "loss": 0.5017,
+      "step": 1097
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797999690648517e-05,
+      "loss": 0.4945,
+      "step": 1098
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797509463068744e-05,
+      "loss": 0.4908,
+      "step": 1099
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79701865364543e-05,
+      "loss": 0.5976,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796527262438105e-05,
+      "loss": 0.5874,
+      "step": 1101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796035289506364e-05,
+      "loss": 0.5198,
+      "step": 1102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795542734909873e-05,
+      "loss": 0.4605,
+      "step": 1103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795049598708369e-05,
+      "loss": 0.4982,
+      "step": 1104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79455588096166e-05,
+      "loss": 0.5909,
+      "step": 1105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794061581729625e-05,
+      "loss": 0.502,
+      "step": 1106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793566701072212e-05,
+      "loss": 0.5481,
+      "step": 1107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793071239049439e-05,
+      "loss": 0.4314,
+      "step": 1108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792575195721398e-05,
+      "loss": 0.4671,
+      "step": 1109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79207857114825e-05,
+      "loss": 0.621,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791581365390223e-05,
+      "loss": 0.5089,
+      "step": 1111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79108357850762e-05,
+      "loss": 0.4933,
+      "step": 1112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790585210560812e-05,
+      "loss": 0.4685,
+      "step": 1113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790086261610243e-05,
+      "loss": 0.4586,
+      "step": 1114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789586731716423e-05,
+      "loss": 0.511,
+      "step": 1115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789086620939936e-05,
+      "loss": 0.5242,
+      "step": 1116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788585929341437e-05,
+      "loss": 0.5033,
+      "step": 1117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788084656981649e-05,
+      "loss": 0.614,
+      "step": 1118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787582803921366e-05,
+      "loss": 0.4589,
+      "step": 1119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787080370221453e-05,
+      "loss": 0.6169,
+      "step": 1120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786577355942847e-05,
+      "loss": 0.5133,
+      "step": 1121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786073761146552e-05,
+      "loss": 0.5011,
+      "step": 1122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785569585893644e-05,
+      "loss": 0.5213,
+      "step": 1123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785064830245272e-05,
+      "loss": 0.5787,
+      "step": 1124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784559494262649e-05,
+      "loss": 0.4669,
+      "step": 1125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784053578007066e-05,
+      "loss": 0.4794,
+      "step": 1126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78354708153988e-05,
+      "loss": 0.4306,
+      "step": 1127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783040004922514e-05,
+      "loss": 0.3665,
+      "step": 1128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782532348216475e-05,
+      "loss": 0.4268,
+      "step": 1129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782024111483325e-05,
+      "loss": 0.4959,
+      "step": 1130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781515294784707e-05,
+      "loss": 0.4713,
+      "step": 1131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781005898182328e-05,
+      "loss": 0.4894,
+      "step": 1132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78049592173797e-05,
+      "loss": 0.4304,
+      "step": 1133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77998536551348e-05,
+      "loss": 0.5198,
+      "step": 1134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779474229570782e-05,
+      "loss": 0.6321,
+      "step": 1135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778962513971865e-05,
+      "loss": 0.4618,
+      "step": 1136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778450218778788e-05,
+      "loss": 0.524,
+      "step": 1137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777937344053688e-05,
+      "loss": 0.4573,
+      "step": 1138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777423889858759e-05,
+      "loss": 0.4874,
+      "step": 1139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77690985625628e-05,
+      "loss": 0.492,
+      "step": 1140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77639524330859e-05,
+      "loss": 0.4274,
+      "step": 1141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775880051078099e-05,
+      "loss": 0.463,
+      "step": 1142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775364279627294e-05,
+      "loss": 0.475,
+      "step": 1143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774847929018725e-05,
+      "loss": 0.4259,
+      "step": 1144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774330999315018e-05,
+      "loss": 0.5325,
+      "step": 1145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773813490578864e-05,
+      "loss": 0.4692,
+      "step": 1146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773295402873026e-05,
+      "loss": 0.4511,
+      "step": 1147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77277673626034e-05,
+      "loss": 0.4202,
+      "step": 1148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772257490803711e-05,
+      "loss": 0.5398,
+      "step": 1149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77173766656611e-05,
+      "loss": 0.4693,
+      "step": 1150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771217263610584e-05,
+      "loss": 0.5304,
+      "step": 1151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770696282000244e-05,
+      "loss": 0.4809,
+      "step": 1152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770174721798279e-05,
+      "loss": 0.5277,
+      "step": 1153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769652583067944e-05,
+      "loss": 0.4097,
+      "step": 1154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769129865872562e-05,
+      "loss": 0.4433,
+      "step": 1155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768606570275527e-05,
+      "loss": 0.5281,
+      "step": 1156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768082696340307e-05,
+      "loss": 0.3894,
+      "step": 1157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767558244130438e-05,
+      "loss": 0.4392,
+      "step": 1158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767033213709524e-05,
+      "loss": 0.4482,
+      "step": 1159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766507605141242e-05,
+      "loss": 0.4472,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765981418489337e-05,
+      "loss": 0.5127,
+      "step": 1161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765454653817626e-05,
+      "loss": 0.5341,
+      "step": 1162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764927311189996e-05,
+      "loss": 0.5011,
+      "step": 1163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7643993906704e-05,
+      "loss": 0.567,
+      "step": 1164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763870892322868e-05,
+      "loss": 0.5647,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763341816211493e-05,
+      "loss": 0.4021,
+      "step": 1166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762812162400445e-05,
+      "loss": 0.5216,
+      "step": 1167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76228193095396e-05,
+      "loss": 0.4596,
+      "step": 1168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761751121936343e-05,
+      "loss": 0.3926,
+      "step": 1169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76121973541197e-05,
+      "loss": 0.5352,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76068777144529e-05,
+      "loss": 0.4794,
+      "step": 1171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760155230100821e-05,
+      "loss": 0.4101,
+      "step": 1172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759622111443146e-05,
+      "loss": 0.4835,
+      "step": 1173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759088415536922e-05,
+      "loss": 0.5218,
+      "step": 1174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758554142446878e-05,
+      "loss": 0.3347,
+      "step": 1175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758019292237812e-05,
+      "loss": 0.5237,
+      "step": 1176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757483864974588e-05,
+      "loss": 0.5479,
+      "step": 1177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756947860722143e-05,
+      "loss": 0.4427,
+      "step": 1178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756411279545485e-05,
+      "loss": 0.3784,
+      "step": 1179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75587412150969e-05,
+      "loss": 0.4996,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755336386679906e-05,
+      "loss": 0.5753,
+      "step": 1181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754798075121347e-05,
+      "loss": 0.5191,
+      "step": 1182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754259186899303e-05,
+      "loss": 0.4564,
+      "step": 1183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753719722079129e-05,
+      "loss": 0.4596,
+      "step": 1184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75317968072625e-05,
+      "loss": 0.4944,
+      "step": 1185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752639062906164e-05,
+      "loss": 0.5262,
+      "step": 1186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752097868684436e-05,
+      "loss": 0.4721,
+      "step": 1187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751556098126704e-05,
+      "loss": 0.4606,
+      "step": 1188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751013751298675e-05,
+      "loss": 0.5306,
+      "step": 1189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750470828266121e-05,
+      "loss": 0.5689,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74992732909489e-05,
+      "loss": 0.5395,
+      "step": 1191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7493832538509e-05,
+      "loss": 0.4229,
+      "step": 1192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748838602600132e-05,
+      "loss": 0.474,
+      "step": 1193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748293375408646e-05,
+      "loss": 0.5116,
+      "step": 1194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747747572342565e-05,
+      "loss": 0.4707,
+      "step": 1195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747201193468083e-05,
+      "loss": 0.5165,
+      "step": 1196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74665423885147e-05,
+      "loss": 0.6003,
+      "step": 1197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746106708559054e-05,
+      "loss": 0.4962,
+      "step": 1198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745558602657244e-05,
+      "loss": 0.4711,
+      "step": 1199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745009921212514e-05,
+      "loss": 0.4502,
+      "step": 1200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744460664291407e-05,
+      "loss": 0.5077,
+      "step": 1201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743910831960539e-05,
+      "loss": 0.4486,
+      "step": 1202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743360424286592e-05,
+      "loss": 0.6012,
+      "step": 1203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74280944133632e-05,
+      "loss": 0.5675,
+      "step": 1204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742257883176548e-05,
+      "loss": 0.4471,
+      "step": 1205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741705749874165e-05,
+      "loss": 0.4408,
+      "step": 1206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74115304149614e-05,
+      "loss": 0.4722,
+      "step": 1207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740599758109501e-05,
+      "loss": 0.4941,
+      "step": 1208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740045899781352e-05,
+      "loss": 0.5026,
+      "step": 1209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739491466578866e-05,
+      "loss": 0.4194,
+      "step": 1210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738936458569284e-05,
+      "loss": 0.444,
+      "step": 1211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738380875819915e-05,
+      "loss": 0.5258,
+      "step": 1212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737824718398143e-05,
+      "loss": 0.5152,
+      "step": 1213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73726798637142e-05,
+      "loss": 0.5044,
+      "step": 1214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736710679807266e-05,
+      "loss": 0.4643,
+      "step": 1215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736152798773268e-05,
+      "loss": 0.5186,
+      "step": 1216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73559434333709e-05,
+      "loss": 0.4238,
+      "step": 1217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73503531356646e-05,
+      "loss": 0.435,
+      "step": 1218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734475709529177e-05,
+      "loss": 0.4608,
+      "step": 1219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73391553129311e-05,
+      "loss": 0.4366,
+      "step": 1220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7333547789262e-05,
+      "loss": 0.5299,
+      "step": 1221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732793452496452e-05,
+      "loss": 0.4655,
+      "step": 1222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732231552071946e-05,
+      "loss": 0.4305,
+      "step": 1223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731669077720828e-05,
+      "loss": 0.5183,
+      "step": 1224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731106029511316e-05,
+      "loss": 0.3953,
+      "step": 1225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730542407511696e-05,
+      "loss": 0.4823,
+      "step": 1226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729978211790324e-05,
+      "loss": 0.4804,
+      "step": 1227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729413442415628e-05,
+      "loss": 0.4335,
+      "step": 1228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7288480994561e-05,
+      "loss": 0.5258,
+      "step": 1229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728282182980308e-05,
+      "loss": 0.5134,
+      "step": 1230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727715693056887e-05,
+      "loss": 0.4385,
+      "step": 1231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727148629754537e-05,
+      "loss": 0.569,
+      "step": 1232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726580993142036e-05,
+      "loss": 0.4255,
+      "step": 1233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726012783288225e-05,
+      "loss": 0.4477,
+      "step": 1234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725444000262016e-05,
+      "loss": 0.4683,
+      "step": 1235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724874644132393e-05,
+      "loss": 0.5206,
+      "step": 1236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724304714968406e-05,
+      "loss": 0.4938,
+      "step": 1237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723734212839177e-05,
+      "loss": 0.51,
+      "step": 1238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723163137813897e-05,
+      "loss": 0.4826,
+      "step": 1239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722591489961827e-05,
+      "loss": 0.5281,
+      "step": 1240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722019269352294e-05,
+      "loss": 0.5754,
+      "step": 1241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7214464760547e-05,
+      "loss": 0.5052,
+      "step": 1242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720873110138512e-05,
+      "loss": 0.4985,
+      "step": 1243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720299171673268e-05,
+      "loss": 0.5225,
+      "step": 1244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719724660728575e-05,
+      "loss": 0.4484,
+      "step": 1245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719149577374111e-05,
+      "loss": 0.4653,
+      "step": 1246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718573921679624e-05,
+      "loss": 0.4771,
+      "step": 1247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717997693714925e-05,
+      "loss": 0.4892,
+      "step": 1248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717420893549902e-05,
+      "loss": 0.4936,
+      "step": 1249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716843521254508e-05,
+      "loss": 0.4201,
+      "step": 1250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716265576898771e-05,
+      "loss": 0.4126,
+      "step": 1251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71568706055278e-05,
+      "loss": 0.5101,
+      "step": 1252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715107972286699e-05,
+      "loss": 0.5156,
+      "step": 1253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71452831217076e-05,
+      "loss": 0.5172,
+      "step": 1254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.713948080275265e-05,
+      "loss": 0.5581,
+      "step": 1255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713367276670584e-05,
+      "loss": 0.52,
+      "step": 1256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712785901427157e-05,
+      "loss": 0.4604,
+      "step": 1257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712203954615493e-05,
+      "loss": 0.3855,
+      "step": 1258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711621436306172e-05,
+      "loss": 0.478,
+      "step": 1259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71103834656984e-05,
+      "loss": 0.4383,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710454685477217e-05,
+      "loss": 0.4829,
+      "step": 1261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709870453099088e-05,
+      "loss": 0.5426,
+      "step": 1262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70928564950631e-05,
+      "loss": 0.5147,
+      "step": 1263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708700274769807e-05,
+      "loss": 0.4382,
+      "step": 1264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708114328960573e-05,
+      "loss": 0.4859,
+      "step": 1265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707527812149673e-05,
+      "loss": 0.3697,
+      "step": 1266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706940724408237e-05,
+      "loss": 0.3856,
+      "step": 1267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706353065807474e-05,
+      "loss": 0.5926,
+      "step": 1268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705764836418647e-05,
+      "loss": 0.5161,
+      "step": 1269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705176036313102e-05,
+      "loss": 0.5012,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70458666556225e-05,
+      "loss": 0.5019,
+      "step": 1271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703996724237565e-05,
+      "loss": 0.4535,
+      "step": 1272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703406212410597e-05,
+      "loss": 0.4381,
+      "step": 1273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702815130152968e-05,
+      "loss": 0.487,
+      "step": 1274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702223477536359e-05,
+      "loss": 0.4557,
+      "step": 1275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701631254632527e-05,
+      "loss": 0.4679,
+      "step": 1276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701038461513299e-05,
+      "loss": 0.4038,
+      "step": 1277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700445098250566e-05,
+      "loss": 0.3985,
+      "step": 1278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699851164916296e-05,
+      "loss": 0.4284,
+      "step": 1279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699256661582516e-05,
+      "loss": 0.4108,
+      "step": 1280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698661588321333e-05,
+      "loss": 0.4847,
+      "step": 1281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698065945204914e-05,
+      "loss": 0.4546,
+      "step": 1282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6974697323055e-05,
+      "loss": 0.4898,
+      "step": 1283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6968729496954e-05,
+      "loss": 0.4953,
+      "step": 1284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69627559744699e-05,
+      "loss": 0.3738,
+      "step": 1285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69567767563272e-05,
+      "loss": 0.4382,
+      "step": 1286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695079184325104e-05,
+      "loss": 0.4488,
+      "step": 1287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694480123596729e-05,
+      "loss": 0.5092,
+      "step": 1288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693880493520249e-05,
+      "loss": 0.5354,
+      "step": 1289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693280294168386e-05,
+      "loss": 0.4769,
+      "step": 1290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692679525613932e-05,
+      "loss": 0.5189,
+      "step": 1291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69207818792975e-05,
+      "loss": 0.4962,
+      "step": 1292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69147628118877e-05,
+      "loss": 0.5254,
+      "step": 1293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690873805463992e-05,
+      "loss": 0.5062,
+      "step": 1294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690270760828483e-05,
+      "loss": 0.4277,
+      "step": 1295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689667147355383e-05,
+      "loss": 0.4546,
+      "step": 1296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689062965117895e-05,
+      "loss": 0.441,
+      "step": 1297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688458214189298e-05,
+      "loss": 0.4612,
+      "step": 1298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687852894642933e-05,
+      "loss": 0.347,
+      "step": 1299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687247006552214e-05,
+      "loss": 0.4371,
+      "step": 1300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686640549990627e-05,
+      "loss": 0.5096,
+      "step": 1301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686033525031719e-05,
+      "loss": 0.5015,
+      "step": 1302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685425931749111e-05,
+      "loss": 0.4971,
+      "step": 1303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684817770216495e-05,
+      "loss": 0.4735,
+      "step": 1304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684209040507624e-05,
+      "loss": 0.5614,
+      "step": 1305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683599742696329e-05,
+      "loss": 0.4913,
+      "step": 1306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682989876856506e-05,
+      "loss": 0.3855,
+      "step": 1307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682379443062116e-05,
+      "loss": 0.4557,
+      "step": 1308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681768441387196e-05,
+      "loss": 0.4903,
+      "step": 1309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681156871905848e-05,
+      "loss": 0.4648,
+      "step": 1310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680544734692241e-05,
+      "loss": 0.4768,
+      "step": 1311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679932029820619e-05,
+      "loss": 0.4602,
+      "step": 1312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679318757365288e-05,
+      "loss": 0.4658,
+      "step": 1313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678704917400628e-05,
+      "loss": 0.4257,
+      "step": 1314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678090510001083e-05,
+      "loss": 0.5203,
+      "step": 1315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67747553524117e-05,
+      "loss": 0.462,
+      "step": 1316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676859993195475e-05,
+      "loss": 0.4902,
+      "step": 1317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67624388393865e-05,
+      "loss": 0.4093,
+      "step": 1318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675627207545415e-05,
+      "loss": 0.5031,
+      "step": 1319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675009964090563e-05,
+      "loss": 0.5371,
+      "step": 1320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674392153648952e-05,
+      "loss": 0.5839,
+      "step": 1321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673773776295512e-05,
+      "loss": 0.5186,
+      "step": 1322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673154832105237e-05,
+      "loss": 0.4068,
+      "step": 1323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672535321153198e-05,
+      "loss": 0.4138,
+      "step": 1324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671915243514526e-05,
+      "loss": 0.4334,
+      "step": 1325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671294599264422e-05,
+      "loss": 0.4673,
+      "step": 1326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670673388478162e-05,
+      "loss": 0.506,
+      "step": 1327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670051611231085e-05,
+      "loss": 0.4482,
+      "step": 1328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669429267598602e-05,
+      "loss": 0.4879,
+      "step": 1329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668806357656188e-05,
+      "loss": 0.3862,
+      "step": 1330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66818288147939e-05,
+      "loss": 0.4815,
+      "step": 1331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667558839143827e-05,
+      "loss": 0.5185,
+      "step": 1332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66693423072518e-05,
+      "loss": 0.4644,
+      "step": 1333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666309056299201e-05,
+      "loss": 0.3765,
+      "step": 1334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665683315941714e-05,
+      "loss": 0.4411,
+      "step": 1335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665057009728609e-05,
+      "loss": 0.5181,
+      "step": 1336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66443013773584e-05,
+      "loss": 0.582,
+      "step": 1337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66380270003944e-05,
+      "loss": 0.5079,
+      "step": 1338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663174696715502e-05,
+      "loss": 0.3722,
+      "step": 1339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662546127840191e-05,
+      "loss": 0.4585,
+      "step": 1340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661916993489742e-05,
+      "loss": 0.4534,
+      "step": 1341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66128729374045e-05,
+      "loss": 0.5267,
+      "step": 1342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660657028668694e-05,
+      "loss": 0.3736,
+      "step": 1343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660026198350907e-05,
+      "loss": 0.4144,
+      "step": 1344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659394802863598e-05,
+      "loss": 0.3936,
+      "step": 1345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658762842283343e-05,
+      "loss": 0.4476,
+      "step": 1346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658130316686786e-05,
+      "loss": 0.4464,
+      "step": 1347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657497226150639e-05,
+      "loss": 0.4476,
+      "step": 1348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656863570751686e-05,
+      "loss": 0.3975,
+      "step": 1349
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656229350566774e-05,
+      "loss": 0.5307,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655594565672823e-05,
+      "loss": 0.438,
+      "step": 1351
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654959216146822e-05,
+      "loss": 0.4475,
+      "step": 1352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654323302065822e-05,
+      "loss": 0.4461,
+      "step": 1353
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653686823506949e-05,
+      "loss": 0.4643,
+      "step": 1354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653049780547397e-05,
+      "loss": 0.4177,
+      "step": 1355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652412173264423e-05,
+      "loss": 0.4637,
+      "step": 1356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651774001735359e-05,
+      "loss": 0.4376,
+      "step": 1357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651135266037602e-05,
+      "loss": 0.528,
+      "step": 1358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650495966248618e-05,
+      "loss": 0.3925,
+      "step": 1359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649856102445942e-05,
+      "loss": 0.5213,
+      "step": 1360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649215674707177e-05,
+      "loss": 0.4681,
+      "step": 1361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648574683109991e-05,
+      "loss": 0.3747,
+      "step": 1362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64793312773213e-05,
+      "loss": 0.4648,
+      "step": 1363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647291008651398e-05,
+      "loss": 0.4799,
+      "step": 1364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646648325945671e-05,
+      "loss": 0.4758,
+      "step": 1365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646005079692894e-05,
+      "loss": 0.5335,
+      "step": 1366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645361269971081e-05,
+      "loss": 0.4433,
+      "step": 1367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644716896858315e-05,
+      "loss": 0.3792,
+      "step": 1368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644071960432741e-05,
+      "loss": 0.4342,
+      "step": 1369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643426460772582e-05,
+      "loss": 0.5137,
+      "step": 1370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642780397956123e-05,
+      "loss": 0.4922,
+      "step": 1371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642133772061718e-05,
+      "loss": 0.4065,
+      "step": 1372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64148658316779e-05,
+      "loss": 0.5344,
+      "step": 1373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64083883135283e-05,
+      "loss": 0.5016,
+      "step": 1374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640190516695399e-05,
+      "loss": 0.4834,
+      "step": 1375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639541639274124e-05,
+      "loss": 0.4871,
+      "step": 1376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638892199167699e-05,
+      "loss": 0.4418,
+      "step": 1377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638242196454891e-05,
+      "loss": 0.4562,
+      "step": 1378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637591631214534e-05,
+      "loss": 0.5018,
+      "step": 1379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636940503525524e-05,
+      "loss": 0.3561,
+      "step": 1380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636288813466834e-05,
+      "loss": 0.5332,
+      "step": 1381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6356365611175e-05,
+      "loss": 0.4839,
+      "step": 1382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634983746556628e-05,
+      "loss": 0.4993,
+      "step": 1383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634330369863389e-05,
+      "loss": 0.5188,
+      "step": 1384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633676431117027e-05,
+      "loss": 0.3886,
+      "step": 1385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63302193039685e-05,
+      "loss": 0.407,
+      "step": 1386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63236686778224e-05,
+      "loss": 0.4387,
+      "step": 1387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631711243352639e-05,
+      "loss": 0.4562,
+      "step": 1388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631055057187564e-05,
+      "loss": 0.5321,
+      "step": 1389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630398309366595e-05,
+      "loss": 0.4503,
+      "step": 1390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629740999969385e-05,
+      "loss": 0.417,
+      "step": 1391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629083129075653e-05,
+      "loss": 0.4792,
+      "step": 1392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628424696765183e-05,
+      "loss": 0.4193,
+      "step": 1393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627765703117831e-05,
+      "loss": 0.5267,
+      "step": 1394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627106148213522e-05,
+      "loss": 0.3999,
+      "step": 1395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626446032132244e-05,
+      "loss": 0.5116,
+      "step": 1396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625785354954057e-05,
+      "loss": 0.5708,
+      "step": 1397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625124116759091e-05,
+      "loss": 0.388,
+      "step": 1398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624462317627539e-05,
+      "loss": 0.4479,
+      "step": 1399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623799957639662e-05,
+      "loss": 0.4161,
+      "step": 1400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623137036875793e-05,
+      "loss": 0.4549,
+      "step": 1401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622473555416333e-05,
+      "loss": 0.4458,
+      "step": 1402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621809513341748e-05,
+      "loss": 0.4246,
+      "step": 1403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621144910732574e-05,
+      "loss": 0.4186,
+      "step": 1404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620479747669413e-05,
+      "loss": 0.3792,
+      "step": 1405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619814024232934e-05,
+      "loss": 0.5073,
+      "step": 1406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61914774050388e-05,
+      "loss": 0.458,
+      "step": 1407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618480896563058e-05,
+      "loss": 0.4984,
+      "step": 1408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61781349249134e-05,
+      "loss": 0.4941,
+      "step": 1409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617145528369673e-05,
+      "loss": 0.4788,
+      "step": 1410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616477004279064e-05,
+      "loss": 0.505,
+      "step": 1411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615807920300594e-05,
+      "loss": 0.3844,
+      "step": 1412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61513827651541e-05,
+      "loss": 0.3651,
+      "step": 1413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614468073004725e-05,
+      "loss": 0.517,
+      "step": 1414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613797309849823e-05,
+      "loss": 0.4112,
+      "step": 1415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613125987132054e-05,
+      "loss": 0.4626,
+      "step": 1416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612454104932835e-05,
+      "loss": 0.466,
+      "step": 1417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611781663333654e-05,
+      "loss": 0.5108,
+      "step": 1418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611108662416065e-05,
+      "loss": 0.5016,
+      "step": 1419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610435102261686e-05,
+      "loss": 0.5468,
+      "step": 1420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609760982952211e-05,
+      "loss": 0.3484,
+      "step": 1421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609086304569396e-05,
+      "loss": 0.4681,
+      "step": 1422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608411067195067e-05,
+      "loss": 0.3974,
+      "step": 1423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607735270911114e-05,
+      "loss": 0.4553,
+      "step": 1424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607058915799501e-05,
+      "loss": 0.5568,
+      "step": 1425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606382001942255e-05,
+      "loss": 0.4504,
+      "step": 1426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605704529421473e-05,
+      "loss": 0.442,
+      "step": 1427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605026498319319e-05,
+      "loss": 0.3867,
+      "step": 1428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604347908718025e-05,
+      "loss": 0.4248,
+      "step": 1429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603668760699891e-05,
+      "loss": 0.4208,
+      "step": 1430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602989054347283e-05,
+      "loss": 0.3588,
+      "step": 1431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602308789742636e-05,
+      "loss": 0.4506,
+      "step": 1432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601627966968454e-05,
+      "loss": 0.4804,
+      "step": 1433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600946586107306e-05,
+      "loss": 0.5048,
+      "step": 1434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60026464724183e-05,
+      "loss": 0.3634,
+      "step": 1435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599582150454735e-05,
+      "loss": 0.4537,
+      "step": 1436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598899095828791e-05,
+      "loss": 0.4502,
+      "step": 1437
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59821548344684e-05,
+      "loss": 0.3671,
+      "step": 1438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59753131339179e-05,
+      "loss": 0.372,
+      "step": 1439
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59684658574662e-05,
+      "loss": 0.3979,
+      "step": 1440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59616130059437e-05,
+      "loss": 0.4952,
+      "step": 1441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595475458018155e-05,
+      "loss": 0.4539,
+      "step": 1442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594789058101153e-05,
+      "loss": 0.3905,
+      "step": 1443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594102100926611e-05,
+      "loss": 0.4745,
+      "step": 1444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593414586577845e-05,
+      "loss": 0.468,
+      "step": 1445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592726515138233e-05,
+      "loss": 0.4531,
+      "step": 1446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592037886691228e-05,
+      "loss": 0.4096,
+      "step": 1447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591348701320343e-05,
+      "loss": 0.4398,
+      "step": 1448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590658959109169e-05,
+      "loss": 0.5122,
+      "step": 1449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589968660141353e-05,
+      "loss": 0.4707,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589277804500616e-05,
+      "loss": 0.5212,
+      "step": 1451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588586392270747e-05,
+      "loss": 0.4789,
+      "step": 1452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587894423535597e-05,
+      "loss": 0.3656,
+      "step": 1453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587201898379092e-05,
+      "loss": 0.4581,
+      "step": 1454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58650881688522e-05,
+      "loss": 0.4928,
+      "step": 1455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585815179138038e-05,
+      "loss": 0.4942,
+      "step": 1456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585120985221671e-05,
+      "loss": 0.3979,
+      "step": 1457
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584426235220311e-05,
+      "loss": 0.4057,
+      "step": 1458
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583730929218218e-05,
+      "loss": 0.3717,
+      "step": 1459
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583035067299719e-05,
+      "loss": 0.4527,
+      "step": 1460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582338649549206e-05,
+      "loss": 0.4388,
+      "step": 1461
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581641676051143e-05,
+      "loss": 0.4103,
+      "step": 1462
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580944146890058e-05,
+      "loss": 0.418,
+      "step": 1463
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58024606215055e-05,
+      "loss": 0.4376,
+      "step": 1464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579547421917281e-05,
+      "loss": 0.4806,
+      "step": 1465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578848226274981e-05,
+      "loss": 0.485,
+      "step": 1466
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578148475308451e-05,
+      "loss": 0.4135,
+      "step": 1467
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577448169102556e-05,
+      "loss": 0.4183,
+      "step": 1468
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576747307742231e-05,
+      "loss": 0.3846,
+      "step": 1469
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576045891312475e-05,
+      "loss": 0.4492,
+      "step": 1470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575343919898355e-05,
+      "loss": 0.5507,
+      "step": 1471
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574641393585009e-05,
+      "loss": 0.4445,
+      "step": 1472
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573938312457637e-05,
+      "loss": 0.436,
+      "step": 1473
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573234676601511e-05,
+      "loss": 0.4406,
+      "step": 1474
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572530486101967e-05,
+      "loss": 0.536,
+      "step": 1475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571825741044411e-05,
+      "loss": 0.3975,
+      "step": 1476
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571120441514312e-05,
+      "loss": 0.4147,
+      "step": 1477
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570414587597211e-05,
+      "loss": 0.3826,
+      "step": 1478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569708179378716e-05,
+      "loss": 0.4798,
+      "step": 1479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569001216944495e-05,
+      "loss": 0.4936,
+      "step": 1480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568293700380293e-05,
+      "loss": 0.5271,
+      "step": 1481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567585629771917e-05,
+      "loss": 0.4021,
+      "step": 1482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566877005205242e-05,
+      "loss": 0.4495,
+      "step": 1483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566167826766208e-05,
+      "loss": 0.4153,
+      "step": 1484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565458094540828e-05,
+      "loss": 0.4785,
+      "step": 1485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564747808615175e-05,
+      "loss": 0.5898,
+      "step": 1486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564036969075396e-05,
+      "loss": 0.4711,
+      "step": 1487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563325576007701e-05,
+      "loss": 0.4578,
+      "step": 1488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562613629498368e-05,
+      "loss": 0.419,
+      "step": 1489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56190112963374e-05,
+      "loss": 0.5078,
+      "step": 1490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56118807650023e-05,
+      "loss": 0.4247,
+      "step": 1491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56047447018432e-05,
+      "loss": 0.4651,
+      "step": 1492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559760310772552e-05,
+      "loss": 0.4585,
+      "step": 1493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559045598351544e-05,
+      "loss": 0.411,
+      "step": 1494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558330333007975e-05,
+      "loss": 0.5112,
+      "step": 1495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55761451482859e-05,
+      "loss": 0.3957,
+      "step": 1496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556898143900207e-05,
+      "loss": 0.4477,
+      "step": 1497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556181220309706e-05,
+      "loss": 0.554,
+      "step": 1498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555463744144037e-05,
+      "loss": 0.4711,
+      "step": 1499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554745715490215e-05,
+      "loss": 0.5204,
+      "step": 1500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554027134435322e-05,
+      "loss": 0.4241,
+      "step": 1501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55330800106651e-05,
+      "loss": 0.3862,
+      "step": 1502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552588315470992e-05,
+      "loss": 0.539,
+      "step": 1503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551868077736055e-05,
+      "loss": 0.4741,
+      "step": 1504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551147287949048e-05,
+      "loss": 0.4812,
+      "step": 1505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55042594619739e-05,
+      "loss": 0.46,
+      "step": 1506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549704052568564e-05,
+      "loss": 0.3797,
+      "step": 1507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548981607150122e-05,
+      "loss": 0.3293,
+      "step": 1508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548258610029684e-05,
+      "loss": 0.3916,
+      "step": 1509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547535061294935e-05,
+      "loss": 0.4646,
+      "step": 1510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546810961033623e-05,
+      "loss": 0.4576,
+      "step": 1511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546086309333573e-05,
+      "loss": 0.4152,
+      "step": 1512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545361106282668e-05,
+      "loss": 0.4871,
+      "step": 1513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54463535196886e-05,
+      "loss": 0.3788,
+      "step": 1514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543909046480171e-05,
+      "loss": 0.4412,
+      "step": 1515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543182189904687e-05,
+      "loss": 0.4612,
+      "step": 1516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54245478233056e-05,
+      "loss": 0.4069,
+      "step": 1517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541726823846011e-05,
+      "loss": 0.3645,
+      "step": 1518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540998314539328e-05,
+      "loss": 0.4645,
+      "step": 1519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540269254498863e-05,
+      "loss": 0.3835,
+      "step": 1520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539539643813039e-05,
+      "loss": 0.4226,
+      "step": 1521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538809482570342e-05,
+      "loss": 0.4987,
+      "step": 1522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538078770859326e-05,
+      "loss": 0.5375,
+      "step": 1523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537347508768612e-05,
+      "loss": 0.475,
+      "step": 1524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536615696386889e-05,
+      "loss": 0.4186,
+      "step": 1525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535883333802908e-05,
+      "loss": 0.5075,
+      "step": 1526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535150421105495e-05,
+      "loss": 0.5279,
+      "step": 1527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534416958383536e-05,
+      "loss": 0.5001,
+      "step": 1528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533682945725983e-05,
+      "loss": 0.3996,
+      "step": 1529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53294838322186e-05,
+      "loss": 0.5564,
+      "step": 1530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532213270960255e-05,
+      "loss": 0.4465,
+      "step": 1531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531477609030323e-05,
+      "loss": 0.5185,
+      "step": 1532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530741397521283e-05,
+      "loss": 0.4445,
+      "step": 1533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530004636522425e-05,
+      "loss": 0.5024,
+      "step": 1534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529267326123102e-05,
+      "loss": 0.4252,
+      "step": 1535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528529466412737e-05,
+      "loss": 0.3782,
+      "step": 1536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527791057480817e-05,
+      "loss": 0.5156,
+      "step": 1537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527052099416898e-05,
+      "loss": 0.4328,
+      "step": 1538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526312592310598e-05,
+      "loss": 0.4798,
+      "step": 1539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525572536251607e-05,
+      "loss": 0.4099,
+      "step": 1540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524831931329679e-05,
+      "loss": 0.422,
+      "step": 1541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524090777634632e-05,
+      "loss": 0.4316,
+      "step": 1542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523349075256359e-05,
+      "loss": 0.4834,
+      "step": 1543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522606824284807e-05,
+      "loss": 0.3856,
+      "step": 1544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521864024810003e-05,
+      "loss": 0.5105,
+      "step": 1545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521120676922031e-05,
+      "loss": 0.5675,
+      "step": 1546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520376780711043e-05,
+      "loss": 0.4242,
+      "step": 1547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519632336267262e-05,
+      "loss": 0.4394,
+      "step": 1548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51888734368097e-05,
+      "loss": 0.4969,
+      "step": 1549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518141803042527e-05,
+      "loss": 0.4031,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517395714442347e-05,
+      "loss": 0.4255,
+      "step": 1551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516649077970917e-05,
+      "loss": 0.4025,
+      "step": 1552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515901893718789e-05,
+      "loss": 0.4618,
+      "step": 1553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515154161776583e-05,
+      "loss": 0.4002,
+      "step": 1554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514405882234983e-05,
+      "loss": 0.4199,
+      "step": 1555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513657055184742e-05,
+      "loss": 0.5072,
+      "step": 1556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512907680716676e-05,
+      "loss": 0.3794,
+      "step": 1557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512157758921672e-05,
+      "loss": 0.3998,
+      "step": 1558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511407289890677e-05,
+      "loss": 0.4339,
+      "step": 1559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51065627371471e-05,
+      "loss": 0.4824,
+      "step": 1560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509904710484858e-05,
+      "loss": 0.4437,
+      "step": 1561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509152600292266e-05,
+      "loss": 0.4374,
+      "step": 1562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508399943228152e-05,
+      "loss": 0.4874,
+      "step": 1563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507646739383797e-05,
+      "loss": 0.4501,
+      "step": 1564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506892988850553e-05,
+      "loss": 0.3844,
+      "step": 1565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506138691719833e-05,
+      "loss": 0.4096,
+      "step": 1566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505383848083118e-05,
+      "loss": 0.4999,
+      "step": 1567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504628458031957e-05,
+      "loss": 0.4634,
+      "step": 1568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503872521657964e-05,
+      "loss": 0.367,
+      "step": 1569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503116039052818e-05,
+      "loss": 0.3874,
+      "step": 1570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502359010308265e-05,
+      "loss": 0.499,
+      "step": 1571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501601435516118e-05,
+      "loss": 0.3707,
+      "step": 1572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50084331476826e-05,
+      "loss": 0.4768,
+      "step": 1573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50008464815663e-05,
+      "loss": 0.4279,
+      "step": 1574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499325435773243e-05,
+      "loss": 0.4103,
+      "step": 1575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498565677710175e-05,
+      "loss": 0.4491,
+      "step": 1576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497805374059571e-05,
+      "loss": 0.4228,
+      "step": 1577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49704452491364e-05,
+      "loss": 0.3645,
+      "step": 1578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496283130364658e-05,
+      "loss": 0.4329,
+      "step": 1579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495521190504967e-05,
+      "loss": 0.4572,
+      "step": 1580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494758705426978e-05,
+      "loss": 0.5107,
+      "step": 1581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49399567522316e-05,
+      "loss": 0.4911,
+      "step": 1582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49323209998606e-05,
+      "loss": 0.4716,
+      "step": 1583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492467979808279e-05,
+      "loss": 0.3638,
+      "step": 1584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491703314782492e-05,
+      "loss": 0.4882,
+      "step": 1585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49093810500144e-05,
+      "loss": 0.3968,
+      "step": 1586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490172350557924e-05,
+      "loss": 0.4394,
+      "step": 1587
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489406051544818e-05,
+      "loss": 0.4455,
+      "step": 1588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488639208055057e-05,
+      "loss": 0.4358,
+      "step": 1589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487871820181647e-05,
+      "loss": 0.4788,
+      "step": 1590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487103888017652e-05,
+      "loss": 0.4284,
+      "step": 1591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486335411656213e-05,
+      "loss": 0.4475,
+      "step": 1592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485566391190526e-05,
+      "loss": 0.4226,
+      "step": 1593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48479682671386e-05,
+      "loss": 0.3631,
+      "step": 1594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484026718319548e-05,
+      "loss": 0.4606,
+      "step": 1595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483256066100989e-05,
+      "loss": 0.4885,
+      "step": 1596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48248487015165e-05,
+      "loss": 0.4423,
+      "step": 1597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48171313056506e-05,
+      "loss": 0.5017,
+      "step": 1598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480940847434814e-05,
+      "loss": 0.4362,
+      "step": 1599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480168020854577e-05,
+      "loss": 0.3734,
+      "step": 1600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479394650918078e-05,
+      "loss": 0.5355,
+      "step": 1601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47862073771911e-05,
+      "loss": 0.4211,
+      "step": 1602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477846281351534e-05,
+      "loss": 0.4317,
+      "step": 1603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477071281909278e-05,
+      "loss": 0.4966,
+      "step": 1604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476295739486331e-05,
+      "loss": 0.4346,
+      "step": 1605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475519654176754e-05,
+      "loss": 0.3904,
+      "step": 1606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47474302607467e-05,
+      "loss": 0.4369,
+      "step": 1607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473965855274268e-05,
+      "loss": 0.4087,
+      "step": 1608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473188141869804e-05,
+      "loss": 0.3474,
+      "step": 1609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4724098859556e-05,
+      "loss": 0.4436,
+      "step": 1610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471631087626042e-05,
+      "loss": 0.5168,
+      "step": 1611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470851746975582e-05,
+      "loss": 0.5424,
+      "step": 1612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470071864098741e-05,
+      "loss": 0.4401,
+      "step": 1613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469291439090103e-05,
+      "loss": 0.3654,
+      "step": 1614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468510472044318e-05,
+      "loss": 0.3892,
+      "step": 1615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4677289630561e-05,
+      "loss": 0.4852,
+      "step": 1616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466946912220233e-05,
+      "loss": 0.4974,
+      "step": 1617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466164319631565e-05,
+      "loss": 0.468,
+      "step": 1618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465381185385007e-05,
+      "loss": 0.4499,
+      "step": 1619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464597509575537e-05,
+      "loss": 0.4233,
+      "step": 1620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463813292298203e-05,
+      "loss": 0.4962,
+      "step": 1621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463028533648113e-05,
+      "loss": 0.4183,
+      "step": 1622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462243233720442e-05,
+      "loss": 0.3742,
+      "step": 1623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461457392610433e-05,
+      "loss": 0.4815,
+      "step": 1624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460671010413394e-05,
+      "loss": 0.4333,
+      "step": 1625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459884087224693e-05,
+      "loss": 0.466,
+      "step": 1626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459096623139774e-05,
+      "loss": 0.388,
+      "step": 1627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458308618254137e-05,
+      "loss": 0.3804,
+      "step": 1628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457520072663352e-05,
+      "loss": 0.4436,
+      "step": 1629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456730986463055e-05,
+      "loss": 0.4275,
+      "step": 1630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455941359748948e-05,
+      "loss": 0.4779,
+      "step": 1631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455151192616796e-05,
+      "loss": 0.3379,
+      "step": 1632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45436048516243e-05,
+      "loss": 0.397,
+      "step": 1633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453569237481747e-05,
+      "loss": 0.4337,
+      "step": 1634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452777449670711e-05,
+      "loss": 0.4039,
+      "step": 1635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451985121825351e-05,
+      "loss": 0.4344,
+      "step": 1636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451192254041758e-05,
+      "loss": 0.3818,
+      "step": 1637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450398846416093e-05,
+      "loss": 0.4086,
+      "step": 1638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449604899044581e-05,
+      "loss": 0.442,
+      "step": 1639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448810412023512e-05,
+      "loss": 0.5048,
+      "step": 1640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448015385449242e-05,
+      "loss": 0.4431,
+      "step": 1641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44721981941819e-05,
+      "loss": 0.3825,
+      "step": 1642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446423714026846e-05,
+      "loss": 0.5055,
+      "step": 1643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445627069371758e-05,
+      "loss": 0.432,
+      "step": 1644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444829885549546e-05,
+      "loss": 0.4153,
+      "step": 1645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444032162656893e-05,
+      "loss": 0.4265,
+      "step": 1646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443233900790544e-05,
+      "loss": 0.4265,
+      "step": 1647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442435100047317e-05,
+      "loss": 0.3813,
+      "step": 1648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441635760524087e-05,
+      "loss": 0.447,
+      "step": 1649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4408358823178e-05,
+      "loss": 0.4652,
+      "step": 1650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440035465525466e-05,
+      "loss": 0.4218,
+      "step": 1651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439234510244156e-05,
+      "loss": 0.4154,
+      "step": 1652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438433016571016e-05,
+      "loss": 0.409,
+      "step": 1653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437630984603246e-05,
+      "loss": 0.4809,
+      "step": 1654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436828414438122e-05,
+      "loss": 0.3723,
+      "step": 1655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436025306172975e-05,
+      "loss": 0.4285,
+      "step": 1656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435221659905208e-05,
+      "loss": 0.3921,
+      "step": 1657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434417475732289e-05,
+      "loss": 0.3541,
+      "step": 1658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433612753751748e-05,
+      "loss": 0.4504,
+      "step": 1659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432807494061182e-05,
+      "loss": 0.4237,
+      "step": 1660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432001696758255e-05,
+      "loss": 0.3724,
+      "step": 1661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431195361940692e-05,
+      "loss": 0.4419,
+      "step": 1662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430388489706286e-05,
+      "loss": 0.3342,
+      "step": 1663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429581080152896e-05,
+      "loss": 0.397,
+      "step": 1664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428773133378444e-05,
+      "loss": 0.3736,
+      "step": 1665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42796464948092e-05,
+      "loss": 0.473,
+      "step": 1666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427155628558372e-05,
+      "loss": 0.44,
+      "step": 1667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426346070708925e-05,
+      "loss": 0.478,
+      "step": 1668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425535976030759e-05,
+      "loss": 0.3701,
+      "step": 1669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424725344622121e-05,
+      "loss": 0.357,
+      "step": 1670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423914176581327e-05,
+      "loss": 0.3535,
+      "step": 1671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423102472006755e-05,
+      "loss": 0.4055,
+      "step": 1672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422290230996849e-05,
+      "loss": 0.4011,
+      "step": 1673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421477453650118e-05,
+      "loss": 0.4499,
+      "step": 1674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420664140065137e-05,
+      "loss": 0.4035,
+      "step": 1675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419850290340541e-05,
+      "loss": 0.4908,
+      "step": 1676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419035904575038e-05,
+      "loss": 0.4738,
+      "step": 1677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418220982867397e-05,
+      "loss": 0.4007,
+      "step": 1678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417405525316448e-05,
+      "loss": 0.4381,
+      "step": 1679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416589532021095e-05,
+      "loss": 0.4739,
+      "step": 1680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415773003080297e-05,
+      "loss": 0.5292,
+      "step": 1681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414955938593087e-05,
+      "loss": 0.4761,
+      "step": 1682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414138338658556e-05,
+      "loss": 0.3605,
+      "step": 1683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413320203375866e-05,
+      "loss": 0.4363,
+      "step": 1684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412501532844237e-05,
+      "loss": 0.4434,
+      "step": 1685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41168232716296e-05,
+      "loss": 0.4401,
+      "step": 1686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410862586431387e-05,
+      "loss": 0.4412,
+      "step": 1687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410042310748939e-05,
+      "loss": 0.4432,
+      "step": 1688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409221500215095e-05,
+      "loss": 0.3987,
+      "step": 1689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408400154929409e-05,
+      "loss": 0.4418,
+      "step": 1690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407578274991489e-05,
+      "loss": 0.3951,
+      "step": 1691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406755860501015e-05,
+      "loss": 0.3258,
+      "step": 1692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405932911557728e-05,
+      "loss": 0.4506,
+      "step": 1693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405109428261439e-05,
+      "loss": 0.4241,
+      "step": 1694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404285410712017e-05,
+      "loss": 0.4428,
+      "step": 1695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403460859009402e-05,
+      "loss": 0.3876,
+      "step": 1696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402635773253593e-05,
+      "loss": 0.3695,
+      "step": 1697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401810153544659e-05,
+      "loss": 0.3864,
+      "step": 1698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400983999982729e-05,
+      "loss": 0.4518,
+      "step": 1699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400157312668001e-05,
+      "loss": 0.4353,
+      "step": 1700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399330091700736e-05,
+      "loss": 0.4157,
+      "step": 1701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398502337181258e-05,
+      "loss": 0.4322,
+      "step": 1702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397674049209959e-05,
+      "loss": 0.3787,
+      "step": 1703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396845227887294e-05,
+      "loss": 0.445,
+      "step": 1704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396015873313781e-05,
+      "loss": 0.3364,
+      "step": 1705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395185985590007e-05,
+      "loss": 0.438,
+      "step": 1706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39435556481662e-05,
+      "loss": 0.4759,
+      "step": 1707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393524611094331e-05,
+      "loss": 0.457,
+      "step": 1708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392693124523924e-05,
+      "loss": 0.4277,
+      "step": 1709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391861105206236e-05,
+      "loss": 0.3511,
+      "step": 1710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391028553242178e-05,
+      "loss": 0.441,
+      "step": 1711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39019546873272e-05,
+      "loss": 0.441,
+      "step": 1712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.389361851778903e-05,
+      "loss": 0.4585,
+      "step": 1713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388527702481825e-05,
+      "loss": 0.3513,
+      "step": 1714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387693020942652e-05,
+      "loss": 0.3739,
+      "step": 1715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386857807262615e-05,
+      "loss": 0.4717,
+      "step": 1716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386022061543008e-05,
+      "loss": 0.4011,
+      "step": 1717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385185783885192e-05,
+      "loss": 0.4618,
+      "step": 1718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384348974390591e-05,
+      "loss": 0.3755,
+      "step": 1719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.383511633160692e-05,
+      "loss": 0.5026,
+      "step": 1720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382673760297049e-05,
+      "loss": 0.3536,
+      "step": 1721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381835355901281e-05,
+      "loss": 0.4319,
+      "step": 1722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380996420075066e-05,
+      "loss": 0.4986,
+      "step": 1723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380156952920156e-05,
+      "loss": 0.4474,
+      "step": 1724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379316954538357e-05,
+      "loss": 0.3919,
+      "step": 1725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378476425031545e-05,
+      "loss": 0.3597,
+      "step": 1726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377635364501662e-05,
+      "loss": 0.4292,
+      "step": 1727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376793773050711e-05,
+      "loss": 0.4461,
+      "step": 1728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375951650780759e-05,
+      "loss": 0.4719,
+      "step": 1729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375108997793942e-05,
+      "loss": 0.3913,
+      "step": 1730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374265814192453e-05,
+      "loss": 0.4186,
+      "step": 1731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373422100078558e-05,
+      "loss": 0.4064,
+      "step": 1732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372577855554581e-05,
+      "loss": 0.5046,
+      "step": 1733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371733080722911e-05,
+      "loss": 0.4591,
+      "step": 1734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370887775686004e-05,
+      "loss": 0.3846,
+      "step": 1735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37004194054638e-05,
+      "loss": 0.3735,
+      "step": 1736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36919557540662e-05,
+      "loss": 0.4627,
+      "step": 1737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368348680369371e-05,
+      "loss": 0.4378,
+      "step": 1738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367501255537346e-05,
+      "loss": 0.3945,
+      "step": 1739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366653301013322e-05,
+      "loss": 0.4589,
+      "step": 1740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365804816900138e-05,
+      "loss": 0.3614,
+      "step": 1741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3649558033007e-05,
+      "loss": 0.404,
+      "step": 1742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364106260317973e-05,
+      "loss": 0.4461,
+      "step": 1743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363256188054991e-05,
+      "loss": 0.4354,
+      "step": 1744
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362405586614855e-05,
+      "loss": 0.4225,
+      "step": 1745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361554456100721e-05,
+      "loss": 0.4835,
+      "step": 1746
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360702796615818e-05,
+      "loss": 0.4188,
+      "step": 1747
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359850608263434e-05,
+      "loss": 0.4896,
+      "step": 1748
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358997891146923e-05,
+      "loss": 0.4094,
+      "step": 1749
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358144645369705e-05,
+      "loss": 0.3758,
+      "step": 1750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357290871035256e-05,
+      "loss": 0.2904,
+      "step": 1751
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356436568247127e-05,
+      "loss": 0.3695,
+      "step": 1752
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355581737108928e-05,
+      "loss": 0.4344,
+      "step": 1753
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35472637772433e-05,
+      "loss": 0.3955,
+      "step": 1754
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353870490197074e-05,
+      "loss": 0.4304,
+      "step": 1755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353014074630962e-05,
+      "loss": 0.2859,
+      "step": 1756
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35215713112986e-05,
+      "loss": 0.3716,
+      "step": 1757
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351299659797699e-05,
+      "loss": 0.3573,
+      "step": 1758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350441660738473e-05,
+      "loss": 0.5407,
+      "step": 1759
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349583134056239e-05,
+      "loss": 0.5387,
+      "step": 1760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34872407985512e-05,
+      "loss": 0.5016,
+      "step": 1761
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347864498239304e-05,
+      "loss": 0.4647,
+      "step": 1762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347004389313041e-05,
+      "loss": 0.4813,
+      "step": 1763
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346143753180644e-05,
+      "loss": 0.5049,
+      "step": 1764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345282589946492e-05,
+      "loss": 0.4336,
+      "step": 1765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344420899715027e-05,
+      "loss": 0.4338,
+      "step": 1766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343558682590756e-05,
+      "loss": 0.4053,
+      "step": 1767
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342695938678247e-05,
+      "loss": 0.4616,
+      "step": 1768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341832668082136e-05,
+      "loss": 0.4985,
+      "step": 1769
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340968870907119e-05,
+      "loss": 0.3746,
+      "step": 1770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340104547257957e-05,
+      "loss": 0.4775,
+      "step": 1771
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339239697239478e-05,
+      "loss": 0.5103,
+      "step": 1772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33837432095657e-05,
+      "loss": 0.3956,
+      "step": 1773
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337508418514187e-05,
+      "loss": 0.3863,
+      "step": 1774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336641990017343e-05,
+      "loss": 0.4855,
+      "step": 1775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335775035571122e-05,
+      "loss": 0.4475,
+      "step": 1776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334907555280668e-05,
+      "loss": 0.4003,
+      "step": 1777
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334039549251189e-05,
+      "loss": 0.4906,
+      "step": 1778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333171017587956e-05,
+      "loss": 0.3974,
+      "step": 1779
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332301960396305e-05,
+      "loss": 0.4456,
+      "step": 1780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331432377781637e-05,
+      "loss": 0.4155,
+      "step": 1781
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330562269849414e-05,
+      "loss": 0.4284,
+      "step": 1782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329691636705163e-05,
+      "loss": 0.3788,
+      "step": 1783
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328820478454476e-05,
+      "loss": 0.3884,
+      "step": 1784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327948795203006e-05,
+      "loss": 0.3394,
+      "step": 1785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32707658705647e-05,
+      "loss": 0.3583,
+      "step": 1786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326203854120652e-05,
+      "loss": 0.4337,
+      "step": 1787
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325330596501397e-05,
+      "loss": 0.3658,
+      "step": 1788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324456814304614e-05,
+      "loss": 0.4906,
+      "step": 1789
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323582507636272e-05,
+      "loss": 0.4501,
+      "step": 1790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322707676602414e-05,
+      "loss": 0.4266,
+      "step": 1791
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321832321309134e-05,
+      "loss": 0.4462,
+      "step": 1792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320956441862597e-05,
+      "loss": 0.3416,
+      "step": 1793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320080038369032e-05,
+      "loss": 0.3913,
+      "step": 1794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319203110934726e-05,
+      "loss": 0.3419,
+      "step": 1795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318325659666036e-05,
+      "loss": 0.4065,
+      "step": 1796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317447684669378e-05,
+      "loss": 0.4628,
+      "step": 1797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316569186051234e-05,
+      "loss": 0.3882,
+      "step": 1798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315690163918148e-05,
+      "loss": 0.3676,
+      "step": 1799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314810618376729e-05,
+      "loss": 0.4668,
+      "step": 1800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313930549533646e-05,
+      "loss": 0.3433,
+      "step": 1801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313049957495637e-05,
+      "loss": 0.3791,
+      "step": 1802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3121688423695e-05,
+      "loss": 0.3068,
+      "step": 1803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311287204262096e-05,
+      "loss": 0.5364,
+      "step": 1804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310405043280349e-05,
+      "loss": 0.4036,
+      "step": 1805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309522359531251e-05,
+      "loss": 0.4285,
+      "step": 1806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308639153121852e-05,
+      "loss": 0.5055,
+      "step": 1807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307755424159269e-05,
+      "loss": 0.5257,
+      "step": 1808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30687117275068e-05,
+      "loss": 0.4132,
+      "step": 1809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305986399003328e-05,
+      "loss": 0.3981,
+      "step": 1810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305101103024516e-05,
+      "loss": 0.4129,
+      "step": 1811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.304215284921618e-05,
+      "loss": 0.3664,
+      "step": 1812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303328944802061e-05,
+      "loss": 0.3861,
+      "step": 1813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302442082773344e-05,
+      "loss": 0.3168,
+      "step": 1814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301554698943025e-05,
+      "loss": 0.4763,
+      "step": 1815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300666793418726e-05,
+      "loss": 0.4904,
+      "step": 1816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299778366308132e-05,
+      "loss": 0.46,
+      "step": 1817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298889417718992e-05,
+      "loss": 0.4327,
+      "step": 1818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29799994775912e-05,
+      "loss": 0.3517,
+      "step": 1819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297109956536389e-05,
+      "loss": 0.3833,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296219444158738e-05,
+      "loss": 0.4123,
+      "step": 1821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295328410734169e-05,
+      "loss": 0.4766,
+      "step": 1822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294436856370745e-05,
+      "loss": 0.3586,
+      "step": 1823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293544781176597e-05,
+      "loss": 0.3782,
+      "step": 1824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292652185259914e-05,
+      "loss": 0.4269,
+      "step": 1825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291759068728952e-05,
+      "loss": 0.4178,
+      "step": 1826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290865431692025e-05,
+      "loss": 0.432,
+      "step": 1827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289971274257517e-05,
+      "loss": 0.3668,
+      "step": 1828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289076596533872e-05,
+      "loss": 0.351,
+      "step": 1829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288181398629594e-05,
+      "loss": 0.4695,
+      "step": 1830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287285680653254e-05,
+      "loss": 0.367,
+      "step": 1831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286389442713488e-05,
+      "loss": 0.4198,
+      "step": 1832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285492684918989e-05,
+      "loss": 0.3511,
+      "step": 1833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284595407378514e-05,
+      "loss": 0.4183,
+      "step": 1834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28369761020089e-05,
+      "loss": 0.4091,
+      "step": 1835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282799293495e-05,
+      "loss": 0.417,
+      "step": 1836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28190045736979e-05,
+      "loss": 0.4069,
+      "step": 1837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281001101934275e-05,
+      "loss": 0.4311,
+      "step": 1838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280101227297525e-05,
+      "loss": 0.41,
+      "step": 1839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279200833568681e-05,
+      "loss": 0.4812,
+      "step": 1840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278299920856941e-05,
+      "loss": 0.3922,
+      "step": 1841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277398489271567e-05,
+      "loss": 0.4213,
+      "step": 1842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.276496538921888e-05,
+      "loss": 0.3972,
+      "step": 1843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275594069917291e-05,
+      "loss": 0.3974,
+      "step": 1844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274691082367227e-05,
+      "loss": 0.3767,
+      "step": 1845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273787576381213e-05,
+      "loss": 0.4573,
+      "step": 1846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272883552068822e-05,
+      "loss": 0.3638,
+      "step": 1847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2719790095397e-05,
+      "loss": 0.3733,
+      "step": 1848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271073948903547e-05,
+      "loss": 0.4172,
+      "step": 1849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27016837027013e-05,
+      "loss": 0.3969,
+      "step": 1850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269262273749277e-05,
+      "loss": 0.3572,
+      "step": 1851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26835565945088e-05,
+      "loss": 0.4657,
+      "step": 1852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267448527484894e-05,
+      "loss": 0.4314,
+      "step": 1853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266540877961336e-05,
+      "loss": 0.3748,
+      "step": 1854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265632710990285e-05,
+      "loss": 0.4281,
+      "step": 1855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264724026681886e-05,
+      "loss": 0.3386,
+      "step": 1856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263814825146342e-05,
+      "loss": 0.3382,
+      "step": 1857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262905106493925e-05,
+      "loss": 0.3139,
+      "step": 1858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261994870834961e-05,
+      "loss": 0.4819,
+      "step": 1859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261084118279847e-05,
+      "loss": 0.4466,
+      "step": 1860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260172848939038e-05,
+      "loss": 0.4074,
+      "step": 1861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259261062923055e-05,
+      "loss": 0.3602,
+      "step": 1862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.258348760342476e-05,
+      "loss": 0.4708,
+      "step": 1863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25743594130795e-05,
+      "loss": 0.3627,
+      "step": 1864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256522605930182e-05,
+      "loss": 0.4915,
+      "step": 1865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25560875431994e-05,
+      "loss": 0.3985,
+      "step": 1866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254694386588059e-05,
+      "loss": 0.3922,
+      "step": 1867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25377950284543e-05,
+      "loss": 0.3758,
+      "step": 1868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252864103203014e-05,
+      "loss": 0.5082,
+      "step": 1869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251948187771831e-05,
+      "loss": 0.3767,
+      "step": 1870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25103175666296e-05,
+      "loss": 0.4242,
+      "step": 1871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25011480998755e-05,
+      "loss": 0.3692,
+      "step": 1872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249197347856807e-05,
+      "loss": 0.4493,
+      "step": 1873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248279370382001e-05,
+      "loss": 0.4033,
+      "step": 1874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247360877674463e-05,
+      "loss": 0.4326,
+      "step": 1875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24644186984559e-05,
+      "loss": 0.4345,
+      "step": 1876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245522347006841e-05,
+      "loss": 0.3756,
+      "step": 1877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244602309269733e-05,
+      "loss": 0.345,
+      "step": 1878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24368175674585e-05,
+      "loss": 0.4148,
+      "step": 1879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242760689546837e-05,
+      "loss": 0.3573,
+      "step": 1880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241839107784403e-05,
+      "loss": 0.33,
+      "step": 1881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240917011570314e-05,
+      "loss": 0.3625,
+      "step": 1882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239994401016403e-05,
+      "loss": 0.4263,
+      "step": 1883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239071276234566e-05,
+      "loss": 0.4312,
+      "step": 1884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23814763733676e-05,
+      "loss": 0.3244,
+      "step": 1885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237223484435005e-05,
+      "loss": 0.4477,
+      "step": 1886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236298817641379e-05,
+      "loss": 0.3732,
+      "step": 1887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23537363706803e-05,
+      "loss": 0.4172,
+      "step": 1888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23444794282716e-05,
+      "loss": 0.4087,
+      "step": 1889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233521735031041e-05,
+      "loss": 0.3603,
+      "step": 1890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232595013792002e-05,
+      "loss": 0.4716,
+      "step": 1891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231667779222437e-05,
+      "loss": 0.3964,
+      "step": 1892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230740031434802e-05,
+      "loss": 0.405,
+      "step": 1893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229811770541612e-05,
+      "loss": 0.4217,
+      "step": 1894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228882996655448e-05,
+      "loss": 0.4712,
+      "step": 1895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227953709888954e-05,
+      "loss": 0.3822,
+      "step": 1896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22702391035483e-05,
+      "loss": 0.3377,
+      "step": 1897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226093598165849e-05,
+      "loss": 0.4444,
+      "step": 1898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225162773434832e-05,
+      "loss": 0.4139,
+      "step": 1899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224231436274673e-05,
+      "loss": 0.4526,
+      "step": 1900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223299586798326e-05,
+      "loss": 0.3891,
+      "step": 1901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222367225118807e-05,
+      "loss": 0.4258,
+      "step": 1902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221434351349188e-05,
+      "loss": 0.4797,
+      "step": 1903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220500965602613e-05,
+      "loss": 0.4075,
+      "step": 1904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219567067992283e-05,
+      "loss": 0.5022,
+      "step": 1905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.218632658631459e-05,
+      "loss": 0.4386,
+      "step": 1906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217697737633466e-05,
+      "loss": 0.3799,
+      "step": 1907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216762305111695e-05,
+      "loss": 0.3819,
+      "step": 1908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215826361179595e-05,
+      "loss": 0.4586,
+      "step": 1909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214889905950675e-05,
+      "loss": 0.3773,
+      "step": 1910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213952939538511e-05,
+      "loss": 0.4764,
+      "step": 1911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213015462056735e-05,
+      "loss": 0.3671,
+      "step": 1912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212077473619049e-05,
+      "loss": 0.4938,
+      "step": 1913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21113897433921e-05,
+      "loss": 0.4227,
+      "step": 1914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210199964331041e-05,
+      "loss": 0.4169,
+      "step": 1915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209260443708424e-05,
+      "loss": 0.3956,
+      "step": 1916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208320412585305e-05,
+      "loss": 0.4461,
+      "step": 1917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207379871075691e-05,
+      "loss": 0.4233,
+      "step": 1918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206438819293654e-05,
+      "loss": 0.3912,
+      "step": 1919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205497257353321e-05,
+      "loss": 0.4129,
+      "step": 1920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.204555185368886e-05,
+      "loss": 0.4844,
+      "step": 1921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203612603454604e-05,
+      "loss": 0.506,
+      "step": 1922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202669511724793e-05,
+      "loss": 0.4244,
+      "step": 1923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20172591029383e-05,
+      "loss": 0.3374,
+      "step": 1924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200781799276154e-05,
+      "loss": 0.3657,
+      "step": 1925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199837178786273e-05,
+      "loss": 0.3744,
+      "step": 1926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198892048938743e-05,
+      "loss": 0.3518,
+      "step": 1927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197946409848194e-05,
+      "loss": 0.4545,
+      "step": 1928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197000261629314e-05,
+      "loss": 0.4065,
+      "step": 1929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19605360439685e-05,
+      "loss": 0.5397,
+      "step": 1930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195106438265615e-05,
+      "loss": 0.3873,
+      "step": 1931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19415876335048e-05,
+      "loss": 0.5202,
+      "step": 1932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193210579766377e-05,
+      "loss": 0.346,
+      "step": 1933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19226188762831e-05,
+      "loss": 0.4468,
+      "step": 1934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191312687051328e-05,
+      "loss": 0.3152,
+      "step": 1935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190362978150555e-05,
+      "loss": 0.4109,
+      "step": 1936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189412761041172e-05,
+      "loss": 0.4875,
+      "step": 1937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188462035838419e-05,
+      "loss": 0.4222,
+      "step": 1938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187510802657602e-05,
+      "loss": 0.3476,
+      "step": 1939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186559061614086e-05,
+      "loss": 0.4932,
+      "step": 1940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185606812823299e-05,
+      "loss": 0.3915,
+      "step": 1941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184654056400732e-05,
+      "loss": 0.3575,
+      "step": 1942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183700792461933e-05,
+      "loss": 0.4113,
+      "step": 1943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182747021122515e-05,
+      "loss": 0.3872,
+      "step": 1944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181792742498152e-05,
+      "loss": 0.3911,
+      "step": 1945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180837956704579e-05,
+      "loss": 0.3961,
+      "step": 1946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179882663857593e-05,
+      "loss": 0.4515,
+      "step": 1947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17892686407305e-05,
+      "loss": 0.4911,
+      "step": 1948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177970557466872e-05,
+      "loss": 0.4929,
+      "step": 1949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177013744155043e-05,
+      "loss": 0.416,
+      "step": 1950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176056424253598e-05,
+      "loss": 0.4416,
+      "step": 1951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175098597878648e-05,
+      "loss": 0.4101,
+      "step": 1952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174140265146356e-05,
+      "loss": 0.3389,
+      "step": 1953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.173181426172949e-05,
+      "loss": 0.3217,
+      "step": 1954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172222081074716e-05,
+      "loss": 0.477,
+      "step": 1955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171262229968003e-05,
+      "loss": 0.4312,
+      "step": 1956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170301872969226e-05,
+      "loss": 0.3904,
+      "step": 1957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169341010194854e-05,
+      "loss": 0.4028,
+      "step": 1958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168379641761424e-05,
+      "loss": 0.4708,
+      "step": 1959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167417767785529e-05,
+      "loss": 0.3738,
+      "step": 1960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166455388383824e-05,
+      "loss": 0.4405,
+      "step": 1961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16549250367303e-05,
+      "loss": 0.3841,
+      "step": 1962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164529113769924e-05,
+      "loss": 0.3529,
+      "step": 1963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163565218791346e-05,
+      "loss": 0.3656,
+      "step": 1964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162600818854199e-05,
+      "loss": 0.4095,
+      "step": 1965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161635914075443e-05,
+      "loss": 0.3657,
+      "step": 1966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160670504572105e-05,
+      "loss": 0.4518,
+      "step": 1967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159704590461268e-05,
+      "loss": 0.4381,
+      "step": 1968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158738171860082e-05,
+      "loss": 0.392,
+      "step": 1969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15777124888575e-05,
+      "loss": 0.401,
+      "step": 1970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156803821655542e-05,
+      "loss": 0.3437,
+      "step": 1971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15583589028679e-05,
+      "loss": 0.3889,
+      "step": 1972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154867454896882e-05,
+      "loss": 0.4114,
+      "step": 1973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153898515603272e-05,
+      "loss": 0.3608,
+      "step": 1974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152929072523474e-05,
+      "loss": 0.4903,
+      "step": 1975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151959125775061e-05,
+      "loss": 0.4192,
+      "step": 1976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15098867547567e-05,
+      "loss": 0.457,
+      "step": 1977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150017721742994e-05,
+      "loss": 0.4013,
+      "step": 1978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149046264694794e-05,
+      "loss": 0.4616,
+      "step": 1979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148074304448888e-05,
+      "loss": 0.4029,
+      "step": 1980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147101841123153e-05,
+      "loss": 0.3366,
+      "step": 1981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146128874835536e-05,
+      "loss": 0.4089,
+      "step": 1982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145155405704032e-05,
+      "loss": 0.3153,
+      "step": 1983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144181433846707e-05,
+      "loss": 0.4075,
+      "step": 1984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143206959381684e-05,
+      "loss": 0.3982,
+      "step": 1985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142231982427147e-05,
+      "loss": 0.4153,
+      "step": 1986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141256503101342e-05,
+      "loss": 0.3915,
+      "step": 1987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140280521522576e-05,
+      "loss": 0.4686,
+      "step": 1988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139304037809216e-05,
+      "loss": 0.3891,
+      "step": 1989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138327052079689e-05,
+      "loss": 0.4643,
+      "step": 1990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137349564452486e-05,
+      "loss": 0.2922,
+      "step": 1991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136371575046157e-05,
+      "loss": 0.3156,
+      "step": 1992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135393083979311e-05,
+      "loss": 0.4561,
+      "step": 1993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134414091370623e-05,
+      "loss": 0.3816,
+      "step": 1994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13343459733882e-05,
+      "loss": 0.3224,
+      "step": 1995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132454602002701e-05,
+      "loss": 0.3597,
+      "step": 1996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131474105481117e-05,
+      "loss": 0.3843,
+      "step": 1997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130493107892986e-05,
+      "loss": 0.331,
+      "step": 1998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.12951160935728e-05,
+      "loss": 0.4223,
+      "step": 1999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128529609993037e-05,
+      "loss": 0.4278,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127547109919355e-05,
+      "loss": 0.3673,
+      "step": 2001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12656410925539e-05,
+      "loss": 0.46,
+      "step": 2002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125580608120364e-05,
+      "loss": 0.3364,
+      "step": 2003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124596606633551e-05,
+      "loss": 0.38,
+      "step": 2004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123612104914296e-05,
+      "loss": 0.5846,
+      "step": 2005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122627103081999e-05,
+      "loss": 0.3986,
+      "step": 2006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12164160125612e-05,
+      "loss": 0.2849,
+      "step": 2007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12065559955618e-05,
+      "loss": 0.5321,
+      "step": 2008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119669098101763e-05,
+      "loss": 0.5119,
+      "step": 2009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118682097012512e-05,
+      "loss": 0.3215,
+      "step": 2010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.117694596408133e-05,
+      "loss": 0.6262,
+      "step": 2011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116706596408386e-05,
+      "loss": 0.3754,
+      "step": 2012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115718097133099e-05,
+      "loss": 0.4283,
+      "step": 2013
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114729098702157e-05,
+      "loss": 0.4488,
+      "step": 2014
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.113739601235507e-05,
+      "loss": 0.3684,
+      "step": 2015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112749604853153e-05,
+      "loss": 0.4604,
+      "step": 2016
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111759109675164e-05,
+      "loss": 0.4366,
+      "step": 2017
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110768115821666e-05,
+      "loss": 0.3819,
+      "step": 2018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10977662341285e-05,
+      "loss": 0.3109,
+      "step": 2019
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108784632568961e-05,
+      "loss": 0.3891,
+      "step": 2020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107792143410311e-05,
+      "loss": 0.3457,
+      "step": 2021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106799156057269e-05,
+      "loss": 0.3187,
+      "step": 2022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105805670630263e-05,
+      "loss": 0.4626,
+      "step": 2023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104811687249783e-05,
+      "loss": 0.3924,
+      "step": 2024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103817206036382e-05,
+      "loss": 0.4272,
+      "step": 2025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10282222711067e-05,
+      "loss": 0.3971,
+      "step": 2026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101826750593318e-05,
+      "loss": 0.57,
+      "step": 2027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100830776605057e-05,
+      "loss": 0.4563,
+      "step": 2028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099834305266681e-05,
+      "loss": 0.3963,
+      "step": 2029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.098837336699042e-05,
+      "loss": 0.394,
+      "step": 2030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09783987102305e-05,
+      "loss": 0.4192,
+      "step": 2031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09684190835968e-05,
+      "loss": 0.4024,
+      "step": 2032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095843448829969e-05,
+      "loss": 0.3968,
+      "step": 2033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094844492555004e-05,
+      "loss": 0.3956,
+      "step": 2034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.093845039655942e-05,
+      "loss": 0.3776,
+      "step": 2035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092845090253995e-05,
+      "loss": 0.3705,
+      "step": 2036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091844644470439e-05,
+      "loss": 0.3711,
+      "step": 2037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09084370242661e-05,
+      "loss": 0.4926,
+      "step": 2038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0898422642439e-05,
+      "loss": 0.3191,
+      "step": 2039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.088840330043764e-05,
+      "loss": 0.3135,
+      "step": 2040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087837899947718e-05,
+      "loss": 0.4921,
+      "step": 2041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086834974077335e-05,
+      "loss": 0.356,
+      "step": 2042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085831552554253e-05,
+      "loss": 0.3713,
+      "step": 2043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.084827635500164e-05,
+      "loss": 0.3842,
+      "step": 2044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.083823223036826e-05,
+      "loss": 0.414,
+      "step": 2045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082818315286055e-05,
+      "loss": 0.3849,
+      "step": 2046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081812912369725e-05,
+      "loss": 0.3936,
+      "step": 2047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080807014409771e-05,
+      "loss": 0.3688,
+      "step": 2048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.079800621528189e-05,
+      "loss": 0.3573,
+      "step": 2049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.078793733847036e-05,
+      "loss": 0.4142,
+      "step": 2050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077786351488427e-05,
+      "loss": 0.3992,
+      "step": 2051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076778474574538e-05,
+      "loss": 0.4469,
+      "step": 2052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075770103227604e-05,
+      "loss": 0.4005,
+      "step": 2053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07476123756992e-05,
+      "loss": 0.3968,
+      "step": 2054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073751877723842e-05,
+      "loss": 0.3834,
+      "step": 2055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072742023811786e-05,
+      "loss": 0.4814,
+      "step": 2056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071731675956227e-05,
+      "loss": 0.4491,
+      "step": 2057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.070720834279702e-05,
+      "loss": 0.3553,
+      "step": 2058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069709498904804e-05,
+      "loss": 0.3706,
+      "step": 2059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068697669954189e-05,
+      "loss": 0.3897,
+      "step": 2060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.067685347550572e-05,
+      "loss": 0.2552,
+      "step": 2061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066672531816726e-05,
+      "loss": 0.3696,
+      "step": 2062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06565922287549e-05,
+      "loss": 0.3673,
+      "step": 2063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064645420849753e-05,
+      "loss": 0.4162,
+      "step": 2064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063631125862475e-05,
+      "loss": 0.4862,
+      "step": 2065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062616338036665e-05,
+      "loss": 0.3697,
+      "step": 2066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061601057495399e-05,
+      "loss": 0.4271,
+      "step": 2067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060585284361809e-05,
+      "loss": 0.4702,
+      "step": 2068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059569018759091e-05,
+      "loss": 0.382,
+      "step": 2069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058552260810497e-05,
+      "loss": 0.4177,
+      "step": 2070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057535010639339e-05,
+      "loss": 0.4438,
+      "step": 2071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05651726836899e-05,
+      "loss": 0.4098,
+      "step": 2072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055499034122881e-05,
+      "loss": 0.4871,
+      "step": 2073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054480308024506e-05,
+      "loss": 0.4199,
+      "step": 2074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053461090197415e-05,
+      "loss": 0.43,
+      "step": 2075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052441380765217e-05,
+      "loss": 0.4404,
+      "step": 2076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051421179851588e-05,
+      "loss": 0.3513,
+      "step": 2077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050400487580255e-05,
+      "loss": 0.3987,
+      "step": 2078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049379304075007e-05,
+      "loss": 0.4258,
+      "step": 2079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048357629459694e-05,
+      "loss": 0.438,
+      "step": 2080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047335463858227e-05,
+      "loss": 0.3785,
+      "step": 2081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046312807394574e-05,
+      "loss": 0.4268,
+      "step": 2082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045289660192761e-05,
+      "loss": 0.3258,
+      "step": 2083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044266022376878e-05,
+      "loss": 0.4022,
+      "step": 2084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043241894071069e-05,
+      "loss": 0.397,
+      "step": 2085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042217275399544e-05,
+      "loss": 0.2881,
+      "step": 2086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.041192166486568e-05,
+      "loss": 0.3403,
+      "step": 2087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040166567456467e-05,
+      "loss": 0.4442,
+      "step": 2088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039140478433625e-05,
+      "loss": 0.3286,
+      "step": 2089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038113899542486e-05,
+      "loss": 0.3614,
+      "step": 2090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037086830907555e-05,
+      "loss": 0.4866,
+      "step": 2091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.036059272653396e-05,
+      "loss": 0.448,
+      "step": 2092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.035031224904628e-05,
+      "loss": 0.3812,
+      "step": 2093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034002687785939e-05,
+      "loss": 0.4226,
+      "step": 2094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032973661422064e-05,
+      "loss": 0.3684,
+      "step": 2095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031944145937807e-05,
+      "loss": 0.381,
+      "step": 2096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.030914141458027e-05,
+      "loss": 0.4162,
+      "step": 2097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029883648107644e-05,
+      "loss": 0.3465,
+      "step": 2098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028852666011638e-05,
+      "loss": 0.4194,
+      "step": 2099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027821195295044e-05,
+      "loss": 0.3715,
+      "step": 2100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026789236082959e-05,
+      "loss": 0.4427,
+      "step": 2101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025756788500542e-05,
+      "loss": 0.3828,
+      "step": 2102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024723852673004e-05,
+      "loss": 0.35,
+      "step": 2103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023690428725626e-05,
+      "loss": 0.3869,
+      "step": 2104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02265651678374e-05,
+      "loss": 0.4531,
+      "step": 2105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021622116972737e-05,
+      "loss": 0.3621,
+      "step": 2106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020587229418069e-05,
+      "loss": 0.3832,
+      "step": 2107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01955185424525e-05,
+      "loss": 0.4815,
+      "step": 2108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01851599157985e-05,
+      "loss": 0.3887,
+      "step": 2109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017479641547499e-05,
+      "loss": 0.4074,
+      "step": 2110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016442804273884e-05,
+      "loss": 0.3628,
+      "step": 2111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015405479884754e-05,
+      "loss": 0.4173,
+      "step": 2112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.014367668505916e-05,
+      "loss": 0.4427,
+      "step": 2113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013329370263239e-05,
+      "loss": 0.3165,
+      "step": 2114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.012290585282644e-05,
+      "loss": 0.2724,
+      "step": 2115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011251313690117e-05,
+      "loss": 0.4616,
+      "step": 2116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0102115556117e-05,
+      "loss": 0.3889,
+      "step": 2117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009171311173496e-05,
+      "loss": 0.3856,
+      "step": 2118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008130580501668e-05,
+      "loss": 0.3977,
+      "step": 2119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007089363722434e-05,
+      "loss": 0.4288,
+      "step": 2120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006047660962072e-05,
+      "loss": 0.4304,
+      "step": 2121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005005472346924e-05,
+      "loss": 0.4289,
+      "step": 2122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003962798003384e-05,
+      "loss": 0.3195,
+      "step": 2123
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002919638057908e-05,
+      "loss": 0.3699,
+      "step": 2124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.001875992637011e-05,
+      "loss": 0.4505,
+      "step": 2125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000831861867267e-05,
+      "loss": 0.4452,
+      "step": 2126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999787245875308e-05,
+      "loss": 0.4339,
+      "step": 2127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998742144787828e-05,
+      "loss": 0.3798,
+      "step": 2128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997696558731575e-05,
+      "loss": 0.2968,
+      "step": 2129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996650487833356e-05,
+      "loss": 0.341,
+      "step": 2130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.995603932220042e-05,
+      "loss": 0.431,
+      "step": 2131
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99455689201856e-05,
+      "loss": 0.3111,
+      "step": 2132
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993509367355893e-05,
+      "loss": 0.4192,
+      "step": 2133
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.992461358359084e-05,
+      "loss": 0.3353,
+      "step": 2134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99141286515524e-05,
+      "loss": 0.4192,
+      "step": 2135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99036388787152e-05,
+      "loss": 0.4657,
+      "step": 2136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989314426635144e-05,
+      "loss": 0.3187,
+      "step": 2137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988264481573393e-05,
+      "loss": 0.4089,
+      "step": 2138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987214052813604e-05,
+      "loss": 0.4656,
+      "step": 2139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986163140483172e-05,
+      "loss": 0.4669,
+      "step": 2140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985111744709555e-05,
+      "loss": 0.3808,
+      "step": 2141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984059865620262e-05,
+      "loss": 0.3472,
+      "step": 2142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983007503342869e-05,
+      "loss": 0.4209,
+      "step": 2143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981954658005005e-05,
+      "loss": 0.4283,
+      "step": 2144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980901329734361e-05,
+      "loss": 0.4028,
+      "step": 2145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979847518658683e-05,
+      "loss": 0.4017,
+      "step": 2146
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978793224905779e-05,
+      "loss": 0.4408,
+      "step": 2147
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977738448603514e-05,
+      "loss": 0.4132,
+      "step": 2148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97668318987981e-05,
+      "loss": 0.3778,
+      "step": 2149
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975627448862653e-05,
+      "loss": 0.3544,
+      "step": 2150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97457122568008e-05,
+      "loss": 0.2978,
+      "step": 2151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97351452046019e-05,
+      "loss": 0.4159,
+      "step": 2152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972457333331143e-05,
+      "loss": 0.4566,
+      "step": 2153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971399664421152e-05,
+      "loss": 0.4021,
+      "step": 2154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.970341513858496e-05,
+      "loss": 0.3808,
+      "step": 2155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969282881771503e-05,
+      "loss": 0.3659,
+      "step": 2156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968223768288566e-05,
+      "loss": 0.3339,
+      "step": 2157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967164173538136e-05,
+      "loss": 0.3558,
+      "step": 2158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96610409764872e-05,
+      "loss": 0.365,
+      "step": 2159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965043540748884e-05,
+      "loss": 0.3532,
+      "step": 2160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963982502967254e-05,
+      "loss": 0.3247,
+      "step": 2161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96292098443251e-05,
+      "loss": 0.4156,
+      "step": 2162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961858985273397e-05,
+      "loss": 0.438,
+      "step": 2163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960796505618711e-05,
+      "loss": 0.3605,
+      "step": 2164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959733545597313e-05,
+      "loss": 0.3809,
+      "step": 2165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958670105338118e-05,
+      "loss": 0.435,
+      "step": 2166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9576061849701e-05,
+      "loss": 0.3793,
+      "step": 2167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956541784622293e-05,
+      "loss": 0.483,
+      "step": 2168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955476904423785e-05,
+      "loss": 0.2925,
+      "step": 2169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954411544503729e-05,
+      "loss": 0.3361,
+      "step": 2170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.953345704991328e-05,
+      "loss": 0.3403,
+      "step": 2171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95227938601585e-05,
+      "loss": 0.3211,
+      "step": 2172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951212587706618e-05,
+      "loss": 0.3414,
+      "step": 2173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950145310193014e-05,
+      "loss": 0.4707,
+      "step": 2174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949077553604477e-05,
+      "loss": 0.3964,
+      "step": 2175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948009318070504e-05,
+      "loss": 0.3127,
+      "step": 2176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946940603720654e-05,
+      "loss": 0.4097,
+      "step": 2177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945871410684539e-05,
+      "loss": 0.4151,
+      "step": 2178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944801739091831e-05,
+      "loss": 0.3786,
+      "step": 2179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94373158907226e-05,
+      "loss": 0.4378,
+      "step": 2180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942660960755613e-05,
+      "loss": 0.2807,
+      "step": 2181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941589854271739e-05,
+      "loss": 0.4037,
+      "step": 2182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940518269750541e-05,
+      "loss": 0.365,
+      "step": 2183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939446207321982e-05,
+      "loss": 0.3889,
+      "step": 2184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.93837366711608e-05,
+      "loss": 0.4957,
+      "step": 2185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937300649262915e-05,
+      "loss": 0.3129,
+      "step": 2186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936227153892621e-05,
+      "loss": 0.3725,
+      "step": 2187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.935153181135391e-05,
+      "loss": 0.4057,
+      "step": 2188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934078731121482e-05,
+      "loss": 0.3792,
+      "step": 2189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.9330038039812e-05,
+      "loss": 0.5201,
+      "step": 2190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.931928399844911e-05,
+      "loss": 0.3572,
+      "step": 2191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930852518843043e-05,
+      "loss": 0.4381,
+      "step": 2192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92977616110608e-05,
+      "loss": 0.3194,
+      "step": 2193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92869932676456e-05,
+      "loss": 0.3928,
+      "step": 2194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927622015949082e-05,
+      "loss": 0.4629,
+      "step": 2195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926544228790306e-05,
+      "loss": 0.3685,
+      "step": 2196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.925465965418945e-05,
+      "loss": 0.3603,
+      "step": 2197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.924387225965768e-05,
+      "loss": 0.3854,
+      "step": 2198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923308010561609e-05,
+      "loss": 0.4339,
+      "step": 2199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922228319337353e-05,
+      "loss": 0.3848,
+      "step": 2200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921148152423946e-05,
+      "loss": 0.398,
+      "step": 2201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920067509952392e-05,
+      "loss": 0.4445,
+      "step": 2202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91898639205375e-05,
+      "loss": 0.3594,
+      "step": 2203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917904798859141e-05,
+      "loss": 0.4015,
+      "step": 2204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916822730499737e-05,
+      "loss": 0.3651,
+      "step": 2205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915740187106776e-05,
+      "loss": 0.4296,
+      "step": 2206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914657168811544e-05,
+      "loss": 0.3433,
+      "step": 2207
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.913573675745396e-05,
+      "loss": 0.3973,
+      "step": 2208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912489708039733e-05,
+      "loss": 0.3994,
+      "step": 2209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911405265826023e-05,
+      "loss": 0.3694,
+      "step": 2210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910320349235786e-05,
+      "loss": 0.3895,
+      "step": 2211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.909234958400598e-05,
+      "loss": 0.4336,
+      "step": 2212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.9081490934521e-05,
+      "loss": 0.4675,
+      "step": 2213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907062754521985e-05,
+      "loss": 0.3426,
+      "step": 2214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905975941742004e-05,
+      "loss": 0.3723,
+      "step": 2215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904888655243966e-05,
+      "loss": 0.4259,
+      "step": 2216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903800895159737e-05,
+      "loss": 0.3669,
+      "step": 2217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902712661621241e-05,
+      "loss": 0.3567,
+      "step": 2218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90162395476046e-05,
+      "loss": 0.4604,
+      "step": 2219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900534774709432e-05,
+      "loss": 0.3561,
+      "step": 2220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899445121600255e-05,
+      "loss": 0.3539,
+      "step": 2221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898354995565077e-05,
+      "loss": 0.3553,
+      "step": 2222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.897264396736117e-05,
+      "loss": 0.3516,
+      "step": 2223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896173325245637e-05,
+      "loss": 0.3502,
+      "step": 2224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895081781225965e-05,
+      "loss": 0.428,
+      "step": 2225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893989764809484e-05,
+      "loss": 0.3782,
+      "step": 2226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892897276128633e-05,
+      "loss": 0.351,
+      "step": 2227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89180431531591e-05,
+      "loss": 0.4147,
+      "step": 2228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89071088250387e-05,
+      "loss": 0.4115,
+      "step": 2229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889616977825124e-05,
+      "loss": 0.4071,
+      "step": 2230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88852260141234e-05,
+      "loss": 0.3739,
+      "step": 2231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.887427753398248e-05,
+      "loss": 0.4073,
+      "step": 2232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886332433915627e-05,
+      "loss": 0.3734,
+      "step": 2233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885236643097322e-05,
+      "loss": 0.3559,
+      "step": 2234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884140381076229e-05,
+      "loss": 0.4196,
+      "step": 2235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883043647985304e-05,
+      "loss": 0.4561,
+      "step": 2236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881946443957555e-05,
+      "loss": 0.4282,
+      "step": 2237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880848769126055e-05,
+      "loss": 0.3928,
+      "step": 2238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879750623623933e-05,
+      "loss": 0.4155,
+      "step": 2239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.878652007584366e-05,
+      "loss": 0.3888,
+      "step": 2240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8775529211406e-05,
+      "loss": 0.3612,
+      "step": 2241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876453364425928e-05,
+      "loss": 0.4061,
+      "step": 2242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875353337573709e-05,
+      "loss": 0.4389,
+      "step": 2243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874252840717351e-05,
+      "loss": 0.4163,
+      "step": 2244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873151873990325e-05,
+      "loss": 0.3789,
+      "step": 2245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872050437526156e-05,
+      "loss": 0.426,
+      "step": 2246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870948531458425e-05,
+      "loss": 0.3499,
+      "step": 2247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869846155920775e-05,
+      "loss": 0.3741,
+      "step": 2248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868743311046898e-05,
+      "loss": 0.3732,
+      "step": 2249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.867639996970551e-05,
+      "loss": 0.3044,
+      "step": 2250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866536213825544e-05,
+      "loss": 0.3954,
+      "step": 2251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865431961745743e-05,
+      "loss": 0.3768,
+      "step": 2252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864327240865072e-05,
+      "loss": 0.3748,
+      "step": 2253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86322205131751e-05,
+      "loss": 0.521,
+      "step": 2254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862116393237099e-05,
+      "loss": 0.3776,
+      "step": 2255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861010266757932e-05,
+      "loss": 0.441,
+      "step": 2256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859903672014159e-05,
+      "loss": 0.4639,
+      "step": 2257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85879660913999e-05,
+      "loss": 0.2986,
+      "step": 2258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.857689078269687e-05,
+      "loss": 0.4298,
+      "step": 2259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.856581079537576e-05,
+      "loss": 0.3519,
+      "step": 2260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855472613078032e-05,
+      "loss": 0.4458,
+      "step": 2261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.854363679025491e-05,
+      "loss": 0.4354,
+      "step": 2262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853254277514446e-05,
+      "loss": 0.3696,
+      "step": 2263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852144408679445e-05,
+      "loss": 0.3808,
+      "step": 2264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851034072655093e-05,
+      "loss": 0.5106,
+      "step": 2265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.849923269576052e-05,
+      "loss": 0.3942,
+      "step": 2266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848811999577041e-05,
+      "loss": 0.3885,
+      "step": 2267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.847700262792835e-05,
+      "loss": 0.4116,
+      "step": 2268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846588059358265e-05,
+      "loss": 0.4096,
+      "step": 2269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84547538940822e-05,
+      "loss": 0.3959,
+      "step": 2270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844362253077645e-05,
+      "loss": 0.3425,
+      "step": 2271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843248650501543e-05,
+      "loss": 0.4432,
+      "step": 2272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84213458181497e-05,
+      "loss": 0.3497,
+      "step": 2273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84102004715304e-05,
+      "loss": 0.4632,
+      "step": 2274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839905046650925e-05,
+      "loss": 0.4736,
+      "step": 2275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838789580443854e-05,
+      "loss": 0.4156,
+      "step": 2276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837673648667111e-05,
+      "loss": 0.3988,
+      "step": 2277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.836557251456035e-05,
+      "loss": 0.3617,
+      "step": 2278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835440388946024e-05,
+      "loss": 0.4002,
+      "step": 2279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834323061272529e-05,
+      "loss": 0.4091,
+      "step": 2280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.833205268571065e-05,
+      "loss": 0.4212,
+      "step": 2281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832087010977193e-05,
+      "loss": 0.3429,
+      "step": 2282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83096828862654e-05,
+      "loss": 0.3107,
+      "step": 2283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.829849101654782e-05,
+      "loss": 0.3218,
+      "step": 2284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828729450197655e-05,
+      "loss": 0.3551,
+      "step": 2285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827609334390951e-05,
+      "loss": 0.2897,
+      "step": 2286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826488754370519e-05,
+      "loss": 0.4638,
+      "step": 2287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825367710272263e-05,
+      "loss": 0.4547,
+      "step": 2288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.824246202232142e-05,
+      "loss": 0.4028,
+      "step": 2289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823124230386172e-05,
+      "loss": 0.3145,
+      "step": 2290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822001794870431e-05,
+      "loss": 0.3379,
+      "step": 2291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820878895821044e-05,
+      "loss": 0.3028,
+      "step": 2292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819755533374199e-05,
+      "loss": 0.4028,
+      "step": 2293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818631707666135e-05,
+      "loss": 0.3676,
+      "step": 2294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817507418833153e-05,
+      "loss": 0.4016,
+      "step": 2295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816382667011604e-05,
+      "loss": 0.3565,
+      "step": 2296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815257452337902e-05,
+      "loss": 0.3867,
+      "step": 2297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.814131774948511e-05,
+      "loss": 0.3681,
+      "step": 2298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.813005634979954e-05,
+      "loss": 0.3313,
+      "step": 2299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.811879032568808e-05,
+      "loss": 0.3117,
+      "step": 2300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81075196785171e-05,
+      "loss": 0.398,
+      "step": 2301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80962444096535e-05,
+      "loss": 0.3873,
+      "step": 2302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808496452046476e-05,
+      "loss": 0.4351,
+      "step": 2303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80736800123189e-05,
+      "loss": 0.298,
+      "step": 2304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806239088658449e-05,
+      "loss": 0.3835,
+      "step": 2305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80510971446307e-05,
+      "loss": 0.3652,
+      "step": 2306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.803979878782724e-05,
+      "loss": 0.3611,
+      "step": 2307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802849581754437e-05,
+      "loss": 0.4353,
+      "step": 2308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801718823515292e-05,
+      "loss": 0.3641,
+      "step": 2309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800587604202428e-05,
+      "loss": 0.3884,
+      "step": 2310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79945592395304e-05,
+      "loss": 0.4269,
+      "step": 2311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798323782904379e-05,
+      "loss": 0.3961,
+      "step": 2312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797191181193748e-05,
+      "loss": 0.4063,
+      "step": 2313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.796058118958512e-05,
+      "loss": 0.4526,
+      "step": 2314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79492459633609e-05,
+      "loss": 0.35,
+      "step": 2315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793790613463955e-05,
+      "loss": 0.325,
+      "step": 2316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.792656170479636e-05,
+      "loss": 0.3506,
+      "step": 2317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79152126752072e-05,
+      "loss": 0.3074,
+      "step": 2318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.790385904724847e-05,
+      "loss": 0.3527,
+      "step": 2319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789250082229716e-05,
+      "loss": 0.4245,
+      "step": 2320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788113800173079e-05,
+      "loss": 0.3991,
+      "step": 2321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786977058692745e-05,
+      "loss": 0.4588,
+      "step": 2322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785839857926575e-05,
+      "loss": 0.4193,
+      "step": 2323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784702198012496e-05,
+      "loss": 0.3935,
+      "step": 2324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.783564079088477e-05,
+      "loss": 0.3309,
+      "step": 2325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782425501292552e-05,
+      "loss": 0.335,
+      "step": 2326
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78128646476281e-05,
+      "loss": 0.3718,
+      "step": 2327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.780146969637391e-05,
+      "loss": 0.3578,
+      "step": 2328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779007016054495e-05,
+      "loss": 0.3565,
+      "step": 2329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.777866604152376e-05,
+      "loss": 0.27,
+      "step": 2330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776725734069342e-05,
+      "loss": 0.3451,
+      "step": 2331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77558440594376e-05,
+      "loss": 0.4655,
+      "step": 2332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774442619914048e-05,
+      "loss": 0.3038,
+      "step": 2333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.773300376118683e-05,
+      "loss": 0.4056,
+      "step": 2334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772157674696199e-05,
+      "loss": 0.3143,
+      "step": 2335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771014515785183e-05,
+      "loss": 0.4373,
+      "step": 2336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769870899524275e-05,
+      "loss": 0.4737,
+      "step": 2337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768726826052175e-05,
+      "loss": 0.319,
+      "step": 2338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.767582295507636e-05,
+      "loss": 0.3575,
+      "step": 2339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766437308029466e-05,
+      "loss": 0.3927,
+      "step": 2340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765291863756533e-05,
+      "loss": 0.2986,
+      "step": 2341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764145962827755e-05,
+      "loss": 0.3506,
+      "step": 2342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762999605382104e-05,
+      "loss": 0.339,
+      "step": 2343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761852791558615e-05,
+      "loss": 0.3075,
+      "step": 2344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.760705521496374e-05,
+      "loss": 0.3707,
+      "step": 2345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759557795334518e-05,
+      "loss": 0.4173,
+      "step": 2346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758409613212247e-05,
+      "loss": 0.4014,
+      "step": 2347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.757260975268813e-05,
+      "loss": 0.267,
+      "step": 2348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75611188164352e-05,
+      "loss": 0.3703,
+      "step": 2349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754962332475735e-05,
+      "loss": 0.4157,
+      "step": 2350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753812327904873e-05,
+      "loss": 0.3425,
+      "step": 2351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752661868070408e-05,
+      "loss": 0.3748,
+      "step": 2352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751510953111868e-05,
+      "loss": 0.3473,
+      "step": 2353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750359583168836e-05,
+      "loss": 0.3934,
+      "step": 2354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749207758380949e-05,
+      "loss": 0.3931,
+      "step": 2355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748055478887904e-05,
+      "loss": 0.3284,
+      "step": 2356
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746902744829446e-05,
+      "loss": 0.3755,
+      "step": 2357
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745749556345383e-05,
+      "loss": 0.3785,
+      "step": 2358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744595913575571e-05,
+      "loss": 0.3894,
+      "step": 2359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.743441816659925e-05,
+      "loss": 0.4034,
+      "step": 2360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742287265738416e-05,
+      "loss": 0.4325,
+      "step": 2361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.741132260951065e-05,
+      "loss": 0.3683,
+      "step": 2362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739976802437955e-05,
+      "loss": 0.3842,
+      "step": 2363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738820890339217e-05,
+      "loss": 0.2414,
+      "step": 2364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.737664524795042e-05,
+      "loss": 0.3852,
+      "step": 2365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736507705945676e-05,
+      "loss": 0.3791,
+      "step": 2366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735350433931416e-05,
+      "loss": 0.4381,
+      "step": 2367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.734192708892617e-05,
+      "loss": 0.4044,
+      "step": 2368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733034530969688e-05,
+      "loss": 0.4079,
+      "step": 2369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731875900303094e-05,
+      "loss": 0.3783,
+      "step": 2370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.730716817033353e-05,
+      "loss": 0.4866,
+      "step": 2371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72955728130104e-05,
+      "loss": 0.4048,
+      "step": 2372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728397293246785e-05,
+      "loss": 0.2844,
+      "step": 2373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.727236853011269e-05,
+      "loss": 0.4282,
+      "step": 2374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726075960735232e-05,
+      "loss": 0.4258,
+      "step": 2375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.724914616559467e-05,
+      "loss": 0.3579,
+      "step": 2376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723752820624824e-05,
+      "loss": 0.397,
+      "step": 2377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722590573072202e-05,
+      "loss": 0.4334,
+      "step": 2378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.721427874042562e-05,
+      "loss": 0.3371,
+      "step": 2379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720264723676918e-05,
+      "loss": 0.3187,
+      "step": 2380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719101122116333e-05,
+      "loss": 0.44,
+      "step": 2381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717937069501932e-05,
+      "loss": 0.3944,
+      "step": 2382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71677256597489e-05,
+      "loss": 0.4017,
+      "step": 2383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71560761167644e-05,
+      "loss": 0.3422,
+      "step": 2384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714442206747865e-05,
+      "loss": 0.432,
+      "step": 2385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71327635133051e-05,
+      "loss": 0.4463,
+      "step": 2386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.712110045565768e-05,
+      "loss": 0.3731,
+      "step": 2387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710943289595088e-05,
+      "loss": 0.3813,
+      "step": 2388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709776083559977e-05,
+      "loss": 0.3549,
+      "step": 2389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708608427601992e-05,
+      "loss": 0.4153,
+      "step": 2390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.707440321862747e-05,
+      "loss": 0.3944,
+      "step": 2391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706271766483911e-05,
+      "loss": 0.4015,
+      "step": 2392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.705102761607207e-05,
+      "loss": 0.4597,
+      "step": 2393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703933307374413e-05,
+      "loss": 0.3871,
+      "step": 2394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702763403927356e-05,
+      "loss": 0.331,
+      "step": 2395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70159305140793e-05,
+      "loss": 0.3895,
+      "step": 2396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700422249958069e-05,
+      "loss": 0.3454,
+      "step": 2397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699250999719772e-05,
+      "loss": 0.4473,
+      "step": 2398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.698079300835086e-05,
+      "loss": 0.3446,
+      "step": 2399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696907153446119e-05,
+      "loss": 0.4301,
+      "step": 2400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695734557695025e-05,
+      "loss": 0.3835,
+      "step": 2401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69456151372402e-05,
+      "loss": 0.3513,
+      "step": 2402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.693388021675368e-05,
+      "loss": 0.2772,
+      "step": 2403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692214081691395e-05,
+      "loss": 0.3971,
+      "step": 2404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691039693914473e-05,
+      "loss": 0.3084,
+      "step": 2405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.689864858487032e-05,
+      "loss": 0.2857,
+      "step": 2406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688689575551559e-05,
+      "loss": 0.3461,
+      "step": 2407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.687513845250591e-05,
+      "loss": 0.4408,
+      "step": 2408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686337667726723e-05,
+      "loss": 0.3784,
+      "step": 2409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685161043122598e-05,
+      "loss": 0.352,
+      "step": 2410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683983971580923e-05,
+      "loss": 0.3137,
+      "step": 2411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.682806453244448e-05,
+      "loss": 0.4584,
+      "step": 2412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681628488255986e-05,
+      "loss": 0.3546,
+      "step": 2413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.680450076758402e-05,
+      "loss": 0.4092,
+      "step": 2414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679271218894611e-05,
+      "loss": 0.3928,
+      "step": 2415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67809191480759e-05,
+      "loss": 0.377,
+      "step": 2416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676912164640358e-05,
+      "loss": 0.4449,
+      "step": 2417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675731968536002e-05,
+      "loss": 0.4471,
+      "step": 2418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.674551326637654e-05,
+      "loss": 0.4297,
+      "step": 2419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673370239088504e-05,
+      "loss": 0.3591,
+      "step": 2420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672188706031791e-05,
+      "loss": 0.3823,
+      "step": 2421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.671006727610817e-05,
+      "loss": 0.3187,
+      "step": 2422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66982430396893e-05,
+      "loss": 0.4048,
+      "step": 2423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.668641435249534e-05,
+      "loss": 0.3081,
+      "step": 2424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667458121596088e-05,
+      "loss": 0.4063,
+      "step": 2425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666274363152107e-05,
+      "loss": 0.3577,
+      "step": 2426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.665090160061156e-05,
+      "loss": 0.3523,
+      "step": 2427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.663905512466854e-05,
+      "loss": 0.3984,
+      "step": 2428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662720420512877e-05,
+      "loss": 0.3415,
+      "step": 2429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.661534884342954e-05,
+      "loss": 0.435,
+      "step": 2430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660348904100864e-05,
+      "loss": 0.2779,
+      "step": 2431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659162479930448e-05,
+      "loss": 0.3374,
+      "step": 2432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657975611975593e-05,
+      "loss": 0.3372,
+      "step": 2433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.656788300380243e-05,
+      "loss": 0.3763,
+      "step": 2434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655600545288396e-05,
+      "loss": 0.363,
+      "step": 2435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654412346844103e-05,
+      "loss": 0.3954,
+      "step": 2436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653223705191468e-05,
+      "loss": 0.4146,
+      "step": 2437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.652034620474652e-05,
+      "loss": 0.37,
+      "step": 2438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650845092837867e-05,
+      "loss": 0.4001,
+      "step": 2439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649655122425378e-05,
+      "loss": 0.3687,
+      "step": 2440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648464709381505e-05,
+      "loss": 0.3267,
+      "step": 2441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647273853850625e-05,
+      "loss": 0.4289,
+      "step": 2442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646082555977162e-05,
+      "loss": 0.4651,
+      "step": 2443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644890815905597e-05,
+      "loss": 0.2983,
+      "step": 2444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.643698633780466e-05,
+      "loss": 0.3922,
+      "step": 2445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642506009746357e-05,
+      "loss": 0.342,
+      "step": 2446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641312943947913e-05,
+      "loss": 0.3813,
+      "step": 2447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.640119436529826e-05,
+      "loss": 0.3314,
+      "step": 2448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.638925487636848e-05,
+      "loss": 0.3129,
+      "step": 2449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.637731097413779e-05,
+      "loss": 0.4191,
+      "step": 2450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63653626600548e-05,
+      "loss": 0.3846,
+      "step": 2451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.635340993556854e-05,
+      "loss": 0.352,
+      "step": 2452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634145280212868e-05,
+      "loss": 0.3905,
+      "step": 2453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632949126118538e-05,
+      "loss": 0.4298,
+      "step": 2454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631752531418934e-05,
+      "loss": 0.4916,
+      "step": 2455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.630555496259179e-05,
+      "loss": 0.3841,
+      "step": 2456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62935802078445e-05,
+      "loss": 0.3493,
+      "step": 2457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.628160105139977e-05,
+      "loss": 0.3847,
+      "step": 2458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626961749471043e-05,
+      "loss": 0.3453,
+      "step": 2459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625762953922987e-05,
+      "loss": 0.4353,
+      "step": 2460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624563718641196e-05,
+      "loss": 0.3935,
+      "step": 2461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623364043771118e-05,
+      "loss": 0.3867,
+      "step": 2462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.622163929458245e-05,
+      "loss": 0.3133,
+      "step": 2463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.620963375848132e-05,
+      "loss": 0.3691,
+      "step": 2464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61976238308638e-05,
+      "loss": 0.3011,
+      "step": 2465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618560951318647e-05,
+      "loss": 0.3238,
+      "step": 2466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617359080690638e-05,
+      "loss": 0.3256,
+      "step": 2467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.616156771348125e-05,
+      "loss": 0.3553,
+      "step": 2468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614954023436918e-05,
+      "loss": 0.3474,
+      "step": 2469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.613750837102891e-05,
+      "loss": 0.3424,
+      "step": 2470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.612547212491964e-05,
+      "loss": 0.4234,
+      "step": 2471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611343149750112e-05,
+      "loss": 0.4156,
+      "step": 2472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610138649023366e-05,
+      "loss": 0.38,
+      "step": 2473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608933710457807e-05,
+      "loss": 0.4402,
+      "step": 2474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607728334199574e-05,
+      "loss": 0.3748,
+      "step": 2475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60652252039485e-05,
+      "loss": 0.3158,
+      "step": 2476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605316269189881e-05,
+      "loss": 0.3583,
+      "step": 2477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60410958073096e-05,
+      "loss": 0.3926,
+      "step": 2478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602902455164433e-05,
+      "loss": 0.3532,
+      "step": 2479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.6016948926367e-05,
+      "loss": 0.3678,
+      "step": 2480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60048689329422e-05,
+      "loss": 0.3428,
+      "step": 2481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599278457283492e-05,
+      "loss": 0.4155,
+      "step": 2482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.598069584751084e-05,
+      "loss": 0.3569,
+      "step": 2483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.596860275843602e-05,
+      "loss": 0.4015,
+      "step": 2484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.595650530707713e-05,
+      "loss": 0.3269,
+      "step": 2485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594440349490137e-05,
+      "loss": 0.3285,
+      "step": 2486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.593229732337643e-05,
+      "loss": 0.3797,
+      "step": 2487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592018679397055e-05,
+      "loss": 0.3122,
+      "step": 2488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.590807190815253e-05,
+      "loss": 0.4196,
+      "step": 2489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.589595266739163e-05,
+      "loss": 0.344,
+      "step": 2490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588382907315768e-05,
+      "loss": 0.3577,
+      "step": 2491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.587170112692107e-05,
+      "loss": 0.2273,
+      "step": 2492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585956883015265e-05,
+      "loss": 0.3023,
+      "step": 2493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.584743218432383e-05,
+      "loss": 0.3272,
+      "step": 2494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583529119090655e-05,
+      "loss": 0.3248,
+      "step": 2495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582314585137328e-05,
+      "loss": 0.3808,
+      "step": 2496
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581099616719698e-05,
+      "loss": 0.3378,
+      "step": 2497
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579884213985123e-05,
+      "loss": 0.4212,
+      "step": 2498
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.578668377081001e-05,
+      "loss": 0.3859,
+      "step": 2499
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577452106154794e-05,
+      "loss": 0.4318,
+      "step": 2500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576235401354009e-05,
+      "loss": 0.3967,
+      "step": 2501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.575018262826209e-05,
+      "loss": 0.3762,
+      "step": 2502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57380069071901e-05,
+      "loss": 0.462,
+      "step": 2503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.572582685180076e-05,
+      "loss": 0.3488,
+      "step": 2504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571364246357131e-05,
+      "loss": 0.3847,
+      "step": 2505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.570145374397947e-05,
+      "loss": 0.3905,
+      "step": 2506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.568926069450347e-05,
+      "loss": 0.3222,
+      "step": 2507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567706331662211e-05,
+      "loss": 0.3969,
+      "step": 2508
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.566486161181469e-05,
+      "loss": 0.4456,
+      "step": 2509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.565265558156101e-05,
+      "loss": 0.3159,
+      "step": 2510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564044522734147e-05,
+      "loss": 0.3454,
+      "step": 2511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562823055063692e-05,
+      "loss": 0.3102,
+      "step": 2512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561601155292873e-05,
+      "loss": 0.4272,
+      "step": 2513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.560378823569885e-05,
+      "loss": 0.3378,
+      "step": 2514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559156060042975e-05,
+      "loss": 0.3493,
+      "step": 2515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557932864860439e-05,
+      "loss": 0.3305,
+      "step": 2516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556709238170624e-05,
+      "loss": 0.409,
+      "step": 2517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.555485180121934e-05,
+      "loss": 0.3324,
+      "step": 2518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554260690862823e-05,
+      "loss": 0.4228,
+      "step": 2519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553035770541797e-05,
+      "loss": 0.3341,
+      "step": 2520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551810419307415e-05,
+      "loss": 0.294,
+      "step": 2521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550584637308288e-05,
+      "loss": 0.3887,
+      "step": 2522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549358424693077e-05,
+      "loss": 0.4035,
+      "step": 2523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.548131781610502e-05,
+      "loss": 0.4195,
+      "step": 2524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.546904708209329e-05,
+      "loss": 0.4059,
+      "step": 2525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545677204638376e-05,
+      "loss": 0.2611,
+      "step": 2526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.544449271046515e-05,
+      "loss": 0.3662,
+      "step": 2527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543220907582672e-05,
+      "loss": 0.3684,
+      "step": 2528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541992114395825e-05,
+      "loss": 0.2869,
+      "step": 2529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540762891634999e-05,
+      "loss": 0.4001,
+      "step": 2530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539533239449275e-05,
+      "loss": 0.4208,
+      "step": 2531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.538303157987786e-05,
+      "loss": 0.3218,
+      "step": 2532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537072647399719e-05,
+      "loss": 0.4516,
+      "step": 2533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535841707834307e-05,
+      "loss": 0.399,
+      "step": 2534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534610339440841e-05,
+      "loss": 0.4439,
+      "step": 2535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533378542368662e-05,
+      "loss": 0.3717,
+      "step": 2536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532146316767161e-05,
+      "loss": 0.3874,
+      "step": 2537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530913662785785e-05,
+      "loss": 0.3179,
+      "step": 2538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.529680580574028e-05,
+      "loss": 0.3645,
+      "step": 2539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528447070281442e-05,
+      "loss": 0.3131,
+      "step": 2540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.527213132057625e-05,
+      "loss": 0.3549,
+      "step": 2541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52597876605223e-05,
+      "loss": 0.3969,
+      "step": 2542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52474397241496e-05,
+      "loss": 0.3508,
+      "step": 2543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523508751295574e-05,
+      "loss": 0.3106,
+      "step": 2544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.522273102843877e-05,
+      "loss": 0.3978,
+      "step": 2545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521037027209733e-05,
+      "loss": 0.4325,
+      "step": 2546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51980052454305e-05,
+      "loss": 0.3598,
+      "step": 2547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518563594993791e-05,
+      "loss": 0.4657,
+      "step": 2548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.517326238711976e-05,
+      "loss": 0.4866,
+      "step": 2549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516088455847665e-05,
+      "loss": 0.3467,
+      "step": 2550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514850246550982e-05,
+      "loss": 0.3166,
+      "step": 2551
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.513611610972098e-05,
+      "loss": 0.3386,
+      "step": 2552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512372549261232e-05,
+      "loss": 0.3797,
+      "step": 2553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511133061568657e-05,
+      "loss": 0.3078,
+      "step": 2554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.509893148044703e-05,
+      "loss": 0.4,
+      "step": 2555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.508652808839744e-05,
+      "loss": 0.3841,
+      "step": 2556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.507412044104207e-05,
+      "loss": 0.2706,
+      "step": 2557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.506170853988576e-05,
+      "loss": 0.4275,
+      "step": 2558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.504929238643381e-05,
+      "loss": 0.351,
+      "step": 2559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.503687198219206e-05,
+      "loss": 0.3637,
+      "step": 2560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.502444732866688e-05,
+      "loss": 0.3193,
+      "step": 2561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501201842736511e-05,
+      "loss": 0.3251,
+      "step": 2562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499958527979413e-05,
+      "loss": 0.2643,
+      "step": 2563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498714788746184e-05,
+      "loss": 0.3932,
+      "step": 2564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.497470625187667e-05,
+      "loss": 0.3855,
+      "step": 2565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496226037454753e-05,
+      "loss": 0.3827,
+      "step": 2566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494981025698385e-05,
+      "loss": 0.375,
+      "step": 2567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.493735590069561e-05,
+      "loss": 0.417,
+      "step": 2568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.492489730719326e-05,
+      "loss": 0.3736,
+      "step": 2569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491243447798778e-05,
+      "loss": 0.3711,
+      "step": 2570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.489996741459067e-05,
+      "loss": 0.4037,
+      "step": 2571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488749611851396e-05,
+      "loss": 0.3608,
+      "step": 2572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.487502059127015e-05,
+      "loss": 0.3091,
+      "step": 2573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486254083437227e-05,
+      "loss": 0.2897,
+      "step": 2574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.485005684933389e-05,
+      "loss": 0.3018,
+      "step": 2575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.483756863766906e-05,
+      "loss": 0.2941,
+      "step": 2576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.482507620089236e-05,
+      "loss": 0.3872,
+      "step": 2577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481257954051887e-05,
+      "loss": 0.2872,
+      "step": 2578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48000786580642e-05,
+      "loss": 0.37,
+      "step": 2579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478757355504446e-05,
+      "loss": 0.374,
+      "step": 2580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477506423297625e-05,
+      "loss": 0.3779,
+      "step": 2581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.476255069337674e-05,
+      "loss": 0.3766,
+      "step": 2582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.475003293776353e-05,
+      "loss": 0.4665,
+      "step": 2583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473751096765484e-05,
+      "loss": 0.4872,
+      "step": 2584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.472498478456928e-05,
+      "loss": 0.3613,
+      "step": 2585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471245439002605e-05,
+      "loss": 0.3011,
+      "step": 2586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469991978554486e-05,
+      "loss": 0.3255,
+      "step": 2587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46873809726459e-05,
+      "loss": 0.3862,
+      "step": 2588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.467483795284987e-05,
+      "loss": 0.3514,
+      "step": 2589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4662290727678e-05,
+      "loss": 0.3843,
+      "step": 2590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464973929865201e-05,
+      "loss": 0.3167,
+      "step": 2591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.463718366729415e-05,
+      "loss": 0.4406,
+      "step": 2592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462462383512719e-05,
+      "loss": 0.3609,
+      "step": 2593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.461205980367435e-05,
+      "loss": 0.3337,
+      "step": 2594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.459949157445945e-05,
+      "loss": 0.3528,
+      "step": 2595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458691914900672e-05,
+      "loss": 0.3533,
+      "step": 2596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.457434252884098e-05,
+      "loss": 0.3896,
+      "step": 2597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456176171548753e-05,
+      "loss": 0.4026,
+      "step": 2598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454917671047214e-05,
+      "loss": 0.366,
+      "step": 2599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453658751532115e-05,
+      "loss": 0.238,
+      "step": 2600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452399413156138e-05,
+      "loss": 0.4292,
+      "step": 2601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451139656072015e-05,
+      "loss": 0.3698,
+      "step": 2602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449879480432532e-05,
+      "loss": 0.3961,
+      "step": 2603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448618886390522e-05,
+      "loss": 0.388,
+      "step": 2604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447357874098869e-05,
+      "loss": 0.3121,
+      "step": 2605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.446096443710512e-05,
+      "loss": 0.2949,
+      "step": 2606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444834595378434e-05,
+      "loss": 0.4898,
+      "step": 2607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443572329255675e-05,
+      "loss": 0.4169,
+      "step": 2608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442309645495321e-05,
+      "loss": 0.316,
+      "step": 2609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441046544250515e-05,
+      "loss": 0.3702,
+      "step": 2610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.439783025674442e-05,
+      "loss": 0.3031,
+      "step": 2611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.438519089920341e-05,
+      "loss": 0.3698,
+      "step": 2612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437254737141509e-05,
+      "loss": 0.2832,
+      "step": 2613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43598996749128e-05,
+      "loss": 0.3181,
+      "step": 2614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43472478112305e-05,
+      "loss": 0.2844,
+      "step": 2615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43345917819026e-05,
+      "loss": 0.3891,
+      "step": 2616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432193158846403e-05,
+      "loss": 0.3996,
+      "step": 2617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430926723245021e-05,
+      "loss": 0.3479,
+      "step": 2618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429659871539708e-05,
+      "loss": 0.3563,
+      "step": 2619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42839260388411e-05,
+      "loss": 0.4163,
+      "step": 2620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.427124920431922e-05,
+      "loss": 0.4581,
+      "step": 2621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.425856821336885e-05,
+      "loss": 0.5004,
+      "step": 2622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424588306752797e-05,
+      "loss": 0.4825,
+      "step": 2623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423319376833505e-05,
+      "loss": 0.3366,
+      "step": 2624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422050031732905e-05,
+      "loss": 0.3334,
+      "step": 2625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.420780271604942e-05,
+      "loss": 0.3255,
+      "step": 2626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419510096603614e-05,
+      "loss": 0.3734,
+      "step": 2627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418239506882966e-05,
+      "loss": 0.4153,
+      "step": 2628
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4169685025971e-05,
+      "loss": 0.3381,
+      "step": 2629
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.415697083900161e-05,
+      "loss": 0.3757,
+      "step": 2630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414425250946345e-05,
+      "loss": 0.2852,
+      "step": 2631
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.413153003889904e-05,
+      "loss": 0.4188,
+      "step": 2632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411880342885134e-05,
+      "loss": 0.4451,
+      "step": 2633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410607268086387e-05,
+      "loss": 0.3711,
+      "step": 2634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40933377964806e-05,
+      "loss": 0.2745,
+      "step": 2635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.408059877724602e-05,
+      "loss": 0.3569,
+      "step": 2636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406785562470511e-05,
+      "loss": 0.3137,
+      "step": 2637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405510834040337e-05,
+      "loss": 0.2891,
+      "step": 2638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.404235692588683e-05,
+      "loss": 0.3913,
+      "step": 2639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.402960138270194e-05,
+      "loss": 0.3128,
+      "step": 2640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.401684171239571e-05,
+      "loss": 0.3352,
+      "step": 2641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400407791651565e-05,
+      "loss": 0.2808,
+      "step": 2642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399130999660974e-05,
+      "loss": 0.3536,
+      "step": 2643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.397853795422648e-05,
+      "loss": 0.4097,
+      "step": 2644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396576179091488e-05,
+      "loss": 0.3344,
+      "step": 2645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395298150822444e-05,
+      "loss": 0.4016,
+      "step": 2646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.394019710770514e-05,
+      "loss": 0.3633,
+      "step": 2647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.392740859090748e-05,
+      "loss": 0.3407,
+      "step": 2648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.391461595938245e-05,
+      "loss": 0.3999,
+      "step": 2649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390181921468157e-05,
+      "loss": 0.4366,
+      "step": 2650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388901835835682e-05,
+      "loss": 0.309,
+      "step": 2651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.387621339196067e-05,
+      "loss": 0.2879,
+      "step": 2652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386340431704616e-05,
+      "loss": 0.3206,
+      "step": 2653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.385059113516673e-05,
+      "loss": 0.3685,
+      "step": 2654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38377738478764e-05,
+      "loss": 0.392,
+      "step": 2655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38249524567296e-05,
+      "loss": 0.3493,
+      "step": 2656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381212696328139e-05,
+      "loss": 0.3769,
+      "step": 2657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37992973690872e-05,
+      "loss": 0.3182,
+      "step": 2658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.378646367570302e-05,
+      "loss": 0.3379,
+      "step": 2659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377362588468533e-05,
+      "loss": 0.3492,
+      "step": 2660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37607839975911e-05,
+      "loss": 0.3944,
+      "step": 2661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37479380159778e-05,
+      "loss": 0.353,
+      "step": 2662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.373508794140337e-05,
+      "loss": 0.3406,
+      "step": 2663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372223377542631e-05,
+      "loss": 0.3216,
+      "step": 2664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.370937551960554e-05,
+      "loss": 0.3748,
+      "step": 2665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.369651317550054e-05,
+      "loss": 0.4043,
+      "step": 2666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368364674467126e-05,
+      "loss": 0.2705,
+      "step": 2667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367077622867814e-05,
+      "loss": 0.3408,
+      "step": 2668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.365790162908211e-05,
+      "loss": 0.3348,
+      "step": 2669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36450229474446e-05,
+      "loss": 0.3756,
+      "step": 2670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.363214018532757e-05,
+      "loss": 0.3121,
+      "step": 2671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361925334429343e-05,
+      "loss": 0.4207,
+      "step": 2672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36063624259051e-05,
+      "loss": 0.3751,
+      "step": 2673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.359346743172599e-05,
+      "loss": 0.2597,
+      "step": 2674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358056836332e-05,
+      "loss": 0.4227,
+      "step": 2675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.356766522225157e-05,
+      "loss": 0.3746,
+      "step": 2676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.355475801008556e-05,
+      "loss": 0.3513,
+      "step": 2677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354184672838739e-05,
+      "loss": 0.429,
+      "step": 2678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35289313787229e-05,
+      "loss": 0.3917,
+      "step": 2679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351601196265853e-05,
+      "loss": 0.2853,
+      "step": 2680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.350308848176109e-05,
+      "loss": 0.3419,
+      "step": 2681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.349016093759801e-05,
+      "loss": 0.4157,
+      "step": 2682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347722933173707e-05,
+      "loss": 0.3667,
+      "step": 2683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.346429366574668e-05,
+      "loss": 0.3122,
+      "step": 2684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.345135394119566e-05,
+      "loss": 0.3914,
+      "step": 2685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343841015965335e-05,
+      "loss": 0.3902,
+      "step": 2686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.342546232268955e-05,
+      "loss": 0.4379,
+      "step": 2687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.341251043187462e-05,
+      "loss": 0.377,
+      "step": 2688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.339955448877935e-05,
+      "loss": 0.3263,
+      "step": 2689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.338659449497501e-05,
+      "loss": 0.4138,
+      "step": 2690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337363045203345e-05,
+      "loss": 0.3516,
+      "step": 2691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.336066236152692e-05,
+      "loss": 0.3604,
+      "step": 2692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33476902250282e-05,
+      "loss": 0.3878,
+      "step": 2693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.333471404411054e-05,
+      "loss": 0.3738,
+      "step": 2694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.332173382034773e-05,
+      "loss": 0.3901,
+      "step": 2695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330874955531399e-05,
+      "loss": 0.3673,
+      "step": 2696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.329576125058406e-05,
+      "loss": 0.4134,
+      "step": 2697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.328276890773317e-05,
+      "loss": 0.3601,
+      "step": 2698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326977252833703e-05,
+      "loss": 0.3798,
+      "step": 2699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.325677211397184e-05,
+      "loss": 0.3381,
+      "step": 2700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.324376766621433e-05,
+      "loss": 0.3848,
+      "step": 2701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323075918664164e-05,
+      "loss": 0.3235,
+      "step": 2702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321774667683146e-05,
+      "loss": 0.3842,
+      "step": 2703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320473013836196e-05,
+      "loss": 0.367,
+      "step": 2704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319170957281177e-05,
+      "loss": 0.3571,
+      "step": 2705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.317868498176004e-05,
+      "loss": 0.4181,
+      "step": 2706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.316565636678641e-05,
+      "loss": 0.4003,
+      "step": 2707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315262372947096e-05,
+      "loss": 0.3671,
+      "step": 2708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313958707139434e-05,
+      "loss": 0.3269,
+      "step": 2709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31265463941376e-05,
+      "loss": 0.4046,
+      "step": 2710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311350169928234e-05,
+      "loss": 0.2751,
+      "step": 2711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31004529884106e-05,
+      "loss": 0.4506,
+      "step": 2712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.308740026310497e-05,
+      "loss": 0.4062,
+      "step": 2713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307434352494845e-05,
+      "loss": 0.3957,
+      "step": 2714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306128277552459e-05,
+      "loss": 0.3221,
+      "step": 2715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30482180164174e-05,
+      "loss": 0.2927,
+      "step": 2716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.303514924921136e-05,
+      "loss": 0.3376,
+      "step": 2717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302207647549147e-05,
+      "loss": 0.3613,
+      "step": 2718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300899969684321e-05,
+      "loss": 0.4443,
+      "step": 2719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299591891485252e-05,
+      "loss": 0.3885,
+      "step": 2720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298283413110586e-05,
+      "loss": 0.3178,
+      "step": 2721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.296974534719011e-05,
+      "loss": 0.3316,
+      "step": 2722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295665256469274e-05,
+      "loss": 0.3551,
+      "step": 2723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294355578520162e-05,
+      "loss": 0.444,
+      "step": 2724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.293045501030512e-05,
+      "loss": 0.3675,
+      "step": 2725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.291735024159212e-05,
+      "loss": 0.4002,
+      "step": 2726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.2904241480652e-05,
+      "loss": 0.297,
+      "step": 2727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.289112872907454e-05,
+      "loss": 0.3947,
+      "step": 2728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.287801198845008e-05,
+      "loss": 0.3806,
+      "step": 2729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.286489126036943e-05,
+      "loss": 0.3137,
+      "step": 2730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285176654642388e-05,
+      "loss": 0.327,
+      "step": 2731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28386378482052e-05,
+      "loss": 0.4342,
+      "step": 2732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282550516730563e-05,
+      "loss": 0.3028,
+      "step": 2733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28123685053179e-05,
+      "loss": 0.3491,
+      "step": 2734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.279922786383525e-05,
+      "loss": 0.3448,
+      "step": 2735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278608324445139e-05,
+      "loss": 0.2723,
+      "step": 2736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277293464876046e-05,
+      "loss": 0.3947,
+      "step": 2737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275978207835715e-05,
+      "loss": 0.447,
+      "step": 2738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274662553483661e-05,
+      "loss": 0.353,
+      "step": 2739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.273346501979449e-05,
+      "loss": 0.3073,
+      "step": 2740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272030053482686e-05,
+      "loss": 0.3357,
+      "step": 2741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270713208153034e-05,
+      "loss": 0.348,
+      "step": 2742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.269395966150201e-05,
+      "loss": 0.3338,
+      "step": 2743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.26807832763394e-05,
+      "loss": 0.4668,
+      "step": 2744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266760292764058e-05,
+      "loss": 0.3009,
+      "step": 2745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265441861700404e-05,
+      "loss": 0.3627,
+      "step": 2746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.264123034602877e-05,
+      "loss": 0.3242,
+      "step": 2747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.262803811631427e-05,
+      "loss": 0.3489,
+      "step": 2748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.26148419294605e-05,
+      "loss": 0.4672,
+      "step": 2749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.260164178706788e-05,
+      "loss": 0.3284,
+      "step": 2750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.258843769073733e-05,
+      "loss": 0.3845,
+      "step": 2751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257522964207028e-05,
+      "loss": 0.3445,
+      "step": 2752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.256201764266856e-05,
+      "loss": 0.4263,
+      "step": 2753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.254880169413455e-05,
+      "loss": 0.3495,
+      "step": 2754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.253558179807109e-05,
+      "loss": 0.3799,
+      "step": 2755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.252235795608148e-05,
+      "loss": 0.464,
+      "step": 2756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25091301697695e-05,
+      "loss": 0.4013,
+      "step": 2757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249589844073944e-05,
+      "loss": 0.5112,
+      "step": 2758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248266277059607e-05,
+      "loss": 0.3717,
+      "step": 2759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.246942316094457e-05,
+      "loss": 0.4899,
+      "step": 2760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245617961339066e-05,
+      "loss": 0.3989,
+      "step": 2761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.244293212954054e-05,
+      "loss": 0.3177,
+      "step": 2762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.242968071100085e-05,
+      "loss": 0.2972,
+      "step": 2763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241642535937873e-05,
+      "loss": 0.3629,
+      "step": 2764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.24031660762818e-05,
+      "loss": 0.3783,
+      "step": 2765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.238990286331815e-05,
+      "loss": 0.3756,
+      "step": 2766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237663572209636e-05,
+      "loss": 0.4001,
+      "step": 2767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236336465422543e-05,
+      "loss": 0.3083,
+      "step": 2768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.235008966131491e-05,
+      "loss": 0.3503,
+      "step": 2769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23368107449748e-05,
+      "loss": 0.3239,
+      "step": 2770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.232352790681556e-05,
+      "loss": 0.454,
+      "step": 2771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231024114844814e-05,
+      "loss": 0.3388,
+      "step": 2772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229695047148396e-05,
+      "loss": 0.3321,
+      "step": 2773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.228365587753493e-05,
+      "loss": 0.2872,
+      "step": 2774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22703573682134e-05,
+      "loss": 0.3874,
+      "step": 2775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225705494513223e-05,
+      "loss": 0.3293,
+      "step": 2776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.224374860990476e-05,
+      "loss": 0.3478,
+      "step": 2777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.223043836414477e-05,
+      "loss": 0.4222,
+      "step": 2778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.221712420946651e-05,
+      "loss": 0.3762,
+      "step": 2779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.220380614748476e-05,
+      "loss": 0.4183,
+      "step": 2780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219048417981471e-05,
+      "loss": 0.2965,
+      "step": 2781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.217715830807209e-05,
+      "loss": 0.3835,
+      "step": 2782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.216382853387304e-05,
+      "loss": 0.3509,
+      "step": 2783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21504948588342e-05,
+      "loss": 0.3885,
+      "step": 2784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.213715728457267e-05,
+      "loss": 0.3432,
+      "step": 2785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.212381581270607e-05,
+      "loss": 0.3435,
+      "step": 2786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211047044485242e-05,
+      "loss": 0.4042,
+      "step": 2787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209712118263029e-05,
+      "loss": 0.3201,
+      "step": 2788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208376802765866e-05,
+      "loss": 0.4153,
+      "step": 2789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2070410981557e-05,
+      "loss": 0.3584,
+      "step": 2790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205705004594526e-05,
+      "loss": 0.3797,
+      "step": 2791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204368522244388e-05,
+      "loss": 0.4343,
+      "step": 2792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.203031651267374e-05,
+      "loss": 0.3837,
+      "step": 2793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201694391825616e-05,
+      "loss": 0.2722,
+      "step": 2794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.200356744081303e-05,
+      "loss": 0.3809,
+      "step": 2795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199018708196664e-05,
+      "loss": 0.4745,
+      "step": 2796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197680284333974e-05,
+      "loss": 0.396,
+      "step": 2797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19634147265556e-05,
+      "loss": 0.3257,
+      "step": 2798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.195002273323791e-05,
+      "loss": 0.3465,
+      "step": 2799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.193662686501089e-05,
+      "loss": 0.3233,
+      "step": 2800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.192322712349917e-05,
+      "loss": 0.4172,
+      "step": 2801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19098235103279e-05,
+      "loss": 0.3793,
+      "step": 2802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189641602712264e-05,
+      "loss": 0.3829,
+      "step": 2803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.188300467550947e-05,
+      "loss": 0.361,
+      "step": 2804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.186958945711491e-05,
+      "loss": 0.3712,
+      "step": 2805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.1856170373566e-05,
+      "loss": 0.3392,
+      "step": 2806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184274742649017e-05,
+      "loss": 0.3019,
+      "step": 2807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.182932061751538e-05,
+      "loss": 0.3397,
+      "step": 2808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.181588994827005e-05,
+      "loss": 0.2646,
+      "step": 2809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180245542038303e-05,
+      "loss": 0.3564,
+      "step": 2810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.178901703548367e-05,
+      "loss": 0.3749,
+      "step": 2811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177557479520179e-05,
+      "loss": 0.3982,
+      "step": 2812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.176212870116766e-05,
+      "loss": 0.3512,
+      "step": 2813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174867875501203e-05,
+      "loss": 0.4157,
+      "step": 2814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173522495836612e-05,
+      "loss": 0.3425,
+      "step": 2815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17217673128616e-05,
+      "loss": 0.3843,
+      "step": 2816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170830582013065e-05,
+      "loss": 0.2702,
+      "step": 2817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.169484048180585e-05,
+      "loss": 0.2774,
+      "step": 2818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168137129952027e-05,
+      "loss": 0.3419,
+      "step": 2819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.166789827490749e-05,
+      "loss": 0.4106,
+      "step": 2820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16544214096015e-05,
+      "loss": 0.3986,
+      "step": 2821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16409407052368e-05,
+      "loss": 0.3742,
+      "step": 2822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16274561634483e-05,
+      "loss": 0.4653,
+      "step": 2823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161396778587145e-05,
+      "loss": 0.3605,
+      "step": 2824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.160047557414212e-05,
+      "loss": 0.4361,
+      "step": 2825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.158697952989662e-05,
+      "loss": 0.4087,
+      "step": 2826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157347965477178e-05,
+      "loss": 0.3366,
+      "step": 2827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.155997595040485e-05,
+      "loss": 0.3533,
+      "step": 2828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.154646841843358e-05,
+      "loss": 0.3865,
+      "step": 2829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153295706049618e-05,
+      "loss": 0.293,
+      "step": 2830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.151944187823126e-05,
+      "loss": 0.3837,
+      "step": 2831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.1505922873278e-05,
+      "loss": 0.2836,
+      "step": 2832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149240004727596e-05,
+      "loss": 0.3583,
+      "step": 2833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.147887340186522e-05,
+      "loss": 0.4913,
+      "step": 2834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.146534293868626e-05,
+      "loss": 0.3908,
+      "step": 2835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.145180865938009e-05,
+      "loss": 0.416,
+      "step": 2836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.143827056558814e-05,
+      "loss": 0.3526,
+      "step": 2837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14247286589523e-05,
+      "loss": 0.3492,
+      "step": 2838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.141118294111496e-05,
+      "loss": 0.4551,
+      "step": 2839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139763341371896e-05,
+      "loss": 0.3705,
+      "step": 2840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138408007840755e-05,
+      "loss": 0.3405,
+      "step": 2841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13705229368245e-05,
+      "loss": 0.3217,
+      "step": 2842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.135696199061404e-05,
+      "loss": 0.2758,
+      "step": 2843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.134339724142083e-05,
+      "loss": 0.4024,
+      "step": 2844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.132982869089e-05,
+      "loss": 0.4343,
+      "step": 2845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.131625634066718e-05,
+      "loss": 0.3363,
+      "step": 2846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.130268019239839e-05,
+      "loss": 0.3223,
+      "step": 2847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12891002477302e-05,
+      "loss": 0.3629,
+      "step": 2848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127551650830954e-05,
+      "loss": 0.4086,
+      "step": 2849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.126192897578388e-05,
+      "loss": 0.3397,
+      "step": 2850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12483376518011e-05,
+      "loss": 0.3978,
+      "step": 2851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.123474253800957e-05,
+      "loss": 0.3612,
+      "step": 2852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12211436360581e-05,
+      "loss": 0.3075,
+      "step": 2853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.1207540947596e-05,
+      "loss": 0.4185,
+      "step": 2854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.1193934474273e-05,
+      "loss": 0.3723,
+      "step": 2855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118032421773925e-05,
+      "loss": 0.357,
+      "step": 2856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.116671017964547e-05,
+      "loss": 0.4078,
+      "step": 2857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115309236164273e-05,
+      "loss": 0.3484,
+      "step": 2858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.113947076538265e-05,
+      "loss": 0.3477,
+      "step": 2859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.11258453925172e-05,
+      "loss": 0.3066,
+      "step": 2860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.111221624469891e-05,
+      "loss": 0.3669,
+      "step": 2861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.109858332358074e-05,
+      "loss": 0.2978,
+      "step": 2862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108494663081605e-05,
+      "loss": 0.362,
+      "step": 2863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.107130616805873e-05,
+      "loss": 0.3083,
+      "step": 2864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.105766193696311e-05,
+      "loss": 0.291,
+      "step": 2865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.104401393918396e-05,
+      "loss": 0.316,
+      "step": 2866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103036217637651e-05,
+      "loss": 0.4084,
+      "step": 2867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101670665019643e-05,
+      "loss": 0.358,
+      "step": 2868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10030473622999e-05,
+      "loss": 0.4413,
+      "step": 2869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.098938431434351e-05,
+      "loss": 0.424,
+      "step": 2870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09757175079843e-05,
+      "loss": 0.3129,
+      "step": 2871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.096204694487982e-05,
+      "loss": 0.3706,
+      "step": 2872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.094837262668801e-05,
+      "loss": 0.2737,
+      "step": 2873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093469455506732e-05,
+      "loss": 0.339,
+      "step": 2874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09210127316766e-05,
+      "loss": 0.278,
+      "step": 2875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.090732715817523e-05,
+      "loss": 0.4336,
+      "step": 2876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089363783622297e-05,
+      "loss": 0.3388,
+      "step": 2877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087994476748007e-05,
+      "loss": 0.3124,
+      "step": 2878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.086624795360723e-05,
+      "loss": 0.271,
+      "step": 2879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08525473962656e-05,
+      "loss": 0.3252,
+      "step": 2880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08388430971168e-05,
+      "loss": 0.274,
+      "step": 2881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.082513505782288e-05,
+      "loss": 0.2921,
+      "step": 2882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.081142328004637e-05,
+      "loss": 0.3065,
+      "step": 2883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.079770776545023e-05,
+      "loss": 0.2797,
+      "step": 2884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.078398851569789e-05,
+      "loss": 0.4107,
+      "step": 2885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.077026553245319e-05,
+      "loss": 0.3868,
+      "step": 2886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.075653881738051e-05,
+      "loss": 0.361,
+      "step": 2887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074280837214461e-05,
+      "loss": 0.395,
+      "step": 2888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07290741984107e-05,
+      "loss": 0.3163,
+      "step": 2889
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.071533629784449e-05,
+      "loss": 0.3267,
+      "step": 2890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.070159467211213e-05,
+      "loss": 0.3148,
+      "step": 2891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068784932288018e-05,
+      "loss": 0.3285,
+      "step": 2892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.067410025181567e-05,
+      "loss": 0.3331,
+      "step": 2893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.066034746058614e-05,
+      "loss": 0.2582,
+      "step": 2894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06465909508595e-05,
+      "loss": 0.4132,
+      "step": 2895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063283072430415e-05,
+      "loss": 0.4987,
+      "step": 2896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061906678258892e-05,
+      "loss": 0.3637,
+      "step": 2897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060529912738315e-05,
+      "loss": 0.3069,
+      "step": 2898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059152776035654e-05,
+      "loss": 0.3046,
+      "step": 2899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05777526831793e-05,
+      "loss": 0.4254,
+      "step": 2900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056397389752208e-05,
+      "loss": 0.3658,
+      "step": 2901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.055019140505597e-05,
+      "loss": 0.3947,
+      "step": 2902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.053640520745253e-05,
+      "loss": 0.3828,
+      "step": 2903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052261530638374e-05,
+      "loss": 0.3461,
+      "step": 2904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.050882170352203e-05,
+      "loss": 0.2869,
+      "step": 2905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.049502440054033e-05,
+      "loss": 0.2964,
+      "step": 2906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048122339911194e-05,
+      "loss": 0.2809,
+      "step": 2907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046741870091071e-05,
+      "loss": 0.291,
+      "step": 2908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04536103076108e-05,
+      "loss": 0.2867,
+      "step": 2909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043979822088696e-05,
+      "loss": 0.3588,
+      "step": 2910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04259824424143e-05,
+      "loss": 0.2879,
+      "step": 2911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04121629738684e-05,
+      "loss": 0.2175,
+      "step": 2912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.039833981692529e-05,
+      "loss": 0.3121,
+      "step": 2913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038451297326145e-05,
+      "loss": 0.3578,
+      "step": 2914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037068244455381e-05,
+      "loss": 0.2592,
+      "step": 2915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.035684823247973e-05,
+      "loss": 0.3344,
+      "step": 2916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.034301033871702e-05,
+      "loss": 0.4163,
+      "step": 2917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.032916876494399e-05,
+      "loss": 0.357,
+      "step": 2918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031532351283929e-05,
+      "loss": 0.4234,
+      "step": 2919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.030147458408212e-05,
+      "loss": 0.4233,
+      "step": 2920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028762198035205e-05,
+      "loss": 0.4045,
+      "step": 2921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027376570332916e-05,
+      "loss": 0.3683,
+      "step": 2922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.025990575469394e-05,
+      "loss": 0.3452,
+      "step": 2923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02460421361273e-05,
+      "loss": 0.4089,
+      "step": 2924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.023217484931065e-05,
+      "loss": 0.322,
+      "step": 2925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.021830389592582e-05,
+      "loss": 0.3487,
+      "step": 2926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.020442927765506e-05,
+      "loss": 0.3875,
+      "step": 2927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019055099618111e-05,
+      "loss": 0.4275,
+      "step": 2928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.017666905318712e-05,
+      "loss": 0.308,
+      "step": 2929
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.016278345035672e-05,
+      "loss": 0.3485,
+      "step": 2930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.014889418937393e-05,
+      "loss": 0.3653,
+      "step": 2931
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013500127192329e-05,
+      "loss": 0.3465,
+      "step": 2932
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.012110469968967e-05,
+      "loss": 0.3368,
+      "step": 2933
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01072044743585e-05,
+      "loss": 0.3439,
+      "step": 2934
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00933005976156e-05,
+      "loss": 0.3599,
+      "step": 2935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00793930711472e-05,
+      "loss": 0.3279,
+      "step": 2936
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.006548189664005e-05,
+      "loss": 0.3437,
+      "step": 2937
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00515670757813e-05,
+      "loss": 0.3265,
+      "step": 2938
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.003764861025853e-05,
+      "loss": 0.362,
+      "step": 2939
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002372650175978e-05,
+      "loss": 0.3739,
+      "step": 2940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.000980075197352e-05,
+      "loss": 0.4093,
+      "step": 2941
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999587136258869e-05,
+      "loss": 0.3246,
+      "step": 2942
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.998193833529462e-05,
+      "loss": 0.4055,
+      "step": 2943
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.996800167178113e-05,
+      "loss": 0.3722,
+      "step": 2944
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.995406137373846e-05,
+      "loss": 0.3087,
+      "step": 2945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994011744285732e-05,
+      "loss": 0.3511,
+      "step": 2946
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.992616988082879e-05,
+      "loss": 0.3742,
+      "step": 2947
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.991221868934447e-05,
+      "loss": 0.3393,
+      "step": 2948
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989826387009633e-05,
+      "loss": 0.3238,
+      "step": 2949
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.988430542477686e-05,
+      "loss": 0.4521,
+      "step": 2950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987034335507891e-05,
+      "loss": 0.4576,
+      "step": 2951
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985637766269582e-05,
+      "loss": 0.3501,
+      "step": 2952
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984240834932136e-05,
+      "loss": 0.3193,
+      "step": 2953
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.982843541664971e-05,
+      "loss": 0.3691,
+      "step": 2954
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981445886637553e-05,
+      "loss": 0.4251,
+      "step": 2955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.980047870019387e-05,
+      "loss": 0.3082,
+      "step": 2956
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97864949198003e-05,
+      "loss": 0.3607,
+      "step": 2957
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.977250752689075e-05,
+      "loss": 0.3458,
+      "step": 2958
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.975851652316162e-05,
+      "loss": 0.3918,
+      "step": 2959
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.974452191030973e-05,
+      "loss": 0.3144,
+      "step": 2960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973052369003238e-05,
+      "loss": 0.3224,
+      "step": 2961
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.971652186402723e-05,
+      "loss": 0.2718,
+      "step": 2962
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97025164339925e-05,
+      "loss": 0.3388,
+      "step": 2963
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96885074016267e-05,
+      "loss": 0.3601,
+      "step": 2964
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96744947686289e-05,
+      "loss": 0.341,
+      "step": 2965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966047853669851e-05,
+      "loss": 0.3061,
+      "step": 2966
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.964645870753546e-05,
+      "loss": 0.3246,
+      "step": 2967
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.963243528284009e-05,
+      "loss": 0.2761,
+      "step": 2968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961840826431313e-05,
+      "loss": 0.2872,
+      "step": 2969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96043776536558e-05,
+      "loss": 0.3612,
+      "step": 2970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.959034345256973e-05,
+      "loss": 0.3997,
+      "step": 2971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9576305662757e-05,
+      "loss": 0.3592,
+      "step": 2972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.956226428592013e-05,
+      "loss": 0.3033,
+      "step": 2973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.954821932376202e-05,
+      "loss": 0.375,
+      "step": 2974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95341707779861e-05,
+      "loss": 0.316,
+      "step": 2975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.952011865029614e-05,
+      "loss": 0.3851,
+      "step": 2976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95060629423964e-05,
+      "loss": 0.3783,
+      "step": 2977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.949200365599158e-05,
+      "loss": 0.286,
+      "step": 2978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.947794079278677e-05,
+      "loss": 0.2956,
+      "step": 2979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.946387435448754e-05,
+      "loss": 0.3293,
+      "step": 2980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944980434279985e-05,
+      "loss": 0.3693,
+      "step": 2981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.943573075943014e-05,
+      "loss": 0.3662,
+      "step": 2982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942165360608524e-05,
+      "loss": 0.3278,
+      "step": 2983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.940757288447244e-05,
+      "loss": 0.333,
+      "step": 2984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939348859629944e-05,
+      "loss": 0.3771,
+      "step": 2985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.937940074327441e-05,
+      "loss": 0.3021,
+      "step": 2986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93653093271059e-05,
+      "loss": 0.3149,
+      "step": 2987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.935121434950297e-05,
+      "loss": 0.2355,
+      "step": 2988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.933711581217501e-05,
+      "loss": 0.3533,
+      "step": 2989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932301371683192e-05,
+      "loss": 0.3145,
+      "step": 2990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930890806518403e-05,
+      "loss": 0.3733,
+      "step": 2991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.929479885894204e-05,
+      "loss": 0.3785,
+      "step": 2992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.928068609981712e-05,
+      "loss": 0.3522,
+      "step": 2993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92665697895209e-05,
+      "loss": 0.2672,
+      "step": 2994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925244992976538e-05,
+      "loss": 0.2788,
+      "step": 2995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923832652226305e-05,
+      "loss": 0.3191,
+      "step": 2996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.922419956872678e-05,
+      "loss": 0.2898,
+      "step": 2997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92100690708699e-05,
+      "loss": 0.3469,
+      "step": 2998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.919593503040616e-05,
+      "loss": 0.297,
+      "step": 2999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.918179744904975e-05,
+      "loss": 0.3332,
+      "step": 3000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.916765632851525e-05,
+      "loss": 0.2951,
+      "step": 3001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.915351167051773e-05,
+      "loss": 0.3249,
+      "step": 3002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.913936347677265e-05,
+      "loss": 0.395,
+      "step": 3003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.912521174899591e-05,
+      "loss": 0.3292,
+      "step": 3004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.911105648890384e-05,
+      "loss": 0.354,
+      "step": 3005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.909689769821319e-05,
+      "loss": 0.2732,
+      "step": 3006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.908273537864113e-05,
+      "loss": 0.3566,
+      "step": 3007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.906856953190529e-05,
+      "loss": 0.4274,
+      "step": 3008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905440015972371e-05,
+      "loss": 0.3265,
+      "step": 3009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904022726381482e-05,
+      "loss": 0.3781,
+      "step": 3010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.902605084589756e-05,
+      "loss": 0.3358,
+      "step": 3011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.901187090769123e-05,
+      "loss": 0.3758,
+      "step": 3012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.899768745091559e-05,
+      "loss": 0.3247,
+      "step": 3013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898350047729077e-05,
+      "loss": 0.4053,
+      "step": 3014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.896930998853744e-05,
+      "loss": 0.3323,
+      "step": 3015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.895511598637657e-05,
+      "loss": 0.3517,
+      "step": 3016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894091847252964e-05,
+      "loss": 0.3372,
+      "step": 3017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892671744871853e-05,
+      "loss": 0.3652,
+      "step": 3018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.891251291666553e-05,
+      "loss": 0.2903,
+      "step": 3019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889830487809338e-05,
+      "loss": 0.3518,
+      "step": 3020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.888409333472524e-05,
+      "loss": 0.3993,
+      "step": 3021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.886987828828468e-05,
+      "loss": 0.3721,
+      "step": 3022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885565974049572e-05,
+      "loss": 0.3596,
+      "step": 3023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.884143769308276e-05,
+      "loss": 0.3857,
+      "step": 3024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.882721214777069e-05,
+      "loss": 0.2802,
+      "step": 3025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881298310628478e-05,
+      "loss": 0.3499,
+      "step": 3026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879875057035072e-05,
+      "loss": 0.3343,
+      "step": 3027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.878451454169465e-05,
+      "loss": 0.3824,
+      "step": 3028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.877027502204311e-05,
+      "loss": 0.4058,
+      "step": 3029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.875603201312307e-05,
+      "loss": 0.3618,
+      "step": 3030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874178551666195e-05,
+      "loss": 0.3151,
+      "step": 3031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872753553438754e-05,
+      "loss": 0.3133,
+      "step": 3032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.87132820680281e-05,
+      "loss": 0.3585,
+      "step": 3033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.869902511931231e-05,
+      "loss": 0.3451,
+      "step": 3034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.868476468996922e-05,
+      "loss": 0.3589,
+      "step": 3035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867050078172836e-05,
+      "loss": 0.2823,
+      "step": 3036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865623339631967e-05,
+      "loss": 0.3421,
+      "step": 3037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86419625354735e-05,
+      "loss": 0.3365,
+      "step": 3038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.862768820092061e-05,
+      "loss": 0.3442,
+      "step": 3039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86134103943922e-05,
+      "loss": 0.3091,
+      "step": 3040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85991291176199e-05,
+      "loss": 0.3882,
+      "step": 3041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.858484437233573e-05,
+      "loss": 0.2983,
+      "step": 3042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.857055616027218e-05,
+      "loss": 0.3764,
+      "step": 3043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.855626448316208e-05,
+      "loss": 0.3333,
+      "step": 3044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854196934273876e-05,
+      "loss": 0.222,
+      "step": 3045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.852767074073595e-05,
+      "loss": 0.3845,
+      "step": 3046
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851336867888779e-05,
+      "loss": 0.4549,
+      "step": 3047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84990631589288e-05,
+      "loss": 0.3353,
+      "step": 3048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.848475418259398e-05,
+      "loss": 0.2881,
+      "step": 3049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847044175161875e-05,
+      "loss": 0.2692,
+      "step": 3050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.845612586773889e-05,
+      "loss": 0.4016,
+      "step": 3051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.844180653269067e-05,
+      "loss": 0.3483,
+      "step": 3052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84274837482107e-05,
+      "loss": 0.3713,
+      "step": 3053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84131575160361e-05,
+      "loss": 0.3426,
+      "step": 3054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.839882783790434e-05,
+      "loss": 0.3102,
+      "step": 3055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838449471555331e-05,
+      "loss": 0.3247,
+      "step": 3056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.837015815072136e-05,
+      "loss": 0.3176,
+      "step": 3057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83558181451472e-05,
+      "loss": 0.3294,
+      "step": 3058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834147470057005e-05,
+      "loss": 0.304,
+      "step": 3059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832712781872943e-05,
+      "loss": 0.374,
+      "step": 3060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.831277750136536e-05,
+      "loss": 0.3279,
+      "step": 3061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.829842375021825e-05,
+      "loss": 0.3831,
+      "step": 3062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.828406656702893e-05,
+      "loss": 0.4299,
+      "step": 3063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.826970595353866e-05,
+      "loss": 0.3504,
+      "step": 3064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.825534191148907e-05,
+      "loss": 0.4139,
+      "step": 3065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.824097444262224e-05,
+      "loss": 0.3746,
+      "step": 3066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.822660354868066e-05,
+      "loss": 0.2929,
+      "step": 3067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.821222923140726e-05,
+      "loss": 0.2334,
+      "step": 3068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.819785149254532e-05,
+      "loss": 0.3805,
+      "step": 3069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.818347033383864e-05,
+      "loss": 0.3425,
+      "step": 3070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.816908575703131e-05,
+      "loss": 0.338,
+      "step": 3071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.815469776386792e-05,
+      "loss": 0.4044,
+      "step": 3072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.814030635609347e-05,
+      "loss": 0.3587,
+      "step": 3073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.812591153545332e-05,
+      "loss": 0.3592,
+      "step": 3074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81115133036933e-05,
+      "loss": 0.3465,
+      "step": 3075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.809711166255965e-05,
+      "loss": 0.2937,
+      "step": 3076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808270661379896e-05,
+      "loss": 0.3538,
+      "step": 3077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80682981591583e-05,
+      "loss": 0.3178,
+      "step": 3078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.805388630038513e-05,
+      "loss": 0.3375,
+      "step": 3079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.803947103922735e-05,
+      "loss": 0.3373,
+      "step": 3080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.802505237743323e-05,
+      "loss": 0.3081,
+      "step": 3081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.801063031675144e-05,
+      "loss": 0.2747,
+      "step": 3082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799620485893113e-05,
+      "loss": 0.3823,
+      "step": 3083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798177600572182e-05,
+      "loss": 0.2899,
+      "step": 3084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.796734375887345e-05,
+      "loss": 0.3734,
+      "step": 3085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.795290812013634e-05,
+      "loss": 0.4142,
+      "step": 3086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793846909126126e-05,
+      "loss": 0.3561,
+      "step": 3087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79240266739994e-05,
+      "loss": 0.294,
+      "step": 3088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790958087010233e-05,
+      "loss": 0.3757,
+      "step": 3089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.789513168132205e-05,
+      "loss": 0.3827,
+      "step": 3090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.788067910941095e-05,
+      "loss": 0.3136,
+      "step": 3091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.786622315612183e-05,
+      "loss": 0.3363,
+      "step": 3092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.785176382320794e-05,
+      "loss": 0.3477,
+      "step": 3093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783730111242289e-05,
+      "loss": 0.3267,
+      "step": 3094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.782283502552072e-05,
+      "loss": 0.3214,
+      "step": 3095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.780836556425592e-05,
+      "loss": 0.3182,
+      "step": 3096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.779389273038333e-05,
+      "loss": 0.3929,
+      "step": 3097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77794165256582e-05,
+      "loss": 0.4021,
+      "step": 3098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.776493695183624e-05,
+      "loss": 0.3081,
+      "step": 3099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77504540106735e-05,
+      "loss": 0.359,
+      "step": 3100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.773596770392651e-05,
+      "loss": 0.2647,
+      "step": 3101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.772147803335217e-05,
+      "loss": 0.4023,
+      "step": 3102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770698500070777e-05,
+      "loss": 0.3338,
+      "step": 3103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.769248860775105e-05,
+      "loss": 0.3816,
+      "step": 3104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.767798885624013e-05,
+      "loss": 0.3511,
+      "step": 3105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766348574793357e-05,
+      "loss": 0.36,
+      "step": 3106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.764897928459025e-05,
+      "loss": 0.3267,
+      "step": 3107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763446946796957e-05,
+      "loss": 0.3285,
+      "step": 3108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761995629983129e-05,
+      "loss": 0.3776,
+      "step": 3109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.760543978193553e-05,
+      "loss": 0.3142,
+      "step": 3110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.75909199160429e-05,
+      "loss": 0.3726,
+      "step": 3111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757639670391435e-05,
+      "loss": 0.3454,
+      "step": 3112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.756187014731127e-05,
+      "loss": 0.36,
+      "step": 3113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.754734024799544e-05,
+      "loss": 0.3058,
+      "step": 3114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.753280700772905e-05,
+      "loss": 0.3394,
+      "step": 3115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.751827042827469e-05,
+      "loss": 0.3713,
+      "step": 3116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.750373051139537e-05,
+      "loss": 0.3656,
+      "step": 3117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74891872588545e-05,
+      "loss": 0.3888,
+      "step": 3118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74746406724159e-05,
+      "loss": 0.3204,
+      "step": 3119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.746009075384376e-05,
+      "loss": 0.3263,
+      "step": 3120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.744553750490272e-05,
+      "loss": 0.3381,
+      "step": 3121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743098092735778e-05,
+      "loss": 0.3238,
+      "step": 3122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74164210229744e-05,
+      "loss": 0.3421,
+      "step": 3123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.740185779351837e-05,
+      "loss": 0.3393,
+      "step": 3124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.738729124075594e-05,
+      "loss": 0.2791,
+      "step": 3125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737272136645377e-05,
+      "loss": 0.2952,
+      "step": 3126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.735814817237887e-05,
+      "loss": 0.3718,
+      "step": 3127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.734357166029868e-05,
+      "loss": 0.3398,
+      "step": 3128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.732899183198106e-05,
+      "loss": 0.436,
+      "step": 3129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.731440868919426e-05,
+      "loss": 0.3897,
+      "step": 3130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.729982223370691e-05,
+      "loss": 0.3705,
+      "step": 3131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.728523246728807e-05,
+      "loss": 0.2566,
+      "step": 3132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.727063939170721e-05,
+      "loss": 0.374,
+      "step": 3133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.725604300873414e-05,
+      "loss": 0.3586,
+      "step": 3134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.724144332013914e-05,
+      "loss": 0.345,
+      "step": 3135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.722684032769287e-05,
+      "loss": 0.3396,
+      "step": 3136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721223403316637e-05,
+      "loss": 0.2839,
+      "step": 3137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.719762443833113e-05,
+      "loss": 0.2913,
+      "step": 3138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.718301154495897e-05,
+      "loss": 0.2823,
+      "step": 3139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.716839535482217e-05,
+      "loss": 0.3705,
+      "step": 3140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.715377586969335e-05,
+      "loss": 0.3154,
+      "step": 3141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713915309134561e-05,
+      "loss": 0.2808,
+      "step": 3142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.712452702155238e-05,
+      "loss": 0.2962,
+      "step": 3143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.710989766208752e-05,
+      "loss": 0.3207,
+      "step": 3144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.709526501472527e-05,
+      "loss": 0.331,
+      "step": 3145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.708062908124032e-05,
+      "loss": 0.2663,
+      "step": 3146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70659898634077e-05,
+      "loss": 0.32,
+      "step": 3147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705134736300284e-05,
+      "loss": 0.3521,
+      "step": 3148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70367015818016e-05,
+      "loss": 0.3233,
+      "step": 3149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.702205252158025e-05,
+      "loss": 0.4491,
+      "step": 3150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70074001841154e-05,
+      "loss": 0.3248,
+      "step": 3151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699274457118411e-05,
+      "loss": 0.3352,
+      "step": 3152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69780856845638e-05,
+      "loss": 0.4319,
+      "step": 3153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.696342352603232e-05,
+      "loss": 0.3793,
+      "step": 3154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69487580973679e-05,
+      "loss": 0.3487,
+      "step": 3155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.693408940034916e-05,
+      "loss": 0.3744,
+      "step": 3156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.691941743675512e-05,
+      "loss": 0.3272,
+      "step": 3157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.690474220836524e-05,
+      "loss": 0.3728,
+      "step": 3158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.689006371695928e-05,
+      "loss": 0.3467,
+      "step": 3159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687538196431747e-05,
+      "loss": 0.4784,
+      "step": 3160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.686069695222044e-05,
+      "loss": 0.3016,
+      "step": 3161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68460086824492e-05,
+      "loss": 0.3381,
+      "step": 3162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.683131715678511e-05,
+      "loss": 0.3305,
+      "step": 3163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681662237700997e-05,
+      "loss": 0.376,
+      "step": 3164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.680192434490598e-05,
+      "loss": 0.241,
+      "step": 3165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.678722306225574e-05,
+      "loss": 0.3581,
+      "step": 3166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.677251853084218e-05,
+      "loss": 0.3502,
+      "step": 3167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675781075244872e-05,
+      "loss": 0.3884,
+      "step": 3168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.674309972885909e-05,
+      "loss": 0.3946,
+      "step": 3169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.672838546185748e-05,
+      "loss": 0.3966,
+      "step": 3170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67136679532284e-05,
+      "loss": 0.2879,
+      "step": 3171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.669894720475683e-05,
+      "loss": 0.3482,
+      "step": 3172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668422321822805e-05,
+      "loss": 0.3236,
+      "step": 3173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666949599542787e-05,
+      "loss": 0.3537,
+      "step": 3174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.665476553814239e-05,
+      "loss": 0.3811,
+      "step": 3175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.664003184815807e-05,
+      "loss": 0.3388,
+      "step": 3176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.662529492726187e-05,
+      "loss": 0.3832,
+      "step": 3177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.661055477724109e-05,
+      "loss": 0.4271,
+      "step": 3178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65958113998834e-05,
+      "loss": 0.2877,
+      "step": 3179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.658106479697688e-05,
+      "loss": 0.2623,
+      "step": 3180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656631497031002e-05,
+      "loss": 0.3075,
+      "step": 3181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.655156192167166e-05,
+      "loss": 0.3233,
+      "step": 3182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.653680565285106e-05,
+      "loss": 0.2831,
+      "step": 3183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.652204616563789e-05,
+      "loss": 0.2593,
+      "step": 3184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.650728346182216e-05,
+      "loss": 0.2956,
+      "step": 3185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.649251754319433e-05,
+      "loss": 0.3373,
+      "step": 3186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.647774841154517e-05,
+      "loss": 0.2921,
+      "step": 3187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64629760686659e-05,
+      "loss": 0.3978,
+      "step": 3188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644820051634812e-05,
+      "loss": 0.3015,
+      "step": 3189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.643342175638382e-05,
+      "loss": 0.3603,
+      "step": 3190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641863979056539e-05,
+      "loss": 0.3079,
+      "step": 3191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.640385462068553e-05,
+      "loss": 0.3301,
+      "step": 3192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.638906624853743e-05,
+      "loss": 0.4032,
+      "step": 3193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.637427467591465e-05,
+      "loss": 0.3609,
+      "step": 3194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.635947990461109e-05,
+      "loss": 0.311,
+      "step": 3195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634468193642105e-05,
+      "loss": 0.3144,
+      "step": 3196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.632988077313928e-05,
+      "loss": 0.3354,
+      "step": 3197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.631507641656084e-05,
+      "loss": 0.3456,
+      "step": 3198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630026886848119e-05,
+      "loss": 0.379,
+      "step": 3199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.628545813069621e-05,
+      "loss": 0.3692,
+      "step": 3200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627064420500217e-05,
+      "loss": 0.3816,
+      "step": 3201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.625582709319568e-05,
+      "loss": 0.2956,
+      "step": 3202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.624100679707379e-05,
+      "loss": 0.3517,
+      "step": 3203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622618331843389e-05,
+      "loss": 0.3125,
+      "step": 3204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621135665907378e-05,
+      "loss": 0.356,
+      "step": 3205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.619652682079165e-05,
+      "loss": 0.321,
+      "step": 3206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.618169380538606e-05,
+      "loss": 0.3601,
+      "step": 3207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.616685761465596e-05,
+      "loss": 0.3723,
+      "step": 3208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61520182504007e-05,
+      "loss": 0.2685,
+      "step": 3209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.613717571442e-05,
+      "loss": 0.3149,
+      "step": 3210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612233000851398e-05,
+      "loss": 0.2948,
+      "step": 3211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610748113448309e-05,
+      "loss": 0.2957,
+      "step": 3212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.609262909412825e-05,
+      "loss": 0.3629,
+      "step": 3213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60777738892507e-05,
+      "loss": 0.4409,
+      "step": 3214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.606291552165208e-05,
+      "loss": 0.284,
+      "step": 3215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.604805399313441e-05,
+      "loss": 0.3675,
+      "step": 3216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.603318930550016e-05,
+      "loss": 0.375,
+      "step": 3217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.601832146055206e-05,
+      "loss": 0.3057,
+      "step": 3218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60034504600933e-05,
+      "loss": 0.3078,
+      "step": 3219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.598857630592746e-05,
+      "loss": 0.4028,
+      "step": 3220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.597369899985847e-05,
+      "loss": 0.2977,
+      "step": 3221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595881854369064e-05,
+      "loss": 0.2884,
+      "step": 3222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.594393493922871e-05,
+      "loss": 0.363,
+      "step": 3223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.592904818827775e-05,
+      "loss": 0.4287,
+      "step": 3224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.591415829264321e-05,
+      "loss": 0.2574,
+      "step": 3225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.589926525413097e-05,
+      "loss": 0.2941,
+      "step": 3226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588436907454726e-05,
+      "loss": 0.3085,
+      "step": 3227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.586946975569871e-05,
+      "loss": 0.3659,
+      "step": 3228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.585456729939226e-05,
+      "loss": 0.2783,
+      "step": 3229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.583966170743532e-05,
+      "loss": 0.3999,
+      "step": 3230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.582475298163565e-05,
+      "loss": 0.2406,
+      "step": 3231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.580984112380137e-05,
+      "loss": 0.3345,
+      "step": 3232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579492613574101e-05,
+      "loss": 0.3063,
+      "step": 3233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.578000801926346e-05,
+      "loss": 0.3197,
+      "step": 3234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.576508677617799e-05,
+      "loss": 0.3619,
+      "step": 3235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.575016240829425e-05,
+      "loss": 0.3222,
+      "step": 3236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.573523491742229e-05,
+      "loss": 0.367,
+      "step": 3237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.572030430537249e-05,
+      "loss": 0.34,
+      "step": 3238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.570537057395566e-05,
+      "loss": 0.3929,
+      "step": 3239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.569043372498297e-05,
+      "loss": 0.3454,
+      "step": 3240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.567549376026597e-05,
+      "loss": 0.3645,
+      "step": 3241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.566055068161655e-05,
+      "loss": 0.3523,
+      "step": 3242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564560449084705e-05,
+      "loss": 0.3184,
+      "step": 3243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.563065518977016e-05,
+      "loss": 0.45,
+      "step": 3244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.561570278019889e-05,
+      "loss": 0.3968,
+      "step": 3245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.56007472639467e-05,
+      "loss": 0.3216,
+      "step": 3246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55857886428274e-05,
+      "loss": 0.3732,
+      "step": 3247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.557082691865516e-05,
+      "loss": 0.299,
+      "step": 3248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.555586209324455e-05,
+      "loss": 0.3162,
+      "step": 3249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.554089416841052e-05,
+      "loss": 0.382,
+      "step": 3250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.552592314596838e-05,
+      "loss": 0.332,
+      "step": 3251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551094902773383e-05,
+      "loss": 0.2819,
+      "step": 3252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549597181552292e-05,
+      "loss": 0.3363,
+      "step": 3253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.548099151115209e-05,
+      "loss": 0.3753,
+      "step": 3254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546600811643816e-05,
+      "loss": 0.3338,
+      "step": 3255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.545102163319833e-05,
+      "loss": 0.3284,
+      "step": 3256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.543603206325015e-05,
+      "loss": 0.3838,
+      "step": 3257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.542103940841157e-05,
+      "loss": 0.3404,
+      "step": 3258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.540604367050091e-05,
+      "loss": 0.2941,
+      "step": 3259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539104485133686e-05,
+      "loss": 0.3035,
+      "step": 3260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537604295273847e-05,
+      "loss": 0.4008,
+      "step": 3261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.536103797652516e-05,
+      "loss": 0.3295,
+      "step": 3262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.534602992451678e-05,
+      "loss": 0.2991,
+      "step": 3263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.533101879853348e-05,
+      "loss": 0.3079,
+      "step": 3264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.531600460039581e-05,
+      "loss": 0.3817,
+      "step": 3265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.530098733192473e-05,
+      "loss": 0.3901,
+      "step": 3266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52859669949415e-05,
+      "loss": 0.3519,
+      "step": 3267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.527094359126783e-05,
+      "loss": 0.3114,
+      "step": 3268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.525591712272574e-05,
+      "loss": 0.3053,
+      "step": 3269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.524088759113764e-05,
+      "loss": 0.2717,
+      "step": 3270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.522585499832633e-05,
+      "loss": 0.3262,
+      "step": 3271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.521081934611496e-05,
+      "loss": 0.367,
+      "step": 3272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.519578063632707e-05,
+      "loss": 0.2887,
+      "step": 3273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.518073887078654e-05,
+      "loss": 0.391,
+      "step": 3274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.516569405131766e-05,
+      "loss": 0.3813,
+      "step": 3275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.515064617974507e-05,
+      "loss": 0.2964,
+      "step": 3276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513559525789378e-05,
+      "loss": 0.3987,
+      "step": 3277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.512054128758914e-05,
+      "loss": 0.3725,
+      "step": 3278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.510548427065693e-05,
+      "loss": 0.3616,
+      "step": 3279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.509042420892328e-05,
+      "loss": 0.3529,
+      "step": 3280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.507536110421465e-05,
+      "loss": 0.368,
+      "step": 3281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.506029495835793e-05,
+      "loss": 0.3488,
+      "step": 3282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.504522577318031e-05,
+      "loss": 0.2894,
+      "step": 3283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.503015355050942e-05,
+      "loss": 0.2996,
+      "step": 3284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50150782921732e-05,
+      "loss": 0.4314,
+      "step": 3285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.3065,
+      "step": 3286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.498491867581852e-05,
+      "loss": 0.3343,
+      "step": 3287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.496983432145781e-05,
+      "loss": 0.3265,
+      "step": 3288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49547469387473e-05,
+      "loss": 0.3457,
+      "step": 3289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.493965652951685e-05,
+      "loss": 0.3837,
+      "step": 3290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492456309559656e-05,
+      "loss": 0.3444,
+      "step": 3291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.490946663881698e-05,
+      "loss": 0.3549,
+      "step": 3292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.489436716100907e-05,
+      "loss": 0.3003,
+      "step": 3293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.487926466400402e-05,
+      "loss": 0.2957,
+      "step": 3294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.486415914963353e-05,
+      "loss": 0.3655,
+      "step": 3295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.484905061972956e-05,
+      "loss": 0.3778,
+      "step": 3296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.48339390761245e-05,
+      "loss": 0.3343,
+      "step": 3297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.481882452065108e-05,
+      "loss": 0.3268,
+      "step": 3298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.480370695514238e-05,
+      "loss": 0.2892,
+      "step": 3299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.478858638143188e-05,
+      "loss": 0.4149,
+      "step": 3300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.477346280135343e-05,
+      "loss": 0.3316,
+      "step": 3301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.47583362167412e-05,
+      "loss": 0.3585,
+      "step": 3302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.474320662942972e-05,
+      "loss": 0.2858,
+      "step": 3303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472807404125394e-05,
+      "loss": 0.3813,
+      "step": 3304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.471293845404916e-05,
+      "loss": 0.3326,
+      "step": 3305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4697799869651e-05,
+      "loss": 0.295,
+      "step": 3306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46826582898955e-05,
+      "loss": 0.373,
+      "step": 3307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4667513716619e-05,
+      "loss": 0.3463,
+      "step": 3308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.465236615165826e-05,
+      "loss": 0.2921,
+      "step": 3309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46372155968504e-05,
+      "loss": 0.3374,
+      "step": 3310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.462206205403284e-05,
+      "loss": 0.3585,
+      "step": 3311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.460690552504345e-05,
+      "loss": 0.2879,
+      "step": 3312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45917460117204e-05,
+      "loss": 0.3289,
+      "step": 3313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.457658351590224e-05,
+      "loss": 0.2948,
+      "step": 3314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.456141803942787e-05,
+      "loss": 0.348,
+      "step": 3315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45462495841366e-05,
+      "loss": 0.2899,
+      "step": 3316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.453107815186803e-05,
+      "loss": 0.2727,
+      "step": 3317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.451590374446217e-05,
+      "loss": 0.3847,
+      "step": 3318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.450072636375939e-05,
+      "loss": 0.25,
+      "step": 3319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.448554601160038e-05,
+      "loss": 0.3174,
+      "step": 3320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447036268982623e-05,
+      "loss": 0.2895,
+      "step": 3321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.44551764002784e-05,
+      "loss": 0.2268,
+      "step": 3322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.443998714479865e-05,
+      "loss": 0.327,
+      "step": 3323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.442479492522918e-05,
+      "loss": 0.3685,
+      "step": 3324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.440959974341246e-05,
+      "loss": 0.3738,
+      "step": 3325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43944016011914e-05,
+      "loss": 0.4244,
+      "step": 3326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.437920050040922e-05,
+      "loss": 0.3696,
+      "step": 3327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.436399644290953e-05,
+      "loss": 0.2833,
+      "step": 3328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434878943053627e-05,
+      "loss": 0.2735,
+      "step": 3329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.433357946513378e-05,
+      "loss": 0.292,
+      "step": 3330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.431836654854667e-05,
+      "loss": 0.3273,
+      "step": 3331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.430315068262001e-05,
+      "loss": 0.2092,
+      "step": 3332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42879318691992e-05,
+      "loss": 0.3788,
+      "step": 3333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.427271011012996e-05,
+      "loss": 0.3092,
+      "step": 3334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.425748540725838e-05,
+      "loss": 0.4146,
+      "step": 3335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424225776243092e-05,
+      "loss": 0.2758,
+      "step": 3336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.422702717749442e-05,
+      "loss": 0.3657,
+      "step": 3337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.421179365429601e-05,
+      "loss": 0.3198,
+      "step": 3338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.419655719468326e-05,
+      "loss": 0.2932,
+      "step": 3339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.418131780050404e-05,
+      "loss": 0.3234,
+      "step": 3340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416607547360655e-05,
+      "loss": 0.3058,
+      "step": 3341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415083021583944e-05,
+      "loss": 0.3383,
+      "step": 3342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413558202905164e-05,
+      "loss": 0.2714,
+      "step": 3343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.412033091509242e-05,
+      "loss": 0.3758,
+      "step": 3344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.41050768758115e-05,
+      "loss": 0.3103,
+      "step": 3345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.408981991305885e-05,
+      "loss": 0.2766,
+      "step": 3346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407456002868485e-05,
+      "loss": 0.3095,
+      "step": 3347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405929722454026e-05,
+      "loss": 0.3299,
+      "step": 3348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.40440315024761e-05,
+      "loss": 0.2235,
+      "step": 3349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.402876286434384e-05,
+      "loss": 0.3419,
+      "step": 3350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401349131199525e-05,
+      "loss": 0.42,
+      "step": 3351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.399821684728247e-05,
+      "loss": 0.3101,
+      "step": 3352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.398293947205801e-05,
+      "loss": 0.2762,
+      "step": 3353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.396765918817467e-05,
+      "loss": 0.3761,
+      "step": 3354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.395237599748571e-05,
+      "loss": 0.2868,
+      "step": 3355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.393708990184463e-05,
+      "loss": 0.2832,
+      "step": 3356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.392180090310535e-05,
+      "loss": 0.3444,
+      "step": 3357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390650900312214e-05,
+      "loss": 0.3522,
+      "step": 3358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.38912142037496e-05,
+      "loss": 0.3153,
+      "step": 3359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.387591650684267e-05,
+      "loss": 0.2809,
+      "step": 3360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.386061591425666e-05,
+      "loss": 0.3684,
+      "step": 3361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.384531242784726e-05,
+      "loss": 0.2957,
+      "step": 3362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383000604947046e-05,
+      "loss": 0.31,
+      "step": 3363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.381469678098261e-05,
+      "loss": 0.2806,
+      "step": 3364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.379938462424046e-05,
+      "loss": 0.3072,
+      "step": 3365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.378406958110105e-05,
+      "loss": 0.2895,
+      "step": 3366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.376875165342179e-05,
+      "loss": 0.3076,
+      "step": 3367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.375343084306046e-05,
+      "loss": 0.2473,
+      "step": 3368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.373810715187516e-05,
+      "loss": 0.3714,
+      "step": 3369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.372278058172434e-05,
+      "loss": 0.3305,
+      "step": 3370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.370745113446684e-05,
+      "loss": 0.2878,
+      "step": 3371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.369211881196181e-05,
+      "loss": 0.3236,
+      "step": 3372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.367678361606875e-05,
+      "loss": 0.2377,
+      "step": 3373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.366144554864752e-05,
+      "loss": 0.3191,
+      "step": 3374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.364610461155835e-05,
+      "loss": 0.3294,
+      "step": 3375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.363076080666177e-05,
+      "loss": 0.3267,
+      "step": 3376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.361541413581866e-05,
+      "loss": 0.3476,
+      "step": 3377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.360006460089031e-05,
+      "loss": 0.3357,
+      "step": 3378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.358471220373832e-05,
+      "loss": 0.2602,
+      "step": 3379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356935694622459e-05,
+      "loss": 0.3255,
+      "step": 3380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.355399883021144e-05,
+      "loss": 0.3318,
+      "step": 3381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.353863785756151e-05,
+      "loss": 0.3313,
+      "step": 3382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35232740301378e-05,
+      "loss": 0.3257,
+      "step": 3383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.350790734980358e-05,
+      "loss": 0.3001,
+      "step": 3384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.349253781842258e-05,
+      "loss": 0.275,
+      "step": 3385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347716543785881e-05,
+      "loss": 0.3089,
+      "step": 3386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.346179020997662e-05,
+      "loss": 0.3612,
+      "step": 3387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.344641213664073e-05,
+      "loss": 0.3127,
+      "step": 3388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.343103121971623e-05,
+      "loss": 0.3557,
+      "step": 3389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.341564746106848e-05,
+      "loss": 0.3618,
+      "step": 3390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.340026086256324e-05,
+      "loss": 0.2829,
+      "step": 3391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338487142606663e-05,
+      "loss": 0.3608,
+      "step": 3392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.336947915344502e-05,
+      "loss": 0.2853,
+      "step": 3393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.335408404656527e-05,
+      "loss": 0.3258,
+      "step": 3394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333868610729445e-05,
+      "loss": 0.3119,
+      "step": 3395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.332328533750005e-05,
+      "loss": 0.4148,
+      "step": 3396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.330788173904985e-05,
+      "loss": 0.3013,
+      "step": 3397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.329247531381205e-05,
+      "loss": 0.3396,
+      "step": 3398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327706606365513e-05,
+      "loss": 0.3713,
+      "step": 3399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.326165399044789e-05,
+      "loss": 0.3535,
+      "step": 3400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324623909605957e-05,
+      "loss": 0.3273,
+      "step": 3401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.323082138235965e-05,
+      "loss": 0.3097,
+      "step": 3402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3215400851218e-05,
+      "loss": 0.2606,
+      "step": 3403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319997750450486e-05,
+      "loss": 0.3364,
+      "step": 3404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.318455134409075e-05,
+      "loss": 0.3261,
+      "step": 3405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.316912237184657e-05,
+      "loss": 0.3366,
+      "step": 3406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315369058964354e-05,
+      "loss": 0.265,
+      "step": 3407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.313825599935327e-05,
+      "loss": 0.3355,
+      "step": 3408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312281860284763e-05,
+      "loss": 0.4157,
+      "step": 3409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.310737840199885e-05,
+      "loss": 0.2792,
+      "step": 3410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.309193539867958e-05,
+      "loss": 0.2595,
+      "step": 3411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.307648959476274e-05,
+      "loss": 0.3656,
+      "step": 3412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.306104099212157e-05,
+      "loss": 0.3497,
+      "step": 3413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.304558959262973e-05,
+      "loss": 0.2871,
+      "step": 3414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.303013539816113e-05,
+      "loss": 0.2378,
+      "step": 3415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.301467841059008e-05,
+      "loss": 0.2398,
+      "step": 3416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.29992186317912e-05,
+      "loss": 0.3804,
+      "step": 3417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.298375606363946e-05,
+      "loss": 0.3314,
+      "step": 3418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.296829070801016e-05,
+      "loss": 0.4095,
+      "step": 3419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.295282256677896e-05,
+      "loss": 0.2937,
+      "step": 3420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.293735164182182e-05,
+      "loss": 0.3833,
+      "step": 3421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292187793501507e-05,
+      "loss": 0.3439,
+      "step": 3422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.290640144823537e-05,
+      "loss": 0.2761,
+      "step": 3423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.28909221833597e-05,
+      "loss": 0.369,
+      "step": 3424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.287544014226543e-05,
+      "loss": 0.4108,
+      "step": 3425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.285995532683016e-05,
+      "loss": 0.3944,
+      "step": 3426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.284446773893195e-05,
+      "loss": 0.3742,
+      "step": 3427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.282897738044912e-05,
+      "loss": 0.3411,
+      "step": 3428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.281348425326033e-05,
+      "loss": 0.2772,
+      "step": 3429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27979883592446e-05,
+      "loss": 0.3877,
+      "step": 3430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.278248970028131e-05,
+      "loss": 0.2846,
+      "step": 3431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27669882782501e-05,
+      "loss": 0.2755,
+      "step": 3432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.275148409503102e-05,
+      "loss": 0.2918,
+      "step": 3433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.273597715250439e-05,
+      "loss": 0.3261,
+      "step": 3434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.272046745255093e-05,
+      "loss": 0.2522,
+      "step": 3435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.270495499705163e-05,
+      "loss": 0.3132,
+      "step": 3436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.268943978788788e-05,
+      "loss": 0.2724,
+      "step": 3437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.267392182694133e-05,
+      "loss": 0.3405,
+      "step": 3438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.265840111609404e-05,
+      "loss": 0.3866,
+      "step": 3439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264287765722835e-05,
+      "loss": 0.402,
+      "step": 3440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.262735145222696e-05,
+      "loss": 0.2699,
+      "step": 3441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.261182250297286e-05,
+      "loss": 0.3356,
+      "step": 3442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.259629081134944e-05,
+      "loss": 0.3129,
+      "step": 3443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.25807563792404e-05,
+      "loss": 0.45,
+      "step": 3444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.256521920852971e-05,
+      "loss": 0.3878,
+      "step": 3445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254967930110176e-05,
+      "loss": 0.2461,
+      "step": 3446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.253413665884125e-05,
+      "loss": 0.279,
+      "step": 3447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251859128363317e-05,
+      "loss": 0.29,
+      "step": 3448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.250304317736286e-05,
+      "loss": 0.3821,
+      "step": 3449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.248749234191601e-05,
+      "loss": 0.3414,
+      "step": 3450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247193877917867e-05,
+      "loss": 0.3216,
+      "step": 3451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.24563824910371e-05,
+      "loss": 0.3306,
+      "step": 3452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.244082347937803e-05,
+      "loss": 0.2822,
+      "step": 3453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.242526174608847e-05,
+      "loss": 0.2916,
+      "step": 3454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.240969729305572e-05,
+      "loss": 0.315,
+      "step": 3455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.239413012216745e-05,
+      "loss": 0.2841,
+      "step": 3456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.237856023531166e-05,
+      "loss": 0.2975,
+      "step": 3457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.236298763437666e-05,
+      "loss": 0.3262,
+      "step": 3458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.234741232125111e-05,
+      "loss": 0.321,
+      "step": 3459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233183429782399e-05,
+      "loss": 0.3041,
+      "step": 3460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23162535659846e-05,
+      "loss": 0.3312,
+      "step": 3461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.230067012762258e-05,
+      "loss": 0.2507,
+      "step": 3462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.228508398462789e-05,
+      "loss": 0.336,
+      "step": 3463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.226949513889083e-05,
+      "loss": 0.3617,
+      "step": 3464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.225390359230202e-05,
+      "loss": 0.4633,
+      "step": 3465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223830934675239e-05,
+      "loss": 0.257,
+      "step": 3466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.222271240413326e-05,
+      "loss": 0.3404,
+      "step": 3467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.220711276633618e-05,
+      "loss": 0.2877,
+      "step": 3468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.21915104352531e-05,
+      "loss": 0.3385,
+      "step": 3469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.21759054127763e-05,
+      "loss": 0.319,
+      "step": 3470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.216029770079831e-05,
+      "loss": 0.293,
+      "step": 3471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.214468730121208e-05,
+      "loss": 0.2618,
+      "step": 3472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.212907421591084e-05,
+      "loss": 0.2478,
+      "step": 3473
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.211345844678816e-05,
+      "loss": 0.3871,
+      "step": 3474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.209783999573788e-05,
+      "loss": 0.253,
+      "step": 3475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.208221886465424e-05,
+      "loss": 0.3089,
+      "step": 3476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.206659505543179e-05,
+      "loss": 0.2589,
+      "step": 3477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.205096856996538e-05,
+      "loss": 0.2421,
+      "step": 3478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.203533941015018e-05,
+      "loss": 0.3114,
+      "step": 3479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.201970757788172e-05,
+      "loss": 0.4257,
+      "step": 3480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.200407307505582e-05,
+      "loss": 0.3651,
+      "step": 3481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.198843590356866e-05,
+      "loss": 0.2552,
+      "step": 3482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197279606531669e-05,
+      "loss": 0.3848,
+      "step": 3483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.195715356219672e-05,
+      "loss": 0.2909,
+      "step": 3484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.194150839610589e-05,
+      "loss": 0.3749,
+      "step": 3485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.192586056894166e-05,
+      "loss": 0.3369,
+      "step": 3486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.191021008260178e-05,
+      "loss": 0.4533,
+      "step": 3487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189455693898436e-05,
+      "loss": 0.3067,
+      "step": 3488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.187890113998782e-05,
+      "loss": 0.2721,
+      "step": 3489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18632426875109e-05,
+      "loss": 0.3793,
+      "step": 3490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.184758158345267e-05,
+      "loss": 0.335,
+      "step": 3491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183191782971248e-05,
+      "loss": 0.2817,
+      "step": 3492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.181625142819007e-05,
+      "loss": 0.3584,
+      "step": 3493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.180058238078544e-05,
+      "loss": 0.3576,
+      "step": 3494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.178491068939896e-05,
+      "loss": 0.3874,
+      "step": 3495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.17692363559313e-05,
+      "loss": 0.2685,
+      "step": 3496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.175355938228343e-05,
+      "loss": 0.2389,
+      "step": 3497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.173787977035667e-05,
+      "loss": 0.3094,
+      "step": 3498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.172219752205264e-05,
+      "loss": 0.2518,
+      "step": 3499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.170651263927331e-05,
+      "loss": 0.2954,
+      "step": 3500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169082512392093e-05,
+      "loss": 0.3543,
+      "step": 3501
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167513497789806e-05,
+      "loss": 0.3661,
+      "step": 3502
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.165944220310767e-05,
+      "loss": 0.3937,
+      "step": 3503
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.164374680145293e-05,
+      "loss": 0.349,
+      "step": 3504
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.16280487748374e-05,
+      "loss": 0.4036,
+      "step": 3505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.161234812516494e-05,
+      "loss": 0.3103,
+      "step": 3506
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.159664485433975e-05,
+      "loss": 0.3524,
+      "step": 3507
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.158093896426631e-05,
+      "loss": 0.3048,
+      "step": 3508
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.156523045684944e-05,
+      "loss": 0.2988,
+      "step": 3509
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.154951933399427e-05,
+      "loss": 0.2509,
+      "step": 3510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.153380559760625e-05,
+      "loss": 0.3019,
+      "step": 3511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.151808924959113e-05,
+      "loss": 0.3258,
+      "step": 3512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.150237029185505e-05,
+      "loss": 0.2995,
+      "step": 3513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.148664872630435e-05,
+      "loss": 0.2627,
+      "step": 3514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.147092455484577e-05,
+      "loss": 0.2956,
+      "step": 3515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.145519777938635e-05,
+      "loss": 0.2727,
+      "step": 3516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.143946840183342e-05,
+      "loss": 0.2809,
+      "step": 3517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.142373642409466e-05,
+      "loss": 0.298,
+      "step": 3518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.140800184807805e-05,
+      "loss": 0.3972,
+      "step": 3519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139226467569188e-05,
+      "loss": 0.31,
+      "step": 3520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137652490884476e-05,
+      "loss": 0.2716,
+      "step": 3521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.13607825494456e-05,
+      "loss": 0.3033,
+      "step": 3522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.134503759940367e-05,
+      "loss": 0.2832,
+      "step": 3523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.132929006062848e-05,
+      "loss": 0.3687,
+      "step": 3524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.131353993502993e-05,
+      "loss": 0.3507,
+      "step": 3525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12977872245182e-05,
+      "loss": 0.3071,
+      "step": 3526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.128203193100377e-05,
+      "loss": 0.3361,
+      "step": 3527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.126627405639744e-05,
+      "loss": 0.3457,
+      "step": 3528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.125051360261036e-05,
+      "loss": 0.2769,
+      "step": 3529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.123475057155393e-05,
+      "loss": 0.3077,
+      "step": 3530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.121898496513993e-05,
+      "loss": 0.3093,
+      "step": 3531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.120321678528039e-05,
+      "loss": 0.3084,
+      "step": 3532
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.118744603388769e-05,
+      "loss": 0.3526,
+      "step": 3533
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117167271287453e-05,
+      "loss": 0.2794,
+      "step": 3534
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115589682415385e-05,
+      "loss": 0.2973,
+      "step": 3535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.114011836963901e-05,
+      "loss": 0.2461,
+      "step": 3536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11243373512436e-05,
+      "loss": 0.2834,
+      "step": 3537
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.110855377088155e-05,
+      "loss": 0.3104,
+      "step": 3538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.109276763046713e-05,
+      "loss": 0.301,
+      "step": 3539
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.107697893191483e-05,
+      "loss": 0.2985,
+      "step": 3540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106118767713954e-05,
+      "loss": 0.3761,
+      "step": 3541
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.104539386805644e-05,
+      "loss": 0.3628,
+      "step": 3542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.102959750658098e-05,
+      "loss": 0.31,
+      "step": 3543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.101379859462898e-05,
+      "loss": 0.2652,
+      "step": 3544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09979971341165e-05,
+      "loss": 0.3294,
+      "step": 3545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.098219312695997e-05,
+      "loss": 0.2704,
+      "step": 3546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096638657507612e-05,
+      "loss": 0.2657,
+      "step": 3547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.095057748038194e-05,
+      "loss": 0.2768,
+      "step": 3548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.093476584479479e-05,
+      "loss": 0.2622,
+      "step": 3549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09189516702323e-05,
+      "loss": 0.2945,
+      "step": 3550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09031349586124e-05,
+      "loss": 0.3256,
+      "step": 3551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.088731571185336e-05,
+      "loss": 0.2951,
+      "step": 3552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087149393187376e-05,
+      "loss": 0.2797,
+      "step": 3553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085566962059247e-05,
+      "loss": 0.3103,
+      "step": 3554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083984277992863e-05,
+      "loss": 0.3002,
+      "step": 3555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.082401341180176e-05,
+      "loss": 0.3736,
+      "step": 3556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.080818151813164e-05,
+      "loss": 0.2122,
+      "step": 3557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079234710083837e-05,
+      "loss": 0.3021,
+      "step": 3558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077651016184235e-05,
+      "loss": 0.3185,
+      "step": 3559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.076067070306429e-05,
+      "loss": 0.3223,
+      "step": 3560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.074482872642518e-05,
+      "loss": 0.3298,
+      "step": 3561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.072898423384639e-05,
+      "loss": 0.2534,
+      "step": 3562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071313722724953e-05,
+      "loss": 0.2699,
+      "step": 3563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.069728770855651e-05,
+      "loss": 0.4213,
+      "step": 3564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.068143567968957e-05,
+      "loss": 0.3302,
+      "step": 3565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.066558114257127e-05,
+      "loss": 0.377,
+      "step": 3566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.064972409912443e-05,
+      "loss": 0.221,
+      "step": 3567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06338645512722e-05,
+      "loss": 0.2986,
+      "step": 3568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061800250093804e-05,
+      "loss": 0.3925,
+      "step": 3569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.060213795004571e-05,
+      "loss": 0.3197,
+      "step": 3570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.058627090051924e-05,
+      "loss": 0.2493,
+      "step": 3571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.057040135428303e-05,
+      "loss": 0.3123,
+      "step": 3572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.055452931326172e-05,
+      "loss": 0.3011,
+      "step": 3573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053865477938027e-05,
+      "loss": 0.362,
+      "step": 3574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.052277775456397e-05,
+      "loss": 0.3465,
+      "step": 3575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05068982407384e-05,
+      "loss": 0.3242,
+      "step": 3576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.049101623982937e-05,
+      "loss": 0.2949,
+      "step": 3577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.047513175376312e-05,
+      "loss": 0.2355,
+      "step": 3578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.045924478446612e-05,
+      "loss": 0.3759,
+      "step": 3579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.044335533386512e-05,
+      "loss": 0.2934,
+      "step": 3580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.04274634038872e-05,
+      "loss": 0.3278,
+      "step": 3581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.041156899645976e-05,
+      "loss": 0.3577,
+      "step": 3582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.039567211351044e-05,
+      "loss": 0.2543,
+      "step": 3583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.037977275696726e-05,
+      "loss": 0.3059,
+      "step": 3584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.036387092875849e-05,
+      "loss": 0.2766,
+      "step": 3585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.034796663081269e-05,
+      "loss": 0.2914,
+      "step": 3586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.033205986505876e-05,
+      "loss": 0.3467,
+      "step": 3587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.031615063342587e-05,
+      "loss": 0.3934,
+      "step": 3588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03002389378435e-05,
+      "loss": 0.315,
+      "step": 3589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02843247802414e-05,
+      "loss": 0.2772,
+      "step": 3590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02684081625497e-05,
+      "loss": 0.3166,
+      "step": 3591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.025248908669873e-05,
+      "loss": 0.2883,
+      "step": 3592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.023656755461916e-05,
+      "loss": 0.2998,
+      "step": 3593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.022064356824197e-05,
+      "loss": 0.3045,
+      "step": 3594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.020471712949842e-05,
+      "loss": 0.285,
+      "step": 3595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.018878824032009e-05,
+      "loss": 0.284,
+      "step": 3596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.017285690263882e-05,
+      "loss": 0.275,
+      "step": 3597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015692311838678e-05,
+      "loss": 0.3153,
+      "step": 3598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.014098688949642e-05,
+      "loss": 0.2296,
+      "step": 3599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.012504821790051e-05,
+      "loss": 0.4254,
+      "step": 3600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.010910710553207e-05,
+      "loss": 0.3395,
+      "step": 3601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009316355432444e-05,
+      "loss": 0.2865,
+      "step": 3602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.007721756621129e-05,
+      "loss": 0.29,
+      "step": 3603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.006126914312653e-05,
+      "loss": 0.3123,
+      "step": 3604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.004531828700442e-05,
+      "loss": 0.3892,
+      "step": 3605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002936499977944e-05,
+      "loss": 0.2869,
+      "step": 3606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.001340928338643e-05,
+      "loss": 0.2766,
+      "step": 3607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.999745113976052e-05,
+      "loss": 0.2912,
+      "step": 3608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99814905708371e-05,
+      "loss": 0.2966,
+      "step": 3609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.996552757855188e-05,
+      "loss": 0.3524,
+      "step": 3610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.994956216484086e-05,
+      "loss": 0.296,
+      "step": 3611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.993359433164032e-05,
+      "loss": 0.3381,
+      "step": 3612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.991762408088686e-05,
+      "loss": 0.4114,
+      "step": 3613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.990165141451736e-05,
+      "loss": 0.2965,
+      "step": 3614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.988567633446897e-05,
+      "loss": 0.3247,
+      "step": 3615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.986969884267915e-05,
+      "loss": 0.2871,
+      "step": 3616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985371894108567e-05,
+      "loss": 0.2679,
+      "step": 3617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983773663162658e-05,
+      "loss": 0.3381,
+      "step": 3618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.982175191624022e-05,
+      "loss": 0.32,
+      "step": 3619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98057647968652e-05,
+      "loss": 0.3443,
+      "step": 3620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978977527544048e-05,
+      "loss": 0.3186,
+      "step": 3621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.977378335390526e-05,
+      "loss": 0.4121,
+      "step": 3622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975778903419901e-05,
+      "loss": 0.2882,
+      "step": 3623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.974179231826157e-05,
+      "loss": 0.2913,
+      "step": 3624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.972579320803303e-05,
+      "loss": 0.3013,
+      "step": 3625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970979170545372e-05,
+      "loss": 0.2518,
+      "step": 3626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.969378781246436e-05,
+      "loss": 0.2911,
+      "step": 3627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.967778153100588e-05,
+      "loss": 0.306,
+      "step": 3628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.966177286301954e-05,
+      "loss": 0.2827,
+      "step": 3629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.964576181044687e-05,
+      "loss": 0.3892,
+      "step": 3630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.962974837522969e-05,
+      "loss": 0.3131,
+      "step": 3631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.961373255931011e-05,
+      "loss": 0.3138,
+      "step": 3632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.959771436463054e-05,
+      "loss": 0.3148,
+      "step": 3633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.958169379313369e-05,
+      "loss": 0.3649,
+      "step": 3634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.956567084676252e-05,
+      "loss": 0.3781,
+      "step": 3635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954964552746031e-05,
+      "loss": 0.3766,
+      "step": 3636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.95336178371706e-05,
+      "loss": 0.3373,
+      "step": 3637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951758777783726e-05,
+      "loss": 0.2795,
+      "step": 3638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.950155535140439e-05,
+      "loss": 0.3648,
+      "step": 3639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.948552055981642e-05,
+      "loss": 0.3357,
+      "step": 3640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.946948340501806e-05,
+      "loss": 0.336,
+      "step": 3641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.945344388895431e-05,
+      "loss": 0.3344,
+      "step": 3642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.94374020135704e-05,
+      "loss": 0.3685,
+      "step": 3643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.942135778081197e-05,
+      "loss": 0.2946,
+      "step": 3644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940531119262483e-05,
+      "loss": 0.2825,
+      "step": 3645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93892622509551e-05,
+      "loss": 0.2924,
+      "step": 3646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.937321095774922e-05,
+      "loss": 0.3943,
+      "step": 3647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.935715731495389e-05,
+      "loss": 0.3376,
+      "step": 3648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93411013245161e-05,
+      "loss": 0.3746,
+      "step": 3649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.932504298838314e-05,
+      "loss": 0.3429,
+      "step": 3650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.930898230850256e-05,
+      "loss": 0.3169,
+      "step": 3651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.92929192868222e-05,
+      "loss": 0.3358,
+      "step": 3652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.927685392529018e-05,
+      "loss": 0.338,
+      "step": 3653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.926078622585495e-05,
+      "loss": 0.3102,
+      "step": 3654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.924471619046517e-05,
+      "loss": 0.3041,
+      "step": 3655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.922864382106984e-05,
+      "loss": 0.3709,
+      "step": 3656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.92125691196182e-05,
+      "loss": 0.2419,
+      "step": 3657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919649208805981e-05,
+      "loss": 0.2811,
+      "step": 3658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.918041272834451e-05,
+      "loss": 0.2412,
+      "step": 3659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.916433104242237e-05,
+      "loss": 0.2894,
+      "step": 3660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.914824703224383e-05,
+      "loss": 0.3675,
+      "step": 3661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.913216069975954e-05,
+      "loss": 0.2927,
+      "step": 3662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.911607204692045e-05,
+      "loss": 0.3538,
+      "step": 3663
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.909998107567782e-05,
+      "loss": 0.3476,
+      "step": 3664
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908388778798313e-05,
+      "loss": 0.4167,
+      "step": 3665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.906779218578822e-05,
+      "loss": 0.3622,
+      "step": 3666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.905169427104515e-05,
+      "loss": 0.3007,
+      "step": 3667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.903559404570628e-05,
+      "loss": 0.3617,
+      "step": 3668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.901949151172426e-05,
+      "loss": 0.3438,
+      "step": 3669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.900338667105199e-05,
+      "loss": 0.2969,
+      "step": 3670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.898727952564271e-05,
+      "loss": 0.3522,
+      "step": 3671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.897117007744986e-05,
+      "loss": 0.3298,
+      "step": 3672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.895505832842722e-05,
+      "loss": 0.2611,
+      "step": 3673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.89389442805288e-05,
+      "loss": 0.2701,
+      "step": 3674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.892282793570897e-05,
+      "loss": 0.3086,
+      "step": 3675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.890670929592227e-05,
+      "loss": 0.3536,
+      "step": 3676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.889058836312362e-05,
+      "loss": 0.2666,
+      "step": 3677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887446513926813e-05,
+      "loss": 0.3342,
+      "step": 3678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885833962631127e-05,
+      "loss": 0.3128,
+      "step": 3679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.884221182620872e-05,
+      "loss": 0.2549,
+      "step": 3680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.882608174091647e-05,
+      "loss": 0.3334,
+      "step": 3681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.880994937239079e-05,
+      "loss": 0.4318,
+      "step": 3682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.879381472258821e-05,
+      "loss": 0.2941,
+      "step": 3683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.877767779346555e-05,
+      "loss": 0.2748,
+      "step": 3684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.87615385869799e-05,
+      "loss": 0.2787,
+      "step": 3685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.874539710508865e-05,
+      "loss": 0.2658,
+      "step": 3686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.872925334974942e-05,
+      "loss": 0.2901,
+      "step": 3687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.871310732292013e-05,
+      "loss": 0.2543,
+      "step": 3688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.869695902655897e-05,
+      "loss": 0.2762,
+      "step": 3689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.868080846262443e-05,
+      "loss": 0.3426,
+      "step": 3690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.866465563307525e-05,
+      "loss": 0.3404,
+      "step": 3691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864850053987044e-05,
+      "loss": 0.3471,
+      "step": 3692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.863234318496931e-05,
+      "loss": 0.2481,
+      "step": 3693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.861618357033142e-05,
+      "loss": 0.3011,
+      "step": 3694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.860002169791662e-05,
+      "loss": 0.3146,
+      "step": 3695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.858385756968501e-05,
+      "loss": 0.3489,
+      "step": 3696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.856769118759701e-05,
+      "loss": 0.2758,
+      "step": 3697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.855152255361326e-05,
+      "loss": 0.3156,
+      "step": 3698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.853535166969469e-05,
+      "loss": 0.372,
+      "step": 3699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851917853780254e-05,
+      "loss": 0.3102,
+      "step": 3700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.850300315989828e-05,
+      "loss": 0.2626,
+      "step": 3701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.848682553794365e-05,
+      "loss": 0.3237,
+      "step": 3702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.847064567390071e-05,
+      "loss": 0.2968,
+      "step": 3703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.845446356973173e-05,
+      "loss": 0.2771,
+      "step": 3704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843827922739928e-05,
+      "loss": 0.3496,
+      "step": 3705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.842209264886622e-05,
+      "loss": 0.3161,
+      "step": 3706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.840590383609567e-05,
+      "loss": 0.234,
+      "step": 3707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.838971279105098e-05,
+      "loss": 0.278,
+      "step": 3708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.837351951569583e-05,
+      "loss": 0.2912,
+      "step": 3709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.835732401199415e-05,
+      "loss": 0.3621,
+      "step": 3710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.834112628191014e-05,
+      "loss": 0.3101,
+      "step": 3711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832492632740823e-05,
+      "loss": 0.3366,
+      "step": 3712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.83087241504532e-05,
+      "loss": 0.3652,
+      "step": 3713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.829251975301003e-05,
+      "loss": 0.2842,
+      "step": 3714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.827631313704399e-05,
+      "loss": 0.2658,
+      "step": 3715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.826010430452064e-05,
+      "loss": 0.3208,
+      "step": 3716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82438932574058e-05,
+      "loss": 0.3242,
+      "step": 3717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.822767999766553e-05,
+      "loss": 0.3046,
+      "step": 3718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.821146452726617e-05,
+      "loss": 0.2799,
+      "step": 3719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.819524684817438e-05,
+      "loss": 0.286,
+      "step": 3720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.817902696235702e-05,
+      "loss": 0.2876,
+      "step": 3721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.816280487178122e-05,
+      "loss": 0.3106,
+      "step": 3722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814658057841444e-05,
+      "loss": 0.2942,
+      "step": 3723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.813035408422436e-05,
+      "loss": 0.3053,
+      "step": 3724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.811412539117889e-05,
+      "loss": 0.3915,
+      "step": 3725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.809789450124633e-05,
+      "loss": 0.3292,
+      "step": 3726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.808166141639511e-05,
+      "loss": 0.2681,
+      "step": 3727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8065426138594e-05,
+      "loss": 0.3569,
+      "step": 3728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.804918866981203e-05,
+      "loss": 0.3514,
+      "step": 3729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.803294901201845e-05,
+      "loss": 0.308,
+      "step": 3730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.801670716718285e-05,
+      "loss": 0.396,
+      "step": 3731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.800046313727503e-05,
+      "loss": 0.3009,
+      "step": 3732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798421692426508e-05,
+      "loss": 0.284,
+      "step": 3733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796796853012333e-05,
+      "loss": 0.2749,
+      "step": 3734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.79517179568204e-05,
+      "loss": 0.3347,
+      "step": 3735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.793546520632719e-05,
+      "loss": 0.3224,
+      "step": 3736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.791921028061479e-05,
+      "loss": 0.2682,
+      "step": 3737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790295318165462e-05,
+      "loss": 0.2786,
+      "step": 3738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.788669391141836e-05,
+      "loss": 0.2789,
+      "step": 3739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.787043247187794e-05,
+      "loss": 0.2929,
+      "step": 3740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.785416886500555e-05,
+      "loss": 0.3183,
+      "step": 3741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.783790309277363e-05,
+      "loss": 0.3002,
+      "step": 3742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.782163515715491e-05,
+      "loss": 0.2464,
+      "step": 3743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.780536506012238e-05,
+      "loss": 0.329,
+      "step": 3744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778909280364927e-05,
+      "loss": 0.256,
+      "step": 3745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.777281838970908e-05,
+      "loss": 0.4026,
+      "step": 3746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77565418202756e-05,
+      "loss": 0.3417,
+      "step": 3747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.774026309732282e-05,
+      "loss": 0.2906,
+      "step": 3748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.772398222282506e-05,
+      "loss": 0.3639,
+      "step": 3749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.770769919875687e-05,
+      "loss": 0.3667,
+      "step": 3750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.769141402709305e-05,
+      "loss": 0.3266,
+      "step": 3751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.767512670980864e-05,
+      "loss": 0.3086,
+      "step": 3752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.765883724887905e-05,
+      "loss": 0.2854,
+      "step": 3753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.764254564627981e-05,
+      "loss": 0.3207,
+      "step": 3754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.762625190398677e-05,
+      "loss": 0.3029,
+      "step": 3755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.760995602397606e-05,
+      "loss": 0.2988,
+      "step": 3756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759365800822405e-05,
+      "loss": 0.2904,
+      "step": 3757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757735785870737e-05,
+      "loss": 0.3307,
+      "step": 3758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.756105557740288e-05,
+      "loss": 0.2398,
+      "step": 3759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.754475116628778e-05,
+      "loss": 0.2996,
+      "step": 3760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.752844462733943e-05,
+      "loss": 0.303,
+      "step": 3761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.75121359625355e-05,
+      "loss": 0.3131,
+      "step": 3762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.749582517385393e-05,
+      "loss": 0.2824,
+      "step": 3763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.747951226327287e-05,
+      "loss": 0.3783,
+      "step": 3764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.746319723277075e-05,
+      "loss": 0.3278,
+      "step": 3765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.744688008432632e-05,
+      "loss": 0.3914,
+      "step": 3766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.743056081991847e-05,
+      "loss": 0.307,
+      "step": 3767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.741423944152641e-05,
+      "loss": 0.2676,
+      "step": 3768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.739791595112965e-05,
+      "loss": 0.3775,
+      "step": 3769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.738159035070785e-05,
+      "loss": 0.3042,
+      "step": 3770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.736526264224101e-05,
+      "loss": 0.2911,
+      "step": 3771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.734893282770937e-05,
+      "loss": 0.4075,
+      "step": 3772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.733260090909341e-05,
+      "loss": 0.3197,
+      "step": 3773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.731626688837387e-05,
+      "loss": 0.3097,
+      "step": 3774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.729993076753171e-05,
+      "loss": 0.2832,
+      "step": 3775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.728359254854825e-05,
+      "loss": 0.2773,
+      "step": 3776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.726725223340494e-05,
+      "loss": 0.3075,
+      "step": 3777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.725090982408355e-05,
+      "loss": 0.2833,
+      "step": 3778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723456532256612e-05,
+      "loss": 0.3346,
+      "step": 3779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.721821873083487e-05,
+      "loss": 0.2479,
+      "step": 3780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720187005087235e-05,
+      "loss": 0.2801,
+      "step": 3781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.718551928466132e-05,
+      "loss": 0.2555,
+      "step": 3782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.716916643418484e-05,
+      "loss": 0.2969,
+      "step": 3783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.715281150142614e-05,
+      "loss": 0.2776,
+      "step": 3784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.713645448836878e-05,
+      "loss": 0.2446,
+      "step": 3785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.712009539699656e-05,
+      "loss": 0.2812,
+      "step": 3786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.710373422929346e-05,
+      "loss": 0.3032,
+      "step": 3787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.708737098724383e-05,
+      "loss": 0.2302,
+      "step": 3788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.707100567283217e-05,
+      "loss": 0.3122,
+      "step": 3789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.705463828804328e-05,
+      "loss": 0.2803,
+      "step": 3790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.70382688348622e-05,
+      "loss": 0.2569,
+      "step": 3791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.702189731527424e-05,
+      "loss": 0.2834,
+      "step": 3792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.700552373126493e-05,
+      "loss": 0.2732,
+      "step": 3793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.698914808482006e-05,
+      "loss": 0.3847,
+      "step": 3794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.697277037792569e-05,
+      "loss": 0.3081,
+      "step": 3795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.695639061256809e-05,
+      "loss": 0.3314,
+      "step": 3796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694000879073383e-05,
+      "loss": 0.3763,
+      "step": 3797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.692362491440966e-05,
+      "loss": 0.2708,
+      "step": 3798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.690723898558267e-05,
+      "loss": 0.3517,
+      "step": 3799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.689085100624013e-05,
+      "loss": 0.3029,
+      "step": 3800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.687446097836957e-05,
+      "loss": 0.2132,
+      "step": 3801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.685806890395881e-05,
+      "loss": 0.2933,
+      "step": 3802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684167478499586e-05,
+      "loss": 0.3479,
+      "step": 3803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682527862346898e-05,
+      "loss": 0.3212,
+      "step": 3804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.680888042136675e-05,
+      "loss": 0.2756,
+      "step": 3805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.679248018067794e-05,
+      "loss": 0.3551,
+      "step": 3806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.677607790339157e-05,
+      "loss": 0.2728,
+      "step": 3807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.675967359149689e-05,
+      "loss": 0.2886,
+      "step": 3808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.674326724698345e-05,
+      "loss": 0.2932,
+      "step": 3809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.672685887184101e-05,
+      "loss": 0.3676,
+      "step": 3810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.671044846805958e-05,
+      "loss": 0.2456,
+      "step": 3811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669403603762943e-05,
+      "loss": 0.4083,
+      "step": 3812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667762158254104e-05,
+      "loss": 0.2238,
+      "step": 3813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.666120510478517e-05,
+      "loss": 0.2854,
+      "step": 3814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.664478660635284e-05,
+      "loss": 0.2769,
+      "step": 3815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.662836608923526e-05,
+      "loss": 0.3146,
+      "step": 3816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.661194355542394e-05,
+      "loss": 0.286,
+      "step": 3817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65955190069106e-05,
+      "loss": 0.2519,
+      "step": 3818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.657909244568721e-05,
+      "loss": 0.2921,
+      "step": 3819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.656266387374598e-05,
+      "loss": 0.2664,
+      "step": 3820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.654623329307939e-05,
+      "loss": 0.3325,
+      "step": 3821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652980070568014e-05,
+      "loss": 0.3143,
+      "step": 3822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65133661135412e-05,
+      "loss": 0.2428,
+      "step": 3823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649692951865571e-05,
+      "loss": 0.28,
+      "step": 3824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.648049092301716e-05,
+      "loss": 0.2937,
+      "step": 3825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.646405032861922e-05,
+      "loss": 0.3691,
+      "step": 3826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.644760773745577e-05,
+      "loss": 0.364,
+      "step": 3827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.643116315152102e-05,
+      "loss": 0.2974,
+      "step": 3828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.641471657280937e-05,
+      "loss": 0.2868,
+      "step": 3829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.63982680033154e-05,
+      "loss": 0.2602,
+      "step": 3830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.63818174450341e-05,
+      "loss": 0.385,
+      "step": 3831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.636536489996054e-05,
+      "loss": 0.3097,
+      "step": 3832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.63489103700901e-05,
+      "loss": 0.2819,
+      "step": 3833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.633245385741838e-05,
+      "loss": 0.3689,
+      "step": 3834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.631599536394126e-05,
+      "loss": 0.2361,
+      "step": 3835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.62995348916548e-05,
+      "loss": 0.3297,
+      "step": 3836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.628307244255533e-05,
+      "loss": 0.2859,
+      "step": 3837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.626660801863946e-05,
+      "loss": 0.3147,
+      "step": 3838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.625014162190397e-05,
+      "loss": 0.3232,
+      "step": 3839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.623367325434591e-05,
+      "loss": 0.3331,
+      "step": 3840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621720291796259e-05,
+      "loss": 0.3062,
+      "step": 3841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.62007306147515e-05,
+      "loss": 0.3339,
+      "step": 3842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618425634671046e-05,
+      "loss": 0.273,
+      "step": 3843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.616778011583743e-05,
+      "loss": 0.349,
+      "step": 3844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615130192413067e-05,
+      "loss": 0.3228,
+      "step": 3845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.613482177358866e-05,
+      "loss": 0.2584,
+      "step": 3846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.611833966621011e-05,
+      "loss": 0.3884,
+      "step": 3847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6101855603994e-05,
+      "loss": 0.2476,
+      "step": 3848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.608536958893948e-05,
+      "loss": 0.2734,
+      "step": 3849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6068881623046e-05,
+      "loss": 0.3415,
+      "step": 3850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.605239170831327e-05,
+      "loss": 0.2564,
+      "step": 3851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.603589984674113e-05,
+      "loss": 0.2913,
+      "step": 3852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.601940604032973e-05,
+      "loss": 0.3772,
+      "step": 3853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.600291029107946e-05,
+      "loss": 0.2882,
+      "step": 3854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.598641260099095e-05,
+      "loss": 0.3366,
+      "step": 3855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596991297206498e-05,
+      "loss": 0.1897,
+      "step": 3856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.595341140630269e-05,
+      "loss": 0.3259,
+      "step": 3857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.593690790570538e-05,
+      "loss": 0.2973,
+      "step": 3858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59204024722746e-05,
+      "loss": 0.2785,
+      "step": 3859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.590389510801212e-05,
+      "loss": 0.2587,
+      "step": 3860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.588738581491997e-05,
+      "loss": 0.2177,
+      "step": 3861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58708745950004e-05,
+      "loss": 0.2929,
+      "step": 3862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58543614502559e-05,
+      "loss": 0.3987,
+      "step": 3863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58378463826892e-05,
+      "loss": 0.3031,
+      "step": 3864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.582132939430322e-05,
+      "loss": 0.3943,
+      "step": 3865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.580481048710119e-05,
+      "loss": 0.2337,
+      "step": 3866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.578828966308649e-05,
+      "loss": 0.3859,
+      "step": 3867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.577176692426279e-05,
+      "loss": 0.2104,
+      "step": 3868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.575524227263397e-05,
+      "loss": 0.3275,
+      "step": 3869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.573871571020413e-05,
+      "loss": 0.3001,
+      "step": 3870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.572218723897764e-05,
+      "loss": 0.2898,
+      "step": 3871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.570565686095907e-05,
+      "loss": 0.3296,
+      "step": 3872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.568912457815321e-05,
+      "loss": 0.2652,
+      "step": 3873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.567259039256516e-05,
+      "loss": 0.2816,
+      "step": 3874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.565605430620013e-05,
+      "loss": 0.296,
+      "step": 3875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.563951632106365e-05,
+      "loss": 0.2835,
+      "step": 3876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.562297643916145e-05,
+      "loss": 0.3442,
+      "step": 3877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.560643466249947e-05,
+      "loss": 0.2594,
+      "step": 3878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.558989099308393e-05,
+      "loss": 0.3531,
+      "step": 3879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.557334543292124e-05,
+      "loss": 0.2334,
+      "step": 3880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.555679798401806e-05,
+      "loss": 0.2896,
+      "step": 3881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554024864838126e-05,
+      "loss": 0.2905,
+      "step": 3882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.552369742801795e-05,
+      "loss": 0.3565,
+      "step": 3883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.550714432493549e-05,
+      "loss": 0.363,
+      "step": 3884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.549058934114141e-05,
+      "loss": 0.2911,
+      "step": 3885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.54740324786435e-05,
+      "loss": 0.3088,
+      "step": 3886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545747373944983e-05,
+      "loss": 0.298,
+      "step": 3887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.544091312556862e-05,
+      "loss": 0.3239,
+      "step": 3888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.542435063900834e-05,
+      "loss": 0.2875,
+      "step": 3889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.540778628177769e-05,
+      "loss": 0.2406,
+      "step": 3890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.539122005588562e-05,
+      "loss": 0.2937,
+      "step": 3891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.537465196334129e-05,
+      "loss": 0.3588,
+      "step": 3892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.535808200615406e-05,
+      "loss": 0.3655,
+      "step": 3893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.534151018633355e-05,
+      "loss": 0.3023,
+      "step": 3894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.532493650588962e-05,
+      "loss": 0.2723,
+      "step": 3895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.530836096683228e-05,
+      "loss": 0.3812,
+      "step": 3896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.529178357117186e-05,
+      "loss": 0.2863,
+      "step": 3897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.527520432091887e-05,
+      "loss": 0.274,
+      "step": 3898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.525862321808402e-05,
+      "loss": 0.3072,
+      "step": 3899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52420402646783e-05,
+      "loss": 0.3005,
+      "step": 3900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.522545546271289e-05,
+      "loss": 0.2948,
+      "step": 3901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.520886881419919e-05,
+      "loss": 0.258,
+      "step": 3902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.519228032114883e-05,
+      "loss": 0.2755,
+      "step": 3903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.517568998557368e-05,
+      "loss": 0.3807,
+      "step": 3904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.515909780948582e-05,
+      "loss": 0.2186,
+      "step": 3905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.514250379489753e-05,
+      "loss": 0.2719,
+      "step": 3906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.512590794382139e-05,
+      "loss": 0.3333,
+      "step": 3907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.510931025827011e-05,
+      "loss": 0.3744,
+      "step": 3908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.509271074025668e-05,
+      "loss": 0.3015,
+      "step": 3909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.507610939179429e-05,
+      "loss": 0.3086,
+      "step": 3910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.505950621489634e-05,
+      "loss": 0.4587,
+      "step": 3911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.504290121157651e-05,
+      "loss": 0.2709,
+      "step": 3912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50262943838486e-05,
+      "loss": 0.2363,
+      "step": 3913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.500968573372673e-05,
+      "loss": 0.4251,
+      "step": 3914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.499307526322522e-05,
+      "loss": 0.238,
+      "step": 3915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.497646297435855e-05,
+      "loss": 0.3299,
+      "step": 3916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.495984886914151e-05,
+      "loss": 0.3126,
+      "step": 3917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.494323294958903e-05,
+      "loss": 0.2509,
+      "step": 3918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.49266152177163e-05,
+      "loss": 0.3286,
+      "step": 3919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.490999567553872e-05,
+      "loss": 0.2905,
+      "step": 3920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.489337432507195e-05,
+      "loss": 0.2365,
+      "step": 3921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.487675116833178e-05,
+      "loss": 0.2287,
+      "step": 3922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.48601262073343e-05,
+      "loss": 0.2682,
+      "step": 3923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.484349944409578e-05,
+      "loss": 0.3361,
+      "step": 3924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.482687088063275e-05,
+      "loss": 0.3276,
+      "step": 3925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.48102405189619e-05,
+      "loss": 0.3003,
+      "step": 3926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.479360836110016e-05,
+      "loss": 0.385,
+      "step": 3927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.47769744090647e-05,
+      "loss": 0.3117,
+      "step": 3928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.476033866487286e-05,
+      "loss": 0.2723,
+      "step": 3929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.474370113054228e-05,
+      "loss": 0.2486,
+      "step": 3930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472706180809074e-05,
+      "loss": 0.2831,
+      "step": 3931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.471042069953625e-05,
+      "loss": 0.2703,
+      "step": 3932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.469377780689708e-05,
+      "loss": 0.3369,
+      "step": 3933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.467713313219166e-05,
+      "loss": 0.2631,
+      "step": 3934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.466048667743865e-05,
+      "loss": 0.2148,
+      "step": 3935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.464383844465698e-05,
+      "loss": 0.2421,
+      "step": 3936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.462718843586571e-05,
+      "loss": 0.2556,
+      "step": 3937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.461053665308419e-05,
+      "loss": 0.345,
+      "step": 3938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.459388309833192e-05,
+      "loss": 0.2998,
+      "step": 3939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.45772277736287e-05,
+      "loss": 0.3368,
+      "step": 3940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456057068099446e-05,
+      "loss": 0.2435,
+      "step": 3941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.454391182244937e-05,
+      "loss": 0.2512,
+      "step": 3942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.452725120001382e-05,
+      "loss": 0.2234,
+      "step": 3943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.451058881570845e-05,
+      "loss": 0.2548,
+      "step": 3944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.449392467155404e-05,
+      "loss": 0.2948,
+      "step": 3945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.447725876957165e-05,
+      "loss": 0.3008,
+      "step": 3946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.446059111178253e-05,
+      "loss": 0.388,
+      "step": 3947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44439217002081e-05,
+      "loss": 0.3256,
+      "step": 3948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.442725053687008e-05,
+      "loss": 0.2341,
+      "step": 3949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.441057762379034e-05,
+      "loss": 0.2661,
+      "step": 3950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.439390296299095e-05,
+      "loss": 0.2654,
+      "step": 3951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.437722655649422e-05,
+      "loss": 0.2539,
+      "step": 3952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.436054840632273e-05,
+      "loss": 0.2482,
+      "step": 3953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.434386851449913e-05,
+      "loss": 0.2877,
+      "step": 3954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.432718688304641e-05,
+      "loss": 0.3095,
+      "step": 3955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.431050351398775e-05,
+      "loss": 0.2937,
+      "step": 3956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.429381840934644e-05,
+      "loss": 0.314,
+      "step": 3957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.427713157114612e-05,
+      "loss": 0.3312,
+      "step": 3958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.426044300141054e-05,
+      "loss": 0.2562,
+      "step": 3959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.42437527021637e-05,
+      "loss": 0.2535,
+      "step": 3960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.422706067542982e-05,
+      "loss": 0.3521,
+      "step": 3961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.421036692323329e-05,
+      "loss": 0.2763,
+      "step": 3962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.419367144759876e-05,
+      "loss": 0.4107,
+      "step": 3963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.417697425055105e-05,
+      "loss": 0.225,
+      "step": 3964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.416027533411519e-05,
+      "loss": 0.3869,
+      "step": 3965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.414357470031648e-05,
+      "loss": 0.2874,
+      "step": 3966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.412687235118032e-05,
+      "loss": 0.2878,
+      "step": 3967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.411016828873239e-05,
+      "loss": 0.3016,
+      "step": 3968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.409346251499857e-05,
+      "loss": 0.3096,
+      "step": 3969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.407675503200496e-05,
+      "loss": 0.336,
+      "step": 3970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406004584177784e-05,
+      "loss": 0.2734,
+      "step": 3971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.404333494634367e-05,
+      "loss": 0.3175,
+      "step": 3972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.40266223477292e-05,
+      "loss": 0.3646,
+      "step": 3973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.400990804796134e-05,
+      "loss": 0.3377,
+      "step": 3974
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.399319204906716e-05,
+      "loss": 0.2329,
+      "step": 3975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.397647435307402e-05,
+      "loss": 0.2799,
+      "step": 3976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.395975496200943e-05,
+      "loss": 0.2392,
+      "step": 3977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.394303387790113e-05,
+      "loss": 0.2875,
+      "step": 3978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.392631110277707e-05,
+      "loss": 0.3349,
+      "step": 3979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.390958663866538e-05,
+      "loss": 0.3003,
+      "step": 3980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.38928604875944e-05,
+      "loss": 0.368,
+      "step": 3981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.387613265159271e-05,
+      "loss": 0.2552,
+      "step": 3982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.385940313268905e-05,
+      "loss": 0.3225,
+      "step": 3983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.384267193291238e-05,
+      "loss": 0.2498,
+      "step": 3984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.382593905429187e-05,
+      "loss": 0.3338,
+      "step": 3985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.380920449885688e-05,
+      "loss": 0.3561,
+      "step": 3986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3792468268637e-05,
+      "loss": 0.2724,
+      "step": 3987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.377573036566198e-05,
+      "loss": 0.2936,
+      "step": 3988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.375899079196184e-05,
+      "loss": 0.3443,
+      "step": 3989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.374224954956673e-05,
+      "loss": 0.263,
+      "step": 3990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.372550664050703e-05,
+      "loss": 0.3659,
+      "step": 3991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.370876206681335e-05,
+      "loss": 0.2877,
+      "step": 3992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.369201583051645e-05,
+      "loss": 0.3501,
+      "step": 3993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.367526793364733e-05,
+      "loss": 0.3289,
+      "step": 3994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.365851837823718e-05,
+      "loss": 0.3264,
+      "step": 3995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.36417671663174e-05,
+      "loss": 0.3523,
+      "step": 3996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362501429991957e-05,
+      "loss": 0.3149,
+      "step": 3997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.360825978107548e-05,
+      "loss": 0.2821,
+      "step": 3998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.359150361181715e-05,
+      "loss": 0.2347,
+      "step": 3999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.357474579417675e-05,
+      "loss": 0.2413,
+      "step": 4000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.355798633018668e-05,
+      "loss": 0.3111,
+      "step": 4001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.354122522187952e-05,
+      "loss": 0.3001,
+      "step": 4002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.352446247128811e-05,
+      "loss": 0.2712,
+      "step": 4003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.350769808044539e-05,
+      "loss": 0.3412,
+      "step": 4004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.349093205138455e-05,
+      "loss": 0.3729,
+      "step": 4005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.347416438613903e-05,
+      "loss": 0.2549,
+      "step": 4006
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.345739508674237e-05,
+      "loss": 0.2769,
+      "step": 4007
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.344062415522837e-05,
+      "loss": 0.3213,
+      "step": 4008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.342385159363103e-05,
+      "loss": 0.4021,
+      "step": 4009
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.34070774039845e-05,
+      "loss": 0.3093,
+      "step": 4010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.339030158832319e-05,
+      "loss": 0.342,
+      "step": 4011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.337352414868166e-05,
+      "loss": 0.2967,
+      "step": 4012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.335674508709469e-05,
+      "loss": 0.2885,
+      "step": 4013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.333996440559725e-05,
+      "loss": 0.306,
+      "step": 4014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.332318210622451e-05,
+      "loss": 0.2446,
+      "step": 4015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.330639819101183e-05,
+      "loss": 0.3604,
+      "step": 4016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.328961266199476e-05,
+      "loss": 0.314,
+      "step": 4017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.327282552120905e-05,
+      "loss": 0.2697,
+      "step": 4018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.325603677069066e-05,
+      "loss": 0.2764,
+      "step": 4019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.323924641247575e-05,
+      "loss": 0.2846,
+      "step": 4020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.322245444860063e-05,
+      "loss": 0.3453,
+      "step": 4021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.320566088110187e-05,
+      "loss": 0.2787,
+      "step": 4022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.318886571201619e-05,
+      "loss": 0.2683,
+      "step": 4023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.317206894338048e-05,
+      "loss": 0.269,
+      "step": 4024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.315527057723189e-05,
+      "loss": 0.3338,
+      "step": 4025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.313847061560773e-05,
+      "loss": 0.3548,
+      "step": 4026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.312166906054549e-05,
+      "loss": 0.3044,
+      "step": 4027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.310486591408287e-05,
+      "loss": 0.2501,
+      "step": 4028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.308806117825778e-05,
+      "loss": 0.4383,
+      "step": 4029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.307125485510828e-05,
+      "loss": 0.2772,
+      "step": 4030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.305444694667268e-05,
+      "loss": 0.3359,
+      "step": 4031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30376374549894e-05,
+      "loss": 0.2609,
+      "step": 4032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.302082638209714e-05,
+      "loss": 0.3911,
+      "step": 4033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.300401373003474e-05,
+      "loss": 0.3323,
+      "step": 4034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.298719950084125e-05,
+      "loss": 0.3251,
+      "step": 4035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.29703836965559e-05,
+      "loss": 0.2792,
+      "step": 4036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.295356631921809e-05,
+      "loss": 0.382,
+      "step": 4037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.293674737086747e-05,
+      "loss": 0.2471,
+      "step": 4038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.291992685354386e-05,
+      "loss": 0.2945,
+      "step": 4039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.290310476928723e-05,
+      "loss": 0.3365,
+      "step": 4040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.28862811201378e-05,
+      "loss": 0.377,
+      "step": 4041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.28694559081359e-05,
+      "loss": 0.2389,
+      "step": 4042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.285262913532211e-05,
+      "loss": 0.3129,
+      "step": 4043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.283580080373721e-05,
+      "loss": 0.3384,
+      "step": 4044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.281897091542212e-05,
+      "loss": 0.2703,
+      "step": 4045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2802139472418e-05,
+      "loss": 0.3194,
+      "step": 4046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.278530647676617e-05,
+      "loss": 0.3339,
+      "step": 4047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.276847193050813e-05,
+      "loss": 0.3504,
+      "step": 4048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.275163583568557e-05,
+      "loss": 0.3098,
+      "step": 4049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.273479819434039e-05,
+      "loss": 0.3134,
+      "step": 4050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.271795900851467e-05,
+      "loss": 0.296,
+      "step": 4051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.270111828025067e-05,
+      "loss": 0.2606,
+      "step": 4052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.268427601159082e-05,
+      "loss": 0.2716,
+      "step": 4053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.266743220457778e-05,
+      "loss": 0.198,
+      "step": 4054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.265058686125435e-05,
+      "loss": 0.3001,
+      "step": 4055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263373998366359e-05,
+      "loss": 0.2558,
+      "step": 4056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.261689157384863e-05,
+      "loss": 0.3294,
+      "step": 4057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.260004163385286e-05,
+      "loss": 0.3112,
+      "step": 4058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.25831901657199e-05,
+      "loss": 0.2822,
+      "step": 4059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.256633717149343e-05,
+      "loss": 0.3545,
+      "step": 4060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.254948265321744e-05,
+      "loss": 0.2987,
+      "step": 4061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.253262661293604e-05,
+      "loss": 0.203,
+      "step": 4062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.251576905269352e-05,
+      "loss": 0.3519,
+      "step": 4063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.249890997453437e-05,
+      "loss": 0.3013,
+      "step": 4064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.24820493805033e-05,
+      "loss": 0.3125,
+      "step": 4065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.246518727264514e-05,
+      "loss": 0.2618,
+      "step": 4066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.24483236530049e-05,
+      "loss": 0.2998,
+      "step": 4067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.243145852362788e-05,
+      "loss": 0.2924,
+      "step": 4068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.241459188655943e-05,
+      "loss": 0.3452,
+      "step": 4069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.239772374384516e-05,
+      "loss": 0.2588,
+      "step": 4070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.238085409753086e-05,
+      "loss": 0.2659,
+      "step": 4071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.236398294966247e-05,
+      "loss": 0.3139,
+      "step": 4072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.234711030228611e-05,
+      "loss": 0.3254,
+      "step": 4073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.233023615744812e-05,
+      "loss": 0.2587,
+      "step": 4074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.231336051719503e-05,
+      "loss": 0.3313,
+      "step": 4075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.229648338357347e-05,
+      "loss": 0.2193,
+      "step": 4076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.227960475863033e-05,
+      "loss": 0.2888,
+      "step": 4077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.226272464441267e-05,
+      "loss": 0.3177,
+      "step": 4078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22458430429677e-05,
+      "loss": 0.231,
+      "step": 4079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22289599563428e-05,
+      "loss": 0.3394,
+      "step": 4080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22120753865856e-05,
+      "loss": 0.2885,
+      "step": 4081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.219518933574386e-05,
+      "loss": 0.321,
+      "step": 4082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.217830180586548e-05,
+      "loss": 0.3227,
+      "step": 4083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.216141279899865e-05,
+      "loss": 0.268,
+      "step": 4084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.214452231719163e-05,
+      "loss": 0.265,
+      "step": 4085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.212763036249293e-05,
+      "loss": 0.3523,
+      "step": 4086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.211073693695119e-05,
+      "loss": 0.3376,
+      "step": 4087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.209384204261526e-05,
+      "loss": 0.3117,
+      "step": 4088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.207694568153417e-05,
+      "loss": 0.2113,
+      "step": 4089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.206004785575709e-05,
+      "loss": 0.2784,
+      "step": 4090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.204314856733342e-05,
+      "loss": 0.2797,
+      "step": 4091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.202624781831268e-05,
+      "loss": 0.3658,
+      "step": 4092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.200934561074464e-05,
+      "loss": 0.2994,
+      "step": 4093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.199244194667918e-05,
+      "loss": 0.3491,
+      "step": 4094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.197553682816637e-05,
+      "loss": 0.2612,
+      "step": 4095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.195863025725649e-05,
+      "loss": 0.3607,
+      "step": 4096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.194172223599998e-05,
+      "loss": 0.2785,
+      "step": 4097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.192481276644743e-05,
+      "loss": 0.3639,
+      "step": 4098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.190790185064964e-05,
+      "loss": 0.2807,
+      "step": 4099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.189098949065753e-05,
+      "loss": 0.2954,
+      "step": 4100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.187407568852231e-05,
+      "loss": 0.3113,
+      "step": 4101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.185716044629526e-05,
+      "loss": 0.2867,
+      "step": 4102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184024376602783e-05,
+      "loss": 0.3251,
+      "step": 4103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.182332564977174e-05,
+      "loss": 0.3187,
+      "step": 4104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.18064060995788e-05,
+      "loss": 0.3718,
+      "step": 4105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.178948511750098e-05,
+      "loss": 0.2717,
+      "step": 4106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.177256270559052e-05,
+      "loss": 0.3129,
+      "step": 4107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.175563886589975e-05,
+      "loss": 0.2888,
+      "step": 4108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.173871360048121e-05,
+      "loss": 0.2729,
+      "step": 4109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.172178691138756e-05,
+      "loss": 0.3107,
+      "step": 4110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.170485880067172e-05,
+      "loss": 0.2396,
+      "step": 4111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.168792927038675e-05,
+      "loss": 0.2667,
+      "step": 4112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.167099832258582e-05,
+      "loss": 0.2558,
+      "step": 4113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.165406595932236e-05,
+      "loss": 0.3808,
+      "step": 4114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.163713218264991e-05,
+      "loss": 0.2912,
+      "step": 4115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.16201969946222e-05,
+      "loss": 0.3133,
+      "step": 4116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.160326039729316e-05,
+      "loss": 0.2789,
+      "step": 4117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.158632239271686e-05,
+      "loss": 0.3278,
+      "step": 4118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.156938298294753e-05,
+      "loss": 0.3038,
+      "step": 4119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.155244217003958e-05,
+      "loss": 0.3191,
+      "step": 4120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.153549995604764e-05,
+      "loss": 0.2324,
+      "step": 4121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.151855634302643e-05,
+      "loss": 0.3347,
+      "step": 4122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.150161133303089e-05,
+      "loss": 0.2928,
+      "step": 4123
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14846649281161e-05,
+      "loss": 0.2888,
+      "step": 4124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.146771713033737e-05,
+      "loss": 0.3002,
+      "step": 4125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.145076794175008e-05,
+      "loss": 0.3329,
+      "step": 4126
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.143381736440987e-05,
+      "loss": 0.333,
+      "step": 4127
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.141686540037251e-05,
+      "loss": 0.2773,
+      "step": 4128
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.13999120516939e-05,
+      "loss": 0.2617,
+      "step": 4129
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.13829573204302e-05,
+      "loss": 0.2152,
+      "step": 4130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.136600120863766e-05,
+      "loss": 0.2391,
+      "step": 4131
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.134904371837272e-05,
+      "loss": 0.3623,
+      "step": 4132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1332084851692e-05,
+      "loss": 0.2487,
+      "step": 4133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.131512461065227e-05,
+      "loss": 0.1903,
+      "step": 4134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.129816299731048e-05,
+      "loss": 0.2452,
+      "step": 4135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.128120001372372e-05,
+      "loss": 0.2753,
+      "step": 4136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.12642356619493e-05,
+      "loss": 0.2626,
+      "step": 4137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.124726994404465e-05,
+      "loss": 0.3103,
+      "step": 4138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.123030286206736e-05,
+      "loss": 0.3249,
+      "step": 4139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.121333441807522e-05,
+      "loss": 0.2982,
+      "step": 4140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.119636461412617e-05,
+      "loss": 0.2616,
+      "step": 4141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11793934522783e-05,
+      "loss": 0.267,
+      "step": 4142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.116242093458989e-05,
+      "loss": 0.268,
+      "step": 4143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.114544706311935e-05,
+      "loss": 0.2453,
+      "step": 4144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.112847183992531e-05,
+      "loss": 0.2602,
+      "step": 4145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.111149526706653e-05,
+      "loss": 0.3164,
+      "step": 4146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.109451734660191e-05,
+      "loss": 0.3108,
+      "step": 4147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.107753808059055e-05,
+      "loss": 0.1942,
+      "step": 4148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.10605574710917e-05,
+      "loss": 0.2354,
+      "step": 4149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.104357552016476e-05,
+      "loss": 0.3278,
+      "step": 4150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.102659222986934e-05,
+      "loss": 0.283,
+      "step": 4151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.100960760226515e-05,
+      "loss": 0.2085,
+      "step": 4152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.099262163941209e-05,
+      "loss": 0.3164,
+      "step": 4153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0975634343370256e-05,
+      "loss": 0.3026,
+      "step": 4154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0958645716199824e-05,
+      "loss": 0.2233,
+      "step": 4155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.094165575996122e-05,
+      "loss": 0.2488,
+      "step": 4156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0924664476714986e-05,
+      "loss": 0.3091,
+      "step": 4157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.090767186852182e-05,
+      "loss": 0.2038,
+      "step": 4158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0890677937442574e-05,
+      "loss": 0.382,
+      "step": 4159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.087368268553831e-05,
+      "loss": 0.3485,
+      "step": 4160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.085668611487022e-05,
+      "loss": 0.2489,
+      "step": 4161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0839688227499626e-05,
+      "loss": 0.2922,
+      "step": 4162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0822689025488054e-05,
+      "loss": 0.2456,
+      "step": 4163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.080568851089717e-05,
+      "loss": 0.276,
+      "step": 4164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0788686685788785e-05,
+      "loss": 0.3001,
+      "step": 4165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0771683552224925e-05,
+      "loss": 0.2518,
+      "step": 4166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.07546791122677e-05,
+      "loss": 0.2918,
+      "step": 4167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.073767336797943e-05,
+      "loss": 0.345,
+      "step": 4168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.072066632142257e-05,
+      "loss": 0.2929,
+      "step": 4169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0703657974659754e-05,
+      "loss": 0.2837,
+      "step": 4170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.068664832975376e-05,
+      "loss": 0.3333,
+      "step": 4171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.066963738876749e-05,
+      "loss": 0.2622,
+      "step": 4172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0652625153764064e-05,
+      "loss": 0.2951,
+      "step": 4173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.063561162680674e-05,
+      "loss": 0.2953,
+      "step": 4174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0618596809958917e-05,
+      "loss": 0.2308,
+      "step": 4175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0601580705284136e-05,
+      "loss": 0.3074,
+      "step": 4176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.058456331484613e-05,
+      "loss": 0.2477,
+      "step": 4177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.056754464070879e-05,
+      "loss": 0.2448,
+      "step": 4178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.055052468493614e-05,
+      "loss": 0.3261,
+      "step": 4179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.053350344959234e-05,
+      "loss": 0.3836,
+      "step": 4180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.051648093674176e-05,
+      "loss": 0.275,
+      "step": 4181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.049945714844888e-05,
+      "loss": 0.3075,
+      "step": 4182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.048243208677835e-05,
+      "loss": 0.3331,
+      "step": 4183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0465405753795e-05,
+      "loss": 0.3581,
+      "step": 4184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.044837815156377e-05,
+      "loss": 0.2813,
+      "step": 4185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.043134928214975e-05,
+      "loss": 0.33,
+      "step": 4186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.041431914761826e-05,
+      "loss": 0.2967,
+      "step": 4187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.039728775003467e-05,
+      "loss": 0.268,
+      "step": 4188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.038025509146459e-05,
+      "loss": 0.3101,
+      "step": 4189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0363221173973726e-05,
+      "loss": 0.2839,
+      "step": 4190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.034618599962796e-05,
+      "loss": 0.2822,
+      "step": 4191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.032914957049332e-05,
+      "loss": 0.2993,
+      "step": 4192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0312111888636005e-05,
+      "loss": 0.262,
+      "step": 4193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.029507295612234e-05,
+      "loss": 0.3173,
+      "step": 4194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0278032775018825e-05,
+      "loss": 0.3261,
+      "step": 4195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.026099134739207e-05,
+      "loss": 0.3145,
+      "step": 4196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0243948675308895e-05,
+      "loss": 0.3203,
+      "step": 4197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0226904760836213e-05,
+      "loss": 0.2411,
+      "step": 4198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.020985960604115e-05,
+      "loss": 0.3741,
+      "step": 4199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.019281321299092e-05,
+      "loss": 0.3027,
+      "step": 4200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.017576558375293e-05,
+      "loss": 0.3127,
+      "step": 4201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0158716720394706e-05,
+      "loss": 0.3297,
+      "step": 4202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.014166662498396e-05,
+      "loss": 0.194,
+      "step": 4203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0124615299588524e-05,
+      "loss": 0.266,
+      "step": 4204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.010756274627638e-05,
+      "loss": 0.311,
+      "step": 4205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.009050896711569e-05,
+      "loss": 0.3195,
+      "step": 4206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.007345396417472e-05,
+      "loss": 0.2512,
+      "step": 4207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.005639773952191e-05,
+      "loss": 0.2866,
+      "step": 4208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.003934029522584e-05,
+      "loss": 0.3207,
+      "step": 4209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.002228163335527e-05,
+      "loss": 0.2654,
+      "step": 4210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.000522175597906e-05,
+      "loss": 0.3181,
+      "step": 4211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9988160665166225e-05,
+      "loss": 0.2974,
+      "step": 4212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.997109836298597e-05,
+      "loss": 0.3362,
+      "step": 4213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.995403485150759e-05,
+      "loss": 0.2342,
+      "step": 4214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.993697013280055e-05,
+      "loss": 0.2345,
+      "step": 4215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.99199042089345e-05,
+      "loss": 0.3059,
+      "step": 4216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.990283708197917e-05,
+      "loss": 0.2797,
+      "step": 4217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9885768754004465e-05,
+      "loss": 0.3061,
+      "step": 4218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.986869922708047e-05,
+      "loss": 0.272,
+      "step": 4219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985162850327737e-05,
+      "loss": 0.2495,
+      "step": 4220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.983455658466547e-05,
+      "loss": 0.3768,
+      "step": 4221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98174834733153e-05,
+      "loss": 0.2626,
+      "step": 4222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.980040917129748e-05,
+      "loss": 0.2947,
+      "step": 4223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.978333368068278e-05,
+      "loss": 0.2406,
+      "step": 4224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.976625700354212e-05,
+      "loss": 0.2972,
+      "step": 4225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.97491791419466e-05,
+      "loss": 0.3138,
+      "step": 4226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9732100097967394e-05,
+      "loss": 0.2362,
+      "step": 4227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9715019873675845e-05,
+      "loss": 0.3883,
+      "step": 4228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.969793847114349e-05,
+      "loss": 0.2789,
+      "step": 4229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.968085589244192e-05,
+      "loss": 0.3307,
+      "step": 4230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9663772139642936e-05,
+      "loss": 0.2468,
+      "step": 4231
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.964668721481849e-05,
+      "loss": 0.2909,
+      "step": 4232
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.96296011200406e-05,
+      "loss": 0.2308,
+      "step": 4233
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.96125138573815e-05,
+      "loss": 0.3101,
+      "step": 4234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.959542542891352e-05,
+      "loss": 0.2876,
+      "step": 4235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9578335836709195e-05,
+      "loss": 0.2471,
+      "step": 4236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.95612450828411e-05,
+      "loss": 0.2279,
+      "step": 4237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9544153169382035e-05,
+      "loss": 0.2623,
+      "step": 4238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9527060098404905e-05,
+      "loss": 0.1785,
+      "step": 4239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.950996587198277e-05,
+      "loss": 0.3497,
+      "step": 4240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.949287049218882e-05,
+      "loss": 0.272,
+      "step": 4241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9475773961096406e-05,
+      "loss": 0.4075,
+      "step": 4242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945867628077897e-05,
+      "loss": 0.3213,
+      "step": 4243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944157745331015e-05,
+      "loss": 0.2922,
+      "step": 4244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.942447748076369e-05,
+      "loss": 0.3088,
+      "step": 4245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9407376365213465e-05,
+      "loss": 0.288,
+      "step": 4246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.939027410873351e-05,
+      "loss": 0.2704,
+      "step": 4247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.937317071339801e-05,
+      "loss": 0.3031,
+      "step": 4248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9356066181281245e-05,
+      "loss": 0.2529,
+      "step": 4249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9338960514457684e-05,
+      "loss": 0.2679,
+      "step": 4250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9321853715001887e-05,
+      "loss": 0.2946,
+      "step": 4251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.930474578498859e-05,
+      "loss": 0.2837,
+      "step": 4252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.928763672649265e-05,
+      "loss": 0.2718,
+      "step": 4253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.927052654158902e-05,
+      "loss": 0.2291,
+      "step": 4254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.925341523235288e-05,
+      "loss": 0.3688,
+      "step": 4255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.923630280085948e-05,
+      "loss": 0.3627,
+      "step": 4256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.921918924918419e-05,
+      "loss": 0.3119,
+      "step": 4257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.920207457940259e-05,
+      "loss": 0.2649,
+      "step": 4258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.918495879359032e-05,
+      "loss": 0.3689,
+      "step": 4259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9167841893823215e-05,
+      "loss": 0.2384,
+      "step": 4260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9150723882177204e-05,
+      "loss": 0.2693,
+      "step": 4261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9133604760728364e-05,
+      "loss": 0.232,
+      "step": 4262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9116484531552916e-05,
+      "loss": 0.2377,
+      "step": 4263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.909936319672719e-05,
+      "loss": 0.2874,
+      "step": 4264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.908224075832769e-05,
+      "loss": 0.2284,
+      "step": 4265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.906511721843102e-05,
+      "loss": 0.228,
+      "step": 4266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.904799257911393e-05,
+      "loss": 0.304,
+      "step": 4267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.903086684245328e-05,
+      "loss": 0.1905,
+      "step": 4268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.901374001052614e-05,
+      "loss": 0.2315,
+      "step": 4269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.899661208540961e-05,
+      "loss": 0.2615,
+      "step": 4270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.897948306918096e-05,
+      "loss": 0.3419,
+      "step": 4271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.896235296391766e-05,
+      "loss": 0.3373,
+      "step": 4272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8945221771697203e-05,
+      "loss": 0.2958,
+      "step": 4273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.892808949459728e-05,
+      "loss": 0.3116,
+      "step": 4274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.891095613469571e-05,
+      "loss": 0.2336,
+      "step": 4275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.889382169407043e-05,
+      "loss": 0.2722,
+      "step": 4276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.887668617479951e-05,
+      "loss": 0.2517,
+      "step": 4277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.885954957896115e-05,
+      "loss": 0.2984,
+      "step": 4278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.884241190863366e-05,
+      "loss": 0.2275,
+      "step": 4279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.882527316589554e-05,
+      "loss": 0.2538,
+      "step": 4280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8808133352825354e-05,
+      "loss": 0.2633,
+      "step": 4281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.879099247150185e-05,
+      "loss": 0.2091,
+      "step": 4282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.877385052400386e-05,
+      "loss": 0.3091,
+      "step": 4283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.875670751241036e-05,
+      "loss": 0.3185,
+      "step": 4284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.873956343880047e-05,
+      "loss": 0.2617,
+      "step": 4285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8722418305253443e-05,
+      "loss": 0.2698,
+      "step": 4286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.870527211384862e-05,
+      "loss": 0.2894,
+      "step": 4287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8688124866665515e-05,
+      "loss": 0.3442,
+      "step": 4288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.867097656578375e-05,
+      "loss": 0.3356,
+      "step": 4289
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.865382721328308e-05,
+      "loss": 0.2776,
+      "step": 4290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.863667681124334e-05,
+      "loss": 0.2759,
+      "step": 4291
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.861952536174461e-05,
+      "loss": 0.2669,
+      "step": 4292
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.860237286686697e-05,
+      "loss": 0.2269,
+      "step": 4293
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.858521932869071e-05,
+      "loss": 0.2656,
+      "step": 4294
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8568064749296195e-05,
+      "loss": 0.3007,
+      "step": 4295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.855090913076395e-05,
+      "loss": 0.3621,
+      "step": 4296
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8533752475174595e-05,
+      "loss": 0.2831,
+      "step": 4297
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.851659478460892e-05,
+      "loss": 0.2604,
+      "step": 4298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.849943606114782e-05,
+      "loss": 0.3357,
+      "step": 4299
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8482276306872287e-05,
+      "loss": 0.3184,
+      "step": 4300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.846511552386347e-05,
+      "loss": 0.259,
+      "step": 4301
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.844795371420264e-05,
+      "loss": 0.1884,
+      "step": 4302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8430790879971174e-05,
+      "loss": 0.2561,
+      "step": 4303
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.841362702325061e-05,
+      "loss": 0.2341,
+      "step": 4304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.839646214612257e-05,
+      "loss": 0.2734,
+      "step": 4305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.837929625066881e-05,
+      "loss": 0.3152,
+      "step": 4306
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8362129338971216e-05,
+      "loss": 0.221,
+      "step": 4307
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.834496141311181e-05,
+      "loss": 0.3131,
+      "step": 4308
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.832779247517273e-05,
+      "loss": 0.3381,
+      "step": 4309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8310622527236204e-05,
+      "loss": 0.3381,
+      "step": 4310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.829345157138463e-05,
+      "loss": 0.3042,
+      "step": 4311
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.82762796097005e-05,
+      "loss": 0.2759,
+      "step": 4312
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.825910664426643e-05,
+      "loss": 0.2355,
+      "step": 4313
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.824193267716517e-05,
+      "loss": 0.3201,
+      "step": 4314
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.822475771047958e-05,
+      "loss": 0.2513,
+      "step": 4315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.820758174629265e-05,
+      "loss": 0.2993,
+      "step": 4316
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.819040478668748e-05,
+      "loss": 0.293,
+      "step": 4317
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8173226833747315e-05,
+      "loss": 0.3284,
+      "step": 4318
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8156047889555484e-05,
+      "loss": 0.3408,
+      "step": 4319
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8138867956195467e-05,
+      "loss": 0.2629,
+      "step": 4320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8121687035750835e-05,
+      "loss": 0.2376,
+      "step": 4321
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810450513030532e-05,
+      "loss": 0.3414,
+      "step": 4322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808732224194271e-05,
+      "loss": 0.3329,
+      "step": 4323
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807013837274701e-05,
+      "loss": 0.2238,
+      "step": 4324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8052953524802233e-05,
+      "loss": 0.2797,
+      "step": 4325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.803576770019259e-05,
+      "loss": 0.3146,
+      "step": 4326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801858090100238e-05,
+      "loss": 0.3408,
+      "step": 4327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8001393129316016e-05,
+      "loss": 0.3321,
+      "step": 4328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7984204387218046e-05,
+      "loss": 0.2987,
+      "step": 4329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796701467679311e-05,
+      "loss": 0.3038,
+      "step": 4330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7949824000125995e-05,
+      "loss": 0.3016,
+      "step": 4331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79326323593016e-05,
+      "loss": 0.2372,
+      "step": 4332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.791543975640491e-05,
+      "loss": 0.3354,
+      "step": 4333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.789824619352107e-05,
+      "loss": 0.2387,
+      "step": 4334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.788105167273532e-05,
+      "loss": 0.2972,
+      "step": 4335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.786385619613299e-05,
+      "loss": 0.2332,
+      "step": 4336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.784665976579959e-05,
+      "loss": 0.2831,
+      "step": 4337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.78294623838207e-05,
+      "loss": 0.2801,
+      "step": 4338
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.781226405228201e-05,
+      "loss": 0.2793,
+      "step": 4339
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.779506477326933e-05,
+      "loss": 0.3229,
+      "step": 4340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.777786454886862e-05,
+      "loss": 0.2425,
+      "step": 4341
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.776066338116593e-05,
+      "loss": 0.2395,
+      "step": 4342
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.77434612722474e-05,
+      "loss": 0.2993,
+      "step": 4343
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.772625822419932e-05,
+      "loss": 0.3742,
+      "step": 4344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7709054239108084e-05,
+      "loss": 0.2138,
+      "step": 4345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.769184931906019e-05,
+      "loss": 0.2202,
+      "step": 4346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7674643466142264e-05,
+      "loss": 0.2797,
+      "step": 4347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.765743668244105e-05,
+      "loss": 0.2449,
+      "step": 4348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7640228970043355e-05,
+      "loss": 0.169,
+      "step": 4349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.762302033103616e-05,
+      "loss": 0.2708,
+      "step": 4350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.760581076750655e-05,
+      "loss": 0.3023,
+      "step": 4351
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.758860028154168e-05,
+      "loss": 0.3156,
+      "step": 4352
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.757138887522884e-05,
+      "loss": 0.2403,
+      "step": 4353
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.755417655065546e-05,
+      "loss": 0.2341,
+      "step": 4354
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.753696330990904e-05,
+      "loss": 0.2064,
+      "step": 4355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7519749155077196e-05,
+      "loss": 0.2905,
+      "step": 4356
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.750253408824771e-05,
+      "loss": 0.3578,
+      "step": 4357
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.748531811150838e-05,
+      "loss": 0.2984,
+      "step": 4358
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.74681012269472e-05,
+      "loss": 0.2661,
+      "step": 4359
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7450883436652236e-05,
+      "loss": 0.3309,
+      "step": 4360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7433664742711645e-05,
+      "loss": 0.2842,
+      "step": 4361
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7416445147213734e-05,
+      "loss": 0.2404,
+      "step": 4362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.739922465224689e-05,
+      "loss": 0.322,
+      "step": 4363
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.738200325989963e-05,
+      "loss": 0.2999,
+      "step": 4364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.736478097226058e-05,
+      "loss": 0.3768,
+      "step": 4365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7347557791418426e-05,
+      "loss": 0.2565,
+      "step": 4366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7330333719462046e-05,
+      "loss": 0.3356,
+      "step": 4367
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.731310875848037e-05,
+      "loss": 0.3297,
+      "step": 4368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.729588291056242e-05,
+      "loss": 0.2533,
+      "step": 4369
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.72786561777974e-05,
+      "loss": 0.2857,
+      "step": 4370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.726142856227452e-05,
+      "loss": 0.298,
+      "step": 4371
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.724420006608319e-05,
+      "loss": 0.2485,
+      "step": 4372
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.722697069131286e-05,
+      "loss": 0.2764,
+      "step": 4373
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.720974044005314e-05,
+      "loss": 0.422,
+      "step": 4374
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.71925093143937e-05,
+      "loss": 0.2936,
+      "step": 4375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.717527731642437e-05,
+      "loss": 0.2951,
+      "step": 4376
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.715804444823501e-05,
+      "loss": 0.3072,
+      "step": 4377
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.714081071191565e-05,
+      "loss": 0.3992,
+      "step": 4378
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7123576109556384e-05,
+      "loss": 0.3383,
+      "step": 4379
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.710634064324747e-05,
+      "loss": 0.2593,
+      "step": 4380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.70891043150792e-05,
+      "loss": 0.3077,
+      "step": 4381
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.707186712714201e-05,
+      "loss": 0.1981,
+      "step": 4382
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.705462908152641e-05,
+      "loss": 0.2628,
+      "step": 4383
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.703739018032309e-05,
+      "loss": 0.2859,
+      "step": 4384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.702015042562274e-05,
+      "loss": 0.2724,
+      "step": 4385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.70029098195162e-05,
+      "loss": 0.237,
+      "step": 4386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.698566836409447e-05,
+      "loss": 0.2884,
+      "step": 4387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.696842606144854e-05,
+      "loss": 0.3176,
+      "step": 4388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.695118291366959e-05,
+      "loss": 0.2336,
+      "step": 4389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6933938922848885e-05,
+      "loss": 0.2227,
+      "step": 4390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.691669409107776e-05,
+      "loss": 0.3106,
+      "step": 4391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6899448420447684e-05,
+      "loss": 0.2884,
+      "step": 4392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6882201913050226e-05,
+      "loss": 0.3196,
+      "step": 4393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.686495457097702e-05,
+      "loss": 0.3089,
+      "step": 4394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.684770639631987e-05,
+      "loss": 0.3167,
+      "step": 4395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.683045739117059e-05,
+      "loss": 0.2658,
+      "step": 4396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6813207557621196e-05,
+      "loss": 0.3021,
+      "step": 4397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6795956897763725e-05,
+      "loss": 0.2795,
+      "step": 4398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6778705413690335e-05,
+      "loss": 0.3117,
+      "step": 4399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.676145310749333e-05,
+      "loss": 0.2831,
+      "step": 4400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6744199981265045e-05,
+      "loss": 0.2837,
+      "step": 4401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.672694603709794e-05,
+      "loss": 0.2779,
+      "step": 4402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.670969127708459e-05,
+      "loss": 0.2531,
+      "step": 4403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.669243570331766e-05,
+      "loss": 0.2637,
+      "step": 4404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.667517931788991e-05,
+      "loss": 0.2819,
+      "step": 4405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.66579221228942e-05,
+      "loss": 0.2369,
+      "step": 4406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.664066412042348e-05,
+      "loss": 0.2042,
+      "step": 4407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.662340531257083e-05,
+      "loss": 0.365,
+      "step": 4408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.660614570142938e-05,
+      "loss": 0.2543,
+      "step": 4409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6588885289092394e-05,
+      "loss": 0.2766,
+      "step": 4410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.657162407765322e-05,
+      "loss": 0.1391,
+      "step": 4411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.655436206920528e-05,
+      "loss": 0.2916,
+      "step": 4412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.653709926584214e-05,
+      "loss": 0.4519,
+      "step": 4413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.651983566965745e-05,
+      "loss": 0.2552,
+      "step": 4414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.650257128274492e-05,
+      "loss": 0.2385,
+      "step": 4415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.648530610719838e-05,
+      "loss": 0.2925,
+      "step": 4416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.646804014511178e-05,
+      "loss": 0.2945,
+      "step": 4417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.645077339857912e-05,
+      "loss": 0.2669,
+      "step": 4418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.643350586969452e-05,
+      "loss": 0.2394,
+      "step": 4419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6416237560552206e-05,
+      "loss": 0.3258,
+      "step": 4420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6398968473246475e-05,
+      "loss": 0.2745,
+      "step": 4421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.63816986098717e-05,
+      "loss": 0.356,
+      "step": 4422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.636442797252243e-05,
+      "loss": 0.2492,
+      "step": 4423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6347156563293215e-05,
+      "loss": 0.2278,
+      "step": 4424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.632988438427875e-05,
+      "loss": 0.244,
+      "step": 4425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6312611437573816e-05,
+      "loss": 0.2608,
+      "step": 4426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6295337725273266e-05,
+      "loss": 0.3009,
+      "step": 4427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.627806324947206e-05,
+      "loss": 0.2828,
+      "step": 4428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6260788012265277e-05,
+      "loss": 0.2523,
+      "step": 4429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.624351201574805e-05,
+      "loss": 0.3475,
+      "step": 4430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6226235262015605e-05,
+      "loss": 0.305,
+      "step": 4431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6208957753163285e-05,
+      "loss": 0.3501,
+      "step": 4432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.619167949128652e-05,
+      "loss": 0.2726,
+      "step": 4433
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.61744004784808e-05,
+      "loss": 0.2819,
+      "step": 4434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6157120716841747e-05,
+      "loss": 0.2243,
+      "step": 4435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.613984020846506e-05,
+      "loss": 0.2815,
+      "step": 4436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.612255895544651e-05,
+      "loss": 0.3077,
+      "step": 4437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.610527695988198e-05,
+      "loss": 0.2268,
+      "step": 4438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.608799422386743e-05,
+      "loss": 0.2589,
+      "step": 4439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6070710749498946e-05,
+      "loss": 0.1879,
+      "step": 4440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6053426538872636e-05,
+      "loss": 0.2511,
+      "step": 4441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.603614159408475e-05,
+      "loss": 0.2885,
+      "step": 4442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.601885591723162e-05,
+      "loss": 0.2749,
+      "step": 4443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.600156951040964e-05,
+      "loss": 0.2321,
+      "step": 4444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5984282375715304e-05,
+      "loss": 0.3435,
+      "step": 4445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.596699451524523e-05,
+      "loss": 0.2507,
+      "step": 4446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5949705931096086e-05,
+      "loss": 0.2694,
+      "step": 4447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5932416625364625e-05,
+      "loss": 0.298,
+      "step": 4448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5915126600147726e-05,
+      "loss": 0.1756,
+      "step": 4449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5897835857542317e-05,
+      "loss": 0.2382,
+      "step": 4450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.588054439964539e-05,
+      "loss": 0.2297,
+      "step": 4451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.586325222855411e-05,
+      "loss": 0.3154,
+      "step": 4452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.584595934636565e-05,
+      "loss": 0.3345,
+      "step": 4453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5828665755177314e-05,
+      "loss": 0.301,
+      "step": 4454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5811371457086435e-05,
+      "loss": 0.2776,
+      "step": 4455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.579407645419053e-05,
+      "loss": 0.2475,
+      "step": 4456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.57767807485871e-05,
+      "loss": 0.2538,
+      "step": 4457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.575948434237378e-05,
+      "loss": 0.3142,
+      "step": 4458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.57421872376483e-05,
+      "loss": 0.2983,
+      "step": 4459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5724889436508443e-05,
+      "loss": 0.2,
+      "step": 4460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.570759094105209e-05,
+      "loss": 0.2859,
+      "step": 4461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.569029175337722e-05,
+      "loss": 0.252,
+      "step": 4462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.567299187558189e-05,
+      "loss": 0.3581,
+      "step": 4463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.565569130976422e-05,
+      "loss": 0.2937,
+      "step": 4464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5638390058022426e-05,
+      "loss": 0.272,
+      "step": 4465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.562108812245484e-05,
+      "loss": 0.3444,
+      "step": 4466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.56037855051598e-05,
+      "loss": 0.2154,
+      "step": 4467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.558648220823579e-05,
+      "loss": 0.267,
+      "step": 4468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.556917823378138e-05,
+      "loss": 0.2169,
+      "step": 4469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.555187358389519e-05,
+      "loss": 0.3239,
+      "step": 4470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5534568260675915e-05,
+      "loss": 0.3584,
+      "step": 4471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.551726226622238e-05,
+      "loss": 0.3336,
+      "step": 4472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5499955602633446e-05,
+      "loss": 0.3058,
+      "step": 4473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.548264827200808e-05,
+      "loss": 0.2809,
+      "step": 4474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.54653402764453e-05,
+      "loss": 0.2876,
+      "step": 4475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.544803161804425e-05,
+      "loss": 0.2546,
+      "step": 4476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5430722298904115e-05,
+      "loss": 0.3233,
+      "step": 4477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.541341232112417e-05,
+      "loss": 0.1907,
+      "step": 4478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.53961016868038e-05,
+      "loss": 0.265,
+      "step": 4479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5378790398042424e-05,
+      "loss": 0.2135,
+      "step": 4480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5361478456939555e-05,
+      "loss": 0.3333,
+      "step": 4481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.534416586559482e-05,
+      "loss": 0.2601,
+      "step": 4482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.532685262610787e-05,
+      "loss": 0.2708,
+      "step": 4483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5309538740578457e-05,
+      "loss": 0.3324,
+      "step": 4484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.529222421110643e-05,
+      "loss": 0.3106,
+      "step": 4485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.527490903979171e-05,
+      "loss": 0.3293,
+      "step": 4486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.525759322873427e-05,
+      "loss": 0.2238,
+      "step": 4487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.524027678003416e-05,
+      "loss": 0.2279,
+      "step": 4488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5222959695791566e-05,
+      "loss": 0.2164,
+      "step": 4489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.520564197810668e-05,
+      "loss": 0.3453,
+      "step": 4490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.518832362907982e-05,
+      "loss": 0.3169,
+      "step": 4491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.517100465081134e-05,
+      "loss": 0.1975,
+      "step": 4492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5153685045401704e-05,
+      "loss": 0.3141,
+      "step": 4493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.513636481495142e-05,
+      "loss": 0.2541,
+      "step": 4494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5119043961561136e-05,
+      "loss": 0.3514,
+      "step": 4495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.510172248733148e-05,
+      "loss": 0.2894,
+      "step": 4496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.508440039436326e-05,
+      "loss": 0.4401,
+      "step": 4497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.506707768475724e-05,
+      "loss": 0.2801,
+      "step": 4498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5049754360614394e-05,
+      "loss": 0.3236,
+      "step": 4499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5032430424035644e-05,
+      "loss": 0.2265,
+      "step": 4500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.501510587712205e-05,
+      "loss": 0.2692,
+      "step": 4501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4997780721974776e-05,
+      "loss": 0.3186,
+      "step": 4502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.498045496069499e-05,
+      "loss": 0.2529,
+      "step": 4503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.496312859538397e-05,
+      "loss": 0.3638,
+      "step": 4504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.494580162814308e-05,
+      "loss": 0.3354,
+      "step": 4505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.492847406107373e-05,
+      "loss": 0.2755,
+      "step": 4506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.49111458962774e-05,
+      "loss": 0.3184,
+      "step": 4507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.489381713585569e-05,
+      "loss": 0.2335,
+      "step": 4508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.48764877819102e-05,
+      "loss": 0.2154,
+      "step": 4509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4859157836542655e-05,
+      "loss": 0.2637,
+      "step": 4510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.484182730185483e-05,
+      "loss": 0.3494,
+      "step": 4511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.482449617994859e-05,
+      "loss": 0.2562,
+      "step": 4512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4807164472925865e-05,
+      "loss": 0.2384,
+      "step": 4513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.478983218288861e-05,
+      "loss": 0.2302,
+      "step": 4514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.477249931193896e-05,
+      "loss": 0.2289,
+      "step": 4515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.475516586217899e-05,
+      "loss": 0.2331,
+      "step": 4516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.473783183571093e-05,
+      "loss": 0.2505,
+      "step": 4517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.472049723463705e-05,
+      "loss": 0.2253,
+      "step": 4518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.470316206105971e-05,
+      "loss": 0.3658,
+      "step": 4519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.468582631708131e-05,
+      "loss": 0.2797,
+      "step": 4520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.466849000480433e-05,
+      "loss": 0.2111,
+      "step": 4521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4651153126331346e-05,
+      "loss": 0.2101,
+      "step": 4522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4633815683764975e-05,
+      "loss": 0.2632,
+      "step": 4523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.46164776792079e-05,
+      "loss": 0.2905,
+      "step": 4524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.459913911476288e-05,
+      "loss": 0.2613,
+      "step": 4525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.458179999253275e-05,
+      "loss": 0.2463,
+      "step": 4526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4564460314620366e-05,
+      "loss": 0.327,
+      "step": 4527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4547120083128755e-05,
+      "loss": 0.2713,
+      "step": 4528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.45297793001609e-05,
+      "loss": 0.3271,
+      "step": 4529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.451243796781992e-05,
+      "loss": 0.3181,
+      "step": 4530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.449509608820895e-05,
+      "loss": 0.2517,
+      "step": 4531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.447775366343125e-05,
+      "loss": 0.3245,
+      "step": 4532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.446041069559009e-05,
+      "loss": 0.3555,
+      "step": 4533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4443067186788854e-05,
+      "loss": 0.3374,
+      "step": 4534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4425723139130955e-05,
+      "loss": 0.2126,
+      "step": 4535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4408378554719884e-05,
+      "loss": 0.2832,
+      "step": 4536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.43910334356592e-05,
+      "loss": 0.2851,
+      "step": 4537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.437368778405253e-05,
+      "loss": 0.236,
+      "step": 4538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4356341602003546e-05,
+      "loss": 0.2724,
+      "step": 4539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4338994891616025e-05,
+      "loss": 0.285,
+      "step": 4540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.432164765499376e-05,
+      "loss": 0.2314,
+      "step": 4541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.430429989424062e-05,
+      "loss": 0.243,
+      "step": 4542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4286951611460565e-05,
+      "loss": 0.2257,
+      "step": 4543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.42696028087576e-05,
+      "loss": 0.1769,
+      "step": 4544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.42522534882358e-05,
+      "loss": 0.347,
+      "step": 4545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4234903651999294e-05,
+      "loss": 0.3438,
+      "step": 4546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4217553302152237e-05,
+      "loss": 0.2885,
+      "step": 4547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.420020244079895e-05,
+      "loss": 0.4236,
+      "step": 4548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4182851070043714e-05,
+      "loss": 0.272,
+      "step": 4549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.41654991919909e-05,
+      "loss": 0.302,
+      "step": 4550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.414814680874497e-05,
+      "loss": 0.2677,
+      "step": 4551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.413079392241042e-05,
+      "loss": 0.2432,
+      "step": 4552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.41134405350918e-05,
+      "loss": 0.4313,
+      "step": 4553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.409608664889376e-05,
+      "loss": 0.2347,
+      "step": 4554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.407873226592098e-05,
+      "loss": 0.3074,
+      "step": 4555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40613773882782e-05,
+      "loss": 0.3105,
+      "step": 4556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4044022018070214e-05,
+      "loss": 0.2027,
+      "step": 4557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4026666157401906e-05,
+      "loss": 0.2749,
+      "step": 4558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.400930980837818e-05,
+      "loss": 0.2957,
+      "step": 4559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.399195297310403e-05,
+      "loss": 0.3243,
+      "step": 4560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3974595653684514e-05,
+      "loss": 0.2532,
+      "step": 4561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3957237852224704e-05,
+      "loss": 0.248,
+      "step": 4562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.39398795708298e-05,
+      "loss": 0.2777,
+      "step": 4563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.392252081160498e-05,
+      "loss": 0.2699,
+      "step": 4564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.390516157665554e-05,
+      "loss": 0.2966,
+      "step": 4565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.388780186808678e-05,
+      "loss": 0.2612,
+      "step": 4566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3870441688004155e-05,
+      "loss": 0.2634,
+      "step": 4567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.385308103851307e-05,
+      "loss": 0.2813,
+      "step": 4568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.383571992171904e-05,
+      "loss": 0.2677,
+      "step": 4569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.381835833972761e-05,
+      "loss": 0.2564,
+      "step": 4570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.380099629464445e-05,
+      "loss": 0.2929,
+      "step": 4571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.378363378857517e-05,
+      "loss": 0.177,
+      "step": 4572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.376627082362554e-05,
+      "loss": 0.3043,
+      "step": 4573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.374890740190134e-05,
+      "loss": 0.2936,
+      "step": 4574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.373154352550841e-05,
+      "loss": 0.2825,
+      "step": 4575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.371417919655263e-05,
+      "loss": 0.2083,
+      "step": 4576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.369681441713997e-05,
+      "loss": 0.3071,
+      "step": 4577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.367944918937644e-05,
+      "loss": 0.4338,
+      "step": 4578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.366208351536809e-05,
+      "loss": 0.2729,
+      "step": 4579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3644717397221025e-05,
+      "loss": 0.2315,
+      "step": 4580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3627350837041446e-05,
+      "loss": 0.3156,
+      "step": 4581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.360998383693554e-05,
+      "loss": 0.2999,
+      "step": 4582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.35926163990096e-05,
+      "loss": 0.2898,
+      "step": 4583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3575248525369956e-05,
+      "loss": 0.2446,
+      "step": 4584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.355788021812298e-05,
+      "loss": 0.325,
+      "step": 4585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.354051147937511e-05,
+      "loss": 0.2561,
+      "step": 4586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.352314231123284e-05,
+      "loss": 0.2651,
+      "step": 4587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3505772715802704e-05,
+      "loss": 0.2366,
+      "step": 4588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.34884026951913e-05,
+      "loss": 0.2472,
+      "step": 4589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.347103225150525e-05,
+      "loss": 0.3188,
+      "step": 4590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.345366138685125e-05,
+      "loss": 0.3182,
+      "step": 4591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.343629010333606e-05,
+      "loss": 0.2839,
+      "step": 4592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.341891840306646e-05,
+      "loss": 0.2911,
+      "step": 4593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.340154628814932e-05,
+      "loss": 0.224,
+      "step": 4594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.33841737606915e-05,
+      "loss": 0.3293,
+      "step": 4595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.336680082279997e-05,
+      "loss": 0.2119,
+      "step": 4596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3349427476581734e-05,
+      "loss": 0.3201,
+      "step": 4597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.333205372414382e-05,
+      "loss": 0.3402,
+      "step": 4598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3314679567593304e-05,
+      "loss": 0.209,
+      "step": 4599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.329730500903737e-05,
+      "loss": 0.3031,
+      "step": 4600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.327993005058318e-05,
+      "loss": 0.267,
+      "step": 4601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.326255469433801e-05,
+      "loss": 0.2633,
+      "step": 4602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.324517894240909e-05,
+      "loss": 0.308,
+      "step": 4603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3227802796903814e-05,
+      "loss": 0.3022,
+      "step": 4604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3210426259929555e-05,
+      "loss": 0.2857,
+      "step": 4605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3193049333593705e-05,
+      "loss": 0.2358,
+      "step": 4606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.31756720200038e-05,
+      "loss": 0.2924,
+      "step": 4607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3158294321267324e-05,
+      "loss": 0.3063,
+      "step": 4608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3140916239491866e-05,
+      "loss": 0.2854,
+      "step": 4609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.312353777678505e-05,
+      "loss": 0.2971,
+      "step": 4610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3106158935254534e-05,
+      "loss": 0.2487,
+      "step": 4611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.308877971700804e-05,
+      "loss": 0.3512,
+      "step": 4612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.30714001241533e-05,
+      "loss": 0.3701,
+      "step": 4613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3054020158798164e-05,
+      "loss": 0.2434,
+      "step": 4614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3036639823050436e-05,
+      "loss": 0.2534,
+      "step": 4615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.301925911901802e-05,
+      "loss": 0.3515,
+      "step": 4616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.300187804880886e-05,
+      "loss": 0.2988,
+      "step": 4617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.298449661453093e-05,
+      "loss": 0.2673,
+      "step": 4618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.296711481829226e-05,
+      "loss": 0.3198,
+      "step": 4619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.294973266220093e-05,
+      "loss": 0.246,
+      "step": 4620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.293235014836504e-05,
+      "loss": 0.2531,
+      "step": 4621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.291496727889276e-05,
+      "loss": 0.355,
+      "step": 4622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.289758405589228e-05,
+      "loss": 0.3536,
+      "step": 4623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.288020048147184e-05,
+      "loss": 0.3114,
+      "step": 4624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2862816557739735e-05,
+      "loss": 0.2371,
+      "step": 4625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.284543228680427e-05,
+      "loss": 0.2805,
+      "step": 4626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.282804767077386e-05,
+      "loss": 0.2789,
+      "step": 4627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.281066271175689e-05,
+      "loss": 0.3114,
+      "step": 4628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.279327741186179e-05,
+      "loss": 0.2471,
+      "step": 4629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2775891773197105e-05,
+      "loss": 0.2699,
+      "step": 4630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.275850579787134e-05,
+      "loss": 0.2698,
+      "step": 4631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.274111948799306e-05,
+      "loss": 0.2752,
+      "step": 4632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2723732845670924e-05,
+      "loss": 0.2391,
+      "step": 4633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.270634587301355e-05,
+      "loss": 0.1913,
+      "step": 4634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2688958572129666e-05,
+      "loss": 0.3284,
+      "step": 4635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2671570945127965e-05,
+      "loss": 0.291,
+      "step": 4636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.265418299411728e-05,
+      "loss": 0.2346,
+      "step": 4637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2636794721206394e-05,
+      "loss": 0.3178,
+      "step": 4638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.261940612850418e-05,
+      "loss": 0.3116,
+      "step": 4639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.260201721811952e-05,
+      "loss": 0.2609,
+      "step": 4640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.258462799216134e-05,
+      "loss": 0.2839,
+      "step": 4641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2567238452738607e-05,
+      "loss": 0.3556,
+      "step": 4642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.254984860196037e-05,
+      "loss": 0.2774,
+      "step": 4643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2532458441935636e-05,
+      "loss": 0.2339,
+      "step": 4644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2515067974773516e-05,
+      "loss": 0.2631,
+      "step": 4645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.249767720258311e-05,
+      "loss": 0.2297,
+      "step": 4646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2480286127473574e-05,
+      "loss": 0.2946,
+      "step": 4647
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.246289475155414e-05,
+      "loss": 0.2426,
+      "step": 4648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2445503076933976e-05,
+      "loss": 0.2522,
+      "step": 4649
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.242811110572242e-05,
+      "loss": 0.3119,
+      "step": 4650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2410718840028736e-05,
+      "loss": 0.2712,
+      "step": 4651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.239332628196226e-05,
+      "loss": 0.2687,
+      "step": 4652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.237593343363238e-05,
+      "loss": 0.2452,
+      "step": 4653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.235854029714853e-05,
+      "loss": 0.4298,
+      "step": 4654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.234114687462012e-05,
+      "loss": 0.2216,
+      "step": 4655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.232375316815663e-05,
+      "loss": 0.2159,
+      "step": 4656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2306359179867594e-05,
+      "loss": 0.2496,
+      "step": 4657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.228896491186255e-05,
+      "loss": 0.2594,
+      "step": 4658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.227157036625108e-05,
+      "loss": 0.2665,
+      "step": 4659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2254175545142805e-05,
+      "loss": 0.3172,
+      "step": 4660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.223678045064737e-05,
+      "loss": 0.2443,
+      "step": 4661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.221938508487446e-05,
+      "loss": 0.2109,
+      "step": 4662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.220198944993381e-05,
+      "loss": 0.2632,
+      "step": 4663
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.218459354793513e-05,
+      "loss": 0.2491,
+      "step": 4664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.216719738098822e-05,
+      "loss": 0.2475,
+      "step": 4665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2149800951202895e-05,
+      "loss": 0.2692,
+      "step": 4666
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2132404260689006e-05,
+      "loss": 0.2489,
+      "step": 4667
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.211500731155642e-05,
+      "loss": 0.2327,
+      "step": 4668
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.209761010591503e-05,
+      "loss": 0.2598,
+      "step": 4669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.20802126458748e-05,
+      "loss": 0.2524,
+      "step": 4670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.20628149335457e-05,
+      "loss": 0.2825,
+      "step": 4671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.204541697103772e-05,
+      "loss": 0.2254,
+      "step": 4672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.202801876046088e-05,
+      "loss": 0.2647,
+      "step": 4673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2010620303925275e-05,
+      "loss": 0.2403,
+      "step": 4674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.199322160354094e-05,
+      "loss": 0.2453,
+      "step": 4675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.197582266141805e-05,
+      "loss": 0.2897,
+      "step": 4676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.195842347966673e-05,
+      "loss": 0.312,
+      "step": 4677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.194102406039716e-05,
+      "loss": 0.2246,
+      "step": 4678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.192362440571955e-05,
+      "loss": 0.347,
+      "step": 4679
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.190622451774414e-05,
+      "loss": 0.3388,
+      "step": 4680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.188882439858117e-05,
+      "loss": 0.2199,
+      "step": 4681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1871424050340955e-05,
+      "loss": 0.2939,
+      "step": 4682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1854023475133826e-05,
+      "loss": 0.3065,
+      "step": 4683
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.183662267507011e-05,
+      "loss": 0.3355,
+      "step": 4684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181922165226017e-05,
+      "loss": 0.2567,
+      "step": 4685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.180182040881444e-05,
+      "loss": 0.3517,
+      "step": 4686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.178441894684335e-05,
+      "loss": 0.2229,
+      "step": 4687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.176701726845731e-05,
+      "loss": 0.2685,
+      "step": 4688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1749615375766846e-05,
+      "loss": 0.2699,
+      "step": 4689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1732213270882446e-05,
+      "loss": 0.2919,
+      "step": 4690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.171481095591463e-05,
+      "loss": 0.3234,
+      "step": 4691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1697408432973996e-05,
+      "loss": 0.2195,
+      "step": 4692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.16800057041711e-05,
+      "loss": 0.2971,
+      "step": 4693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.166260277161654e-05,
+      "loss": 0.3229,
+      "step": 4694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.164519963742097e-05,
+      "loss": 0.2185,
+      "step": 4695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.162779630369507e-05,
+      "loss": 0.2746,
+      "step": 4696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1610392772549465e-05,
+      "loss": 0.2932,
+      "step": 4697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.15929890460949e-05,
+      "loss": 0.2604,
+      "step": 4698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.15755851264421e-05,
+      "loss": 0.1943,
+      "step": 4699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.15581810157018e-05,
+      "loss": 0.2535,
+      "step": 4700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.154077671598479e-05,
+      "loss": 0.2812,
+      "step": 4701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.152337222940188e-05,
+      "loss": 0.3618,
+      "step": 4702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.150596755806387e-05,
+      "loss": 0.3118,
+      "step": 4703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.148856270408163e-05,
+      "loss": 0.3032,
+      "step": 4704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1471157669566004e-05,
+      "loss": 0.2629,
+      "step": 4705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.145375245662789e-05,
+      "loss": 0.2552,
+      "step": 4706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.14363470673782e-05,
+      "loss": 0.2892,
+      "step": 4707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1418941503927855e-05,
+      "loss": 0.2455,
+      "step": 4708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.140153576838781e-05,
+      "loss": 0.2004,
+      "step": 4709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.138412986286905e-05,
+      "loss": 0.2241,
+      "step": 4710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1366723789482554e-05,
+      "loss": 0.2385,
+      "step": 4711
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.134931755033936e-05,
+      "loss": 0.2993,
+      "step": 4712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.133191114755047e-05,
+      "loss": 0.2957,
+      "step": 4713
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.131450458322695e-05,
+      "loss": 0.2625,
+      "step": 4714
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.129709785947989e-05,
+      "loss": 0.1736,
+      "step": 4715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1279690978420384e-05,
+      "loss": 0.2733,
+      "step": 4716
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.126228394215952e-05,
+      "loss": 0.3636,
+      "step": 4717
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.124487675280842e-05,
+      "loss": 0.2978,
+      "step": 4718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.122746941247828e-05,
+      "loss": 0.2735,
+      "step": 4719
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.121006192328025e-05,
+      "loss": 0.2774,
+      "step": 4720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1192654287325484e-05,
+      "loss": 0.1915,
+      "step": 4721
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1175246506725226e-05,
+      "loss": 0.3099,
+      "step": 4722
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1157838583590675e-05,
+      "loss": 0.2559,
+      "step": 4723
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114043052003308e-05,
+      "loss": 0.3108,
+      "step": 4724
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.112302231816369e-05,
+      "loss": 0.2441,
+      "step": 4725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.110561398009378e-05,
+      "loss": 0.289,
+      "step": 4726
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.108820550793465e-05,
+      "loss": 0.2684,
+      "step": 4727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1070796903797593e-05,
+      "loss": 0.253,
+      "step": 4728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.105338816979392e-05,
+      "loss": 0.2995,
+      "step": 4729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1035979308034997e-05,
+      "loss": 0.2374,
+      "step": 4730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1018570320632156e-05,
+      "loss": 0.3144,
+      "step": 4731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.100116120969676e-05,
+      "loss": 0.3243,
+      "step": 4732
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.098375197734021e-05,
+      "loss": 0.3704,
+      "step": 4733
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.096634262567388e-05,
+      "loss": 0.3134,
+      "step": 4734
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.094893315680921e-05,
+      "loss": 0.2722,
+      "step": 4735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0931523572857606e-05,
+      "loss": 0.2075,
+      "step": 4736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0914113875930516e-05,
+      "loss": 0.3171,
+      "step": 4737
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.08967040681394e-05,
+      "loss": 0.2418,
+      "step": 4738
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.087929415159571e-05,
+      "loss": 0.3161,
+      "step": 4739
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.086188412841093e-05,
+      "loss": 0.2761,
+      "step": 4740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.084447400069655e-05,
+      "loss": 0.2478,
+      "step": 4741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.08270637705641e-05,
+      "loss": 0.264,
+      "step": 4742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.080965344012508e-05,
+      "loss": 0.2379,
+      "step": 4743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0792243011491014e-05,
+      "loss": 0.3428,
+      "step": 4744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.077483248677347e-05,
+      "loss": 0.25,
+      "step": 4745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0757421868083966e-05,
+      "loss": 0.328,
+      "step": 4746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.074001115753408e-05,
+      "loss": 0.2265,
+      "step": 4747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0722600357235415e-05,
+      "loss": 0.2694,
+      "step": 4748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.070518946929954e-05,
+      "loss": 0.2609,
+      "step": 4749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.068777849583804e-05,
+      "loss": 0.2197,
+      "step": 4750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.067036743896254e-05,
+      "loss": 0.3259,
+      "step": 4751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.065295630078466e-05,
+      "loss": 0.3524,
+      "step": 4752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0635545083416025e-05,
+      "loss": 0.2291,
+      "step": 4753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0618133788968255e-05,
+      "loss": 0.2396,
+      "step": 4754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0600722419553035e-05,
+      "loss": 0.2762,
+      "step": 4755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0583310977282e-05,
+      "loss": 0.2376,
+      "step": 4756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.056589946426679e-05,
+      "loss": 0.3318,
+      "step": 4757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.054848788261912e-05,
+      "loss": 0.2051,
+      "step": 4758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.053107623445067e-05,
+      "loss": 0.2888,
+      "step": 4759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0513664521873105e-05,
+      "loss": 0.2231,
+      "step": 4760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.049625274699814e-05,
+      "loss": 0.3117,
+      "step": 4761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.047884091193749e-05,
+      "loss": 0.2504,
+      "step": 4762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.046142901880284e-05,
+      "loss": 0.2014,
+      "step": 4763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.044401706970592e-05,
+      "loss": 0.294,
+      "step": 4764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.042660506675847e-05,
+      "loss": 0.2984,
+      "step": 4765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.040919301207222e-05,
+      "loss": 0.2692,
+      "step": 4766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.039178090775889e-05,
+      "loss": 0.3141,
+      "step": 4767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.037436875593026e-05,
+      "loss": 0.3208,
+      "step": 4768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.035695655869808e-05,
+      "loss": 0.344,
+      "step": 4769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.033954431817405e-05,
+      "loss": 0.2273,
+      "step": 4770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.032213203646999e-05,
+      "loss": 0.2722,
+      "step": 4771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0304719715697655e-05,
+      "loss": 0.237,
+      "step": 4772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.028730735796882e-05,
+      "loss": 0.2928,
+      "step": 4773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0269894965395225e-05,
+      "loss": 0.2622,
+      "step": 4774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0252482540088696e-05,
+      "loss": 0.2887,
+      "step": 4775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0235070084161e-05,
+      "loss": 0.3165,
+      "step": 4776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.021765759972392e-05,
+      "loss": 0.2645,
+      "step": 4777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0200245088889275e-05,
+      "loss": 0.2522,
+      "step": 4778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.018283255376882e-05,
+      "loss": 0.3219,
+      "step": 4779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.016541999647437e-05,
+      "loss": 0.4352,
+      "step": 4780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.014800741911774e-05,
+      "loss": 0.2821,
+      "step": 4781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.013059482381072e-05,
+      "loss": 0.3298,
+      "step": 4782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.011318221266511e-05,
+      "loss": 0.2892,
+      "step": 4783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.009576958779272e-05,
+      "loss": 0.2502,
+      "step": 4784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0078356951305364e-05,
+      "loss": 0.2388,
+      "step": 4785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.006094430531486e-05,
+      "loss": 0.3161,
+      "step": 4786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0043531651933003e-05,
+      "loss": 0.3317,
+      "step": 4787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.002611899327162e-05,
+      "loss": 0.2151,
+      "step": 4788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.000870633144251e-05,
+      "loss": 0.1821,
+      "step": 4789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.999129366855749e-05,
+      "loss": 0.4209,
+      "step": 4790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.997388100672839e-05,
+      "loss": 0.2691,
+      "step": 4791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9956468348067e-05,
+      "loss": 0.2022,
+      "step": 4792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.993905569468515e-05,
+      "loss": 0.2046,
+      "step": 4793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9921643048694634e-05,
+      "loss": 0.2568,
+      "step": 4794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.990423041220729e-05,
+      "loss": 0.31,
+      "step": 4795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.98868177873349e-05,
+      "loss": 0.2602,
+      "step": 4796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9869405176189285e-05,
+      "loss": 0.277,
+      "step": 4797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.985199258088228e-05,
+      "loss": 0.2427,
+      "step": 4798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.983458000352565e-05,
+      "loss": 0.2238,
+      "step": 4799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.98171674462312e-05,
+      "loss": 0.275,
+      "step": 4800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.979975491111075e-05,
+      "loss": 0.2202,
+      "step": 4801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.978234240027609e-05,
+      "loss": 0.2362,
+      "step": 4802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.976492991583901e-05,
+      "loss": 0.2586,
+      "step": 4803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974751745991132e-05,
+      "loss": 0.3025,
+      "step": 4804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.973010503460479e-05,
+      "loss": 0.2662,
+      "step": 4805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.97126926420312e-05,
+      "loss": 0.3171,
+      "step": 4806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.969528028430235e-05,
+      "loss": 0.285,
+      "step": 4807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9677867963530014e-05,
+      "loss": 0.2973,
+      "step": 4808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9660455681825954e-05,
+      "loss": 0.1986,
+      "step": 4809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.964304344130194e-05,
+      "loss": 0.3114,
+      "step": 4810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.962563124406974e-05,
+      "loss": 0.2424,
+      "step": 4811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9608219092241104e-05,
+      "loss": 0.2552,
+      "step": 4812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.959080698792779e-05,
+      "loss": 0.3642,
+      "step": 4813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.957339493324153e-05,
+      "loss": 0.202,
+      "step": 4814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9555982930294106e-05,
+      "loss": 0.2852,
+      "step": 4815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9538570981197184e-05,
+      "loss": 0.2154,
+      "step": 4816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.952115908806254e-05,
+      "loss": 0.223,
+      "step": 4817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.950374725300188e-05,
+      "loss": 0.2433,
+      "step": 4818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.948633547812691e-05,
+      "loss": 0.2536,
+      "step": 4819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9468923765549346e-05,
+      "loss": 0.2585,
+      "step": 4820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.945151211738089e-05,
+      "loss": 0.2357,
+      "step": 4821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.943410053573322e-05,
+      "loss": 0.234,
+      "step": 4822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.941668902271802e-05,
+      "loss": 0.2899,
+      "step": 4823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9399277580446976e-05,
+      "loss": 0.2575,
+      "step": 4824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.938186621103175e-05,
+      "loss": 0.2141,
+      "step": 4825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.936445491658398e-05,
+      "loss": 0.3649,
+      "step": 4826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.934704369921534e-05,
+      "loss": 0.2363,
+      "step": 4827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.932963256103746e-05,
+      "loss": 0.3236,
+      "step": 4828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.931222150416197e-05,
+      "loss": 0.2255,
+      "step": 4829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929481053070046e-05,
+      "loss": 0.2629,
+      "step": 4830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9277399642764603e-05,
+      "loss": 0.2358,
+      "step": 4831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.925998884246592e-05,
+      "loss": 0.2661,
+      "step": 4832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.924257813191605e-05,
+      "loss": 0.2076,
+      "step": 4833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9225167513226554e-05,
+      "loss": 0.3192,
+      "step": 4834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9207756988509004e-05,
+      "loss": 0.2782,
+      "step": 4835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.919034655987493e-05,
+      "loss": 0.2617,
+      "step": 4836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9172936229435906e-05,
+      "loss": 0.2762,
+      "step": 4837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.915552599930345e-05,
+      "loss": 0.3383,
+      "step": 4838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9138115871589073e-05,
+      "loss": 0.2976,
+      "step": 4839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.91207058484043e-05,
+      "loss": 0.3078,
+      "step": 4840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9103295931860614e-05,
+      "loss": 0.2375,
+      "step": 4841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.908588612406949e-05,
+      "loss": 0.2216,
+      "step": 4842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.90684764271424e-05,
+      "loss": 0.281,
+      "step": 4843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9051066843190795e-05,
+      "loss": 0.2782,
+      "step": 4844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.903365737432612e-05,
+      "loss": 0.2582,
+      "step": 4845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9016248022659796e-05,
+      "loss": 0.2958,
+      "step": 4846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.899883879030326e-05,
+      "loss": 0.2049,
+      "step": 4847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8981429679367875e-05,
+      "loss": 0.273,
+      "step": 4848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8964020691965015e-05,
+      "loss": 0.2859,
+      "step": 4849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.894661183020609e-05,
+      "loss": 0.2694,
+      "step": 4850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.892920309620243e-05,
+      "loss": 0.3496,
+      "step": 4851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.891179449206536e-05,
+      "loss": 0.2831,
+      "step": 4852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.889438601990622e-05,
+      "loss": 0.2153,
+      "step": 4853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8876977681836325e-05,
+      "loss": 0.3007,
+      "step": 4854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.885956947996694e-05,
+      "loss": 0.2249,
+      "step": 4855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.884216141640933e-05,
+      "loss": 0.3256,
+      "step": 4856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8824753493274785e-05,
+      "loss": 0.2527,
+      "step": 4857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.880734571267452e-05,
+      "loss": 0.208,
+      "step": 4858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.878993807671976e-05,
+      "loss": 0.252,
+      "step": 4859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.877253058752172e-05,
+      "loss": 0.2697,
+      "step": 4860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8755123247191576e-05,
+      "loss": 0.2847,
+      "step": 4861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87377160578405e-05,
+      "loss": 0.3819,
+      "step": 4862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.872030902157962e-05,
+      "loss": 0.2796,
+      "step": 4863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8702902140520126e-05,
+      "loss": 0.2384,
+      "step": 4864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8685495416773054e-05,
+      "loss": 0.3346,
+      "step": 4865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.866808885244954e-05,
+      "loss": 0.288,
+      "step": 4866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.865068244966066e-05,
+      "loss": 0.307,
+      "step": 4867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.863327621051746e-05,
+      "loss": 0.2636,
+      "step": 4868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8615870137130956e-05,
+      "loss": 0.3283,
+      "step": 4869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.85984642316122e-05,
+      "loss": 0.2204,
+      "step": 4870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8581058496072164e-05,
+      "loss": 0.2163,
+      "step": 4871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.856365293262181e-05,
+      "loss": 0.2644,
+      "step": 4872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.854624754337212e-05,
+      "loss": 0.2294,
+      "step": 4873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.852884233043401e-05,
+      "loss": 0.1855,
+      "step": 4874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8511437295918386e-05,
+      "loss": 0.2513,
+      "step": 4875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849403244193613e-05,
+      "loss": 0.292,
+      "step": 4876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.847662777059813e-05,
+      "loss": 0.2202,
+      "step": 4877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.845922328401522e-05,
+      "loss": 0.2045,
+      "step": 4878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8441818984298204e-05,
+      "loss": 0.2935,
+      "step": 4879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8424414873557935e-05,
+      "loss": 0.2295,
+      "step": 4880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.840701095390512e-05,
+      "loss": 0.3493,
+      "step": 4881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.838960722745055e-05,
+      "loss": 0.1874,
+      "step": 4882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.837220369630496e-05,
+      "loss": 0.1983,
+      "step": 4883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.835480036257904e-05,
+      "loss": 0.2468,
+      "step": 4884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.833739722838346e-05,
+      "loss": 0.2998,
+      "step": 4885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.831999429582892e-05,
+      "loss": 0.3134,
+      "step": 4886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8302591567026016e-05,
+      "loss": 0.2343,
+      "step": 4887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8285189044085374e-05,
+      "loss": 0.2841,
+      "step": 4888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.826778672911757e-05,
+      "loss": 0.2367,
+      "step": 4889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8250384624233166e-05,
+      "loss": 0.2275,
+      "step": 4890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.82329827315427e-05,
+      "loss": 0.1931,
+      "step": 4891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821558105315666e-05,
+      "loss": 0.3199,
+      "step": 4892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.819817959118555e-05,
+      "loss": 0.3016,
+      "step": 4893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.818077834773983e-05,
+      "loss": 0.2225,
+      "step": 4894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8163377324929906e-05,
+      "loss": 0.3327,
+      "step": 4895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.814597652486618e-05,
+      "loss": 0.2477,
+      "step": 4896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.812857594965906e-05,
+      "loss": 0.2909,
+      "step": 4897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8111175601418844e-05,
+      "loss": 0.2529,
+      "step": 4898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8093775482255886e-05,
+      "loss": 0.2414,
+      "step": 4899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.807637559428047e-05,
+      "loss": 0.2318,
+      "step": 4900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8058975939602854e-05,
+      "loss": 0.29,
+      "step": 4901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.804157652033328e-05,
+      "loss": 0.2919,
+      "step": 4902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.802417733858196e-05,
+      "loss": 0.2548,
+      "step": 4903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8006778396459066e-05,
+      "loss": 0.2577,
+      "step": 4904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.798937969607474e-05,
+      "loss": 0.3473,
+      "step": 4905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7971981239539124e-05,
+      "loss": 0.303,
+      "step": 4906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.795458302896229e-05,
+      "loss": 0.2745,
+      "step": 4907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.793718506645431e-05,
+      "loss": 0.2494,
+      "step": 4908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7919787354125196e-05,
+      "loss": 0.2576,
+      "step": 4909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.790238989408497e-05,
+      "loss": 0.2681,
+      "step": 4910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.788499268844359e-05,
+      "loss": 0.3071,
+      "step": 4911
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.786759573931099e-05,
+      "loss": 0.2527,
+      "step": 4912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.785019904879712e-05,
+      "loss": 0.2749,
+      "step": 4913
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.783280261901179e-05,
+      "loss": 0.3141,
+      "step": 4914
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.781540645206489e-05,
+      "loss": 0.2761,
+      "step": 4915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.779801055006622e-05,
+      "loss": 0.2748,
+      "step": 4916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.778061491512555e-05,
+      "loss": 0.3051,
+      "step": 4917
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.776321954935264e-05,
+      "loss": 0.2767,
+      "step": 4918
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7745824454857206e-05,
+      "loss": 0.2241,
+      "step": 4919
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7728429633748934e-05,
+      "loss": 0.318,
+      "step": 4920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.771103508813746e-05,
+      "loss": 0.2051,
+      "step": 4921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.769364082013242e-05,
+      "loss": 0.2692,
+      "step": 4922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.767624683184338e-05,
+      "loss": 0.3142,
+      "step": 4923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.76588531253799e-05,
+      "loss": 0.2519,
+      "step": 4924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7641459702851476e-05,
+      "loss": 0.2263,
+      "step": 4925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.762406656636761e-05,
+      "loss": 0.2524,
+      "step": 4926
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.760667371803774e-05,
+      "loss": 0.2663,
+      "step": 4927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7589281159971276e-05,
+      "loss": 0.2414,
+      "step": 4928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7571888894277604e-05,
+      "loss": 0.2085,
+      "step": 4929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7554496923066036e-05,
+      "loss": 0.2784,
+      "step": 4930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.753710524844589e-05,
+      "loss": 0.2478,
+      "step": 4931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.751971387252644e-05,
+      "loss": 0.3248,
+      "step": 4932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7502322797416916e-05,
+      "loss": 0.3175,
+      "step": 4933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7484932025226495e-05,
+      "loss": 0.2748,
+      "step": 4934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.746754155806437e-05,
+      "loss": 0.2913,
+      "step": 4935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.745015139803964e-05,
+      "loss": 0.2466,
+      "step": 4936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74327615472614e-05,
+      "loss": 0.1915,
+      "step": 4937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741537200783867e-05,
+      "loss": 0.2573,
+      "step": 4938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7397982781880494e-05,
+      "loss": 0.2401,
+      "step": 4939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7380593871495834e-05,
+      "loss": 0.2618,
+      "step": 4940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.736320527879361e-05,
+      "loss": 0.2105,
+      "step": 4941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.734581700588272e-05,
+      "loss": 0.242,
+      "step": 4942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.732842905487203e-05,
+      "loss": 0.2486,
+      "step": 4943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.731104142787035e-05,
+      "loss": 0.2027,
+      "step": 4944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7293654126986446e-05,
+      "loss": 0.2602,
+      "step": 4945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.72762671543291e-05,
+      "loss": 0.2271,
+      "step": 4946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.725888051200695e-05,
+      "loss": 0.3111,
+      "step": 4947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.724149420212868e-05,
+      "loss": 0.2834,
+      "step": 4948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7224108226802913e-05,
+      "loss": 0.3087,
+      "step": 4949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.720672258813822e-05,
+      "loss": 0.237,
+      "step": 4950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.718933728824313e-05,
+      "loss": 0.2687,
+      "step": 4951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.717195232922615e-05,
+      "loss": 0.2987,
+      "step": 4952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.715456771319573e-05,
+      "loss": 0.3035,
+      "step": 4953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7137183442260276e-05,
+      "loss": 0.2837,
+      "step": 4954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7119799518528166e-05,
+      "loss": 0.3725,
+      "step": 4955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.710241594410773e-05,
+      "loss": 0.3364,
+      "step": 4956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7085032721107244e-05,
+      "loss": 0.289,
+      "step": 4957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7067649851634956e-05,
+      "loss": 0.3438,
+      "step": 4958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.705026733779907e-05,
+      "loss": 0.2319,
+      "step": 4959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.703288518170774e-05,
+      "loss": 0.2503,
+      "step": 4960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.701550338546908e-05,
+      "loss": 0.243,
+      "step": 4961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.699812195119116e-05,
+      "loss": 0.2346,
+      "step": 4962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6980740880982006e-05,
+      "loss": 0.2952,
+      "step": 4963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.696336017694958e-05,
+      "loss": 0.2751,
+      "step": 4964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6945979841201854e-05,
+      "loss": 0.2315,
+      "step": 4965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6928599875846705e-05,
+      "loss": 0.2135,
+      "step": 4966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.691122028299197e-05,
+      "loss": 0.2853,
+      "step": 4967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.689384106474548e-05,
+      "loss": 0.3255,
+      "step": 4968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.687646222321496e-05,
+      "loss": 0.2879,
+      "step": 4969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6859083760508146e-05,
+      "loss": 0.3524,
+      "step": 4970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.684170567873268e-05,
+      "loss": 0.2586,
+      "step": 4971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.682432797999622e-05,
+      "loss": 0.3219,
+      "step": 4972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.68069506664063e-05,
+      "loss": 0.2736,
+      "step": 4973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6789573740070456e-05,
+      "loss": 0.2543,
+      "step": 4974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6772197203096184e-05,
+      "loss": 0.2258,
+      "step": 4975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.675482105759091e-05,
+      "loss": 0.3665,
+      "step": 4976
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.673744530566201e-05,
+      "loss": 0.2149,
+      "step": 4977
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.672006994941681e-05,
+      "loss": 0.3005,
+      "step": 4978
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6702694990962655e-05,
+      "loss": 0.3626,
+      "step": 4979
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.66853204324067e-05,
+      "loss": 0.2959,
+      "step": 4980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.666794627585621e-05,
+      "loss": 0.2489,
+      "step": 4981
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6650572523418284e-05,
+      "loss": 0.3853,
+      "step": 4982
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663319917720004e-05,
+      "loss": 0.2772,
+      "step": 4983
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6615826239308505e-05,
+      "loss": 0.2216,
+      "step": 4984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.659845371185069e-05,
+      "loss": 0.2634,
+      "step": 4985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.658108159693355e-05,
+      "loss": 0.2117,
+      "step": 4986
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.656370989666394e-05,
+      "loss": 0.2099,
+      "step": 4987
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6546338613148755e-05,
+      "loss": 0.2446,
+      "step": 4988
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.652896774849477e-05,
+      "loss": 0.3132,
+      "step": 4989
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.651159730480872e-05,
+      "loss": 0.2283,
+      "step": 4990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6494227284197294e-05,
+      "loss": 0.1729,
+      "step": 4991
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6476857688767156e-05,
+      "loss": 0.2307,
+      "step": 4992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.645948852062489e-05,
+      "loss": 0.2439,
+      "step": 4993
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.644211978187701e-05,
+      "loss": 0.2801,
+      "step": 4994
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.642475147463007e-05,
+      "loss": 0.2166,
+      "step": 4995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6407383600990414e-05,
+      "loss": 0.2269,
+      "step": 4996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.639001616306447e-05,
+      "loss": 0.2432,
+      "step": 4997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.637264916295857e-05,
+      "loss": 0.1702,
+      "step": 4998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.635528260277899e-05,
+      "loss": 0.2535,
+      "step": 4999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6337916484631926e-05,
+      "loss": 0.282,
+      "step": 5000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6320550810623576e-05,
+      "loss": 0.2737,
+      "step": 5001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.630318558286004e-05,
+      "loss": 0.2611,
+      "step": 5002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.628582080344739e-05,
+      "loss": 0.3188,
+      "step": 5003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6268456474491606e-05,
+      "loss": 0.2135,
+      "step": 5004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6251092598098665e-05,
+      "loss": 0.2971,
+      "step": 5005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.623372917637447e-05,
+      "loss": 0.2253,
+      "step": 5006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6216366211424825e-05,
+      "loss": 0.3579,
+      "step": 5007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.619900370535556e-05,
+      "loss": 0.2506,
+      "step": 5008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.618164166027238e-05,
+      "loss": 0.3174,
+      "step": 5009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616428007828097e-05,
+      "loss": 0.3106,
+      "step": 5010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614691896148695e-05,
+      "loss": 0.2645,
+      "step": 5011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.612955831199587e-05,
+      "loss": 0.2246,
+      "step": 5012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.611219813191322e-05,
+      "loss": 0.2706,
+      "step": 5013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609483842334449e-05,
+      "loss": 0.2958,
+      "step": 5014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6077479188395045e-05,
+      "loss": 0.3311,
+      "step": 5015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.606012042917022e-05,
+      "loss": 0.2787,
+      "step": 5016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.60427621477753e-05,
+      "loss": 0.2829,
+      "step": 5017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.60254043463155e-05,
+      "loss": 0.3065,
+      "step": 5018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.600804702689598e-05,
+      "loss": 0.2102,
+      "step": 5019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.599069019162183e-05,
+      "loss": 0.2089,
+      "step": 5020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.597333384259811e-05,
+      "loss": 0.2333,
+      "step": 5021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.59559779819298e-05,
+      "loss": 0.236,
+      "step": 5022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5938622611721815e-05,
+      "loss": 0.2809,
+      "step": 5023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5921267734079023e-05,
+      "loss": 0.2968,
+      "step": 5024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.590391335110624e-05,
+      "loss": 0.3244,
+      "step": 5025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.58865594649082e-05,
+      "loss": 0.2462,
+      "step": 5026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.586920607758958e-05,
+      "loss": 0.3178,
+      "step": 5027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.585185319125506e-05,
+      "loss": 0.2681,
+      "step": 5028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.583450080800912e-05,
+      "loss": 0.2823,
+      "step": 5029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.581714892995631e-05,
+      "loss": 0.232,
+      "step": 5030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.579979755920107e-05,
+      "loss": 0.2456,
+      "step": 5031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.578244669784777e-05,
+      "loss": 0.2803,
+      "step": 5032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5765096348000724e-05,
+      "loss": 0.2755,
+      "step": 5033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.57477465117642e-05,
+      "loss": 0.2934,
+      "step": 5034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5730397191242405e-05,
+      "loss": 0.3315,
+      "step": 5035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5713048388539446e-05,
+      "loss": 0.2877,
+      "step": 5036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5695700105759385e-05,
+      "loss": 0.2644,
+      "step": 5037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.567835234500626e-05,
+      "loss": 0.2497,
+      "step": 5038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5661005108383994e-05,
+      "loss": 0.2385,
+      "step": 5039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5643658397996445e-05,
+      "loss": 0.2827,
+      "step": 5040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.562631221594747e-05,
+      "loss": 0.2149,
+      "step": 5041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.560896656434081e-05,
+      "loss": 0.2871,
+      "step": 5042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.559162144528012e-05,
+      "loss": 0.2254,
+      "step": 5043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5574276860869064e-05,
+      "loss": 0.2741,
+      "step": 5044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5556932813211164e-05,
+      "loss": 0.2413,
+      "step": 5045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5539589304409916e-05,
+      "loss": 0.2555,
+      "step": 5046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.552224633656876e-05,
+      "loss": 0.2789,
+      "step": 5047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.550490391179106e-05,
+      "loss": 0.2119,
+      "step": 5048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.54875620321801e-05,
+      "loss": 0.2459,
+      "step": 5049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.547022069983911e-05,
+      "loss": 0.2351,
+      "step": 5050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.545287991687126e-05,
+      "loss": 0.2394,
+      "step": 5051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.543553968537964e-05,
+      "loss": 0.2042,
+      "step": 5052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.541820000746727e-05,
+      "loss": 0.3055,
+      "step": 5053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.540086088523713e-05,
+      "loss": 0.2625,
+      "step": 5054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5383522320792116e-05,
+      "loss": 0.2774,
+      "step": 5055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.536618431623504e-05,
+      "loss": 0.2923,
+      "step": 5056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.534884687366865e-05,
+      "loss": 0.2176,
+      "step": 5057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.533150999519567e-05,
+      "loss": 0.264,
+      "step": 5058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.53141736829187e-05,
+      "loss": 0.2829,
+      "step": 5059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5296837938940295e-05,
+      "loss": 0.2786,
+      "step": 5060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.527950276536298e-05,
+      "loss": 0.2413,
+      "step": 5061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.526216816428909e-05,
+      "loss": 0.2816,
+      "step": 5062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5244834137821026e-05,
+      "loss": 0.3374,
+      "step": 5063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5227500688061065e-05,
+      "loss": 0.2396,
+      "step": 5064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.521016781711139e-05,
+      "loss": 0.2046,
+      "step": 5065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5192835527074154e-05,
+      "loss": 0.2855,
+      "step": 5066
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.517550382005142e-05,
+      "loss": 0.3049,
+      "step": 5067
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.515817269814518e-05,
+      "loss": 0.2699,
+      "step": 5068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.514084216345736e-05,
+      "loss": 0.255,
+      "step": 5069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.512351221808982e-05,
+      "loss": 0.2723,
+      "step": 5070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.510618286414432e-05,
+      "loss": 0.3174,
+      "step": 5071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.508885410372261e-05,
+      "loss": 0.2752,
+      "step": 5072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5071525938926276e-05,
+      "loss": 0.2802,
+      "step": 5073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.505419837185692e-05,
+      "loss": 0.2435,
+      "step": 5074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.503687140461603e-05,
+      "loss": 0.3409,
+      "step": 5075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5019545039305013e-05,
+      "loss": 0.2729,
+      "step": 5076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.500221927802524e-05,
+      "loss": 0.207,
+      "step": 5077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4984894122877966e-05,
+      "loss": 0.4158,
+      "step": 5078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.496756957596438e-05,
+      "loss": 0.2368,
+      "step": 5079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.495024563938564e-05,
+      "loss": 0.2773,
+      "step": 5080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.493292231524277e-05,
+      "loss": 0.2374,
+      "step": 5081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4915599605636755e-05,
+      "loss": 0.2959,
+      "step": 5082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.489827751266852e-05,
+      "loss": 0.2362,
+      "step": 5083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4880956038438876e-05,
+      "loss": 0.3576,
+      "step": 5084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.486363518504859e-05,
+      "loss": 0.2238,
+      "step": 5085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.484631495459831e-05,
+      "loss": 0.1814,
+      "step": 5086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4828995349188675e-05,
+      "loss": 0.2962,
+      "step": 5087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.48116763709202e-05,
+      "loss": 0.2604,
+      "step": 5088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.479435802189332e-05,
+      "loss": 0.2298,
+      "step": 5089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477704030420843e-05,
+      "loss": 0.2634,
+      "step": 5090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.475972321996584e-05,
+      "loss": 0.2085,
+      "step": 5091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4742406771265745e-05,
+      "loss": 0.2827,
+      "step": 5092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.472509096020831e-05,
+      "loss": 0.2913,
+      "step": 5093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.470777578889358e-05,
+      "loss": 0.2378,
+      "step": 5094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.469046125942155e-05,
+      "loss": 0.2711,
+      "step": 5095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.467314737389215e-05,
+      "loss": 0.3066,
+      "step": 5096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4655834134405196e-05,
+      "loss": 0.269,
+      "step": 5097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.463852154306045e-05,
+      "loss": 0.2336,
+      "step": 5098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462120960195759e-05,
+      "loss": 0.2545,
+      "step": 5099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4603898313196215e-05,
+      "loss": 0.2804,
+      "step": 5100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4586587678875835e-05,
+      "loss": 0.3264,
+      "step": 5101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.456927770109589e-05,
+      "loss": 0.2098,
+      "step": 5102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.455196838195576e-05,
+      "loss": 0.1978,
+      "step": 5103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4534659723554714e-05,
+      "loss": 0.2673,
+      "step": 5104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.451735172799194e-05,
+      "loss": 0.2954,
+      "step": 5105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.450004439736656e-05,
+      "loss": 0.2587,
+      "step": 5106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4482737733777623e-05,
+      "loss": 0.2464,
+      "step": 5107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.446543173932409e-05,
+      "loss": 0.2985,
+      "step": 5108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.444812641610482e-05,
+      "loss": 0.2584,
+      "step": 5109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4430821766218644e-05,
+      "loss": 0.2162,
+      "step": 5110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.441351779176422e-05,
+      "loss": 0.29,
+      "step": 5111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4396214494840224e-05,
+      "loss": 0.3157,
+      "step": 5112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4378911877545195e-05,
+      "loss": 0.2271,
+      "step": 5113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4361609941977585e-05,
+      "loss": 0.2146,
+      "step": 5114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.434430869023579e-05,
+      "loss": 0.2424,
+      "step": 5115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.432700812441812e-05,
+      "loss": 0.2745,
+      "step": 5116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4309708246622786e-05,
+      "loss": 0.2943,
+      "step": 5117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.429240905894792e-05,
+      "loss": 0.2774,
+      "step": 5118
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.427511056349157e-05,
+      "loss": 0.2491,
+      "step": 5119
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.425781276235171e-05,
+      "loss": 0.2843,
+      "step": 5120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.424051565762623e-05,
+      "loss": 0.3022,
+      "step": 5121
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.42232192514129e-05,
+      "loss": 0.2396,
+      "step": 5122
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.420592354580948e-05,
+      "loss": 0.2261,
+      "step": 5123
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4188628542913556e-05,
+      "loss": 0.2417,
+      "step": 5124
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.41713342448227e-05,
+      "loss": 0.2294,
+      "step": 5125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.415404065363437e-05,
+      "loss": 0.2411,
+      "step": 5126
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.413674777144591e-05,
+      "loss": 0.2526,
+      "step": 5127
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.411945560035462e-05,
+      "loss": 0.2648,
+      "step": 5128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.410216414245771e-05,
+      "loss": 0.2107,
+      "step": 5129
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.408487339985229e-05,
+      "loss": 0.3119,
+      "step": 5130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4067583374635386e-05,
+      "loss": 0.2638,
+      "step": 5131
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4050294068903926e-05,
+      "loss": 0.2679,
+      "step": 5132
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.403300548475478e-05,
+      "loss": 0.26,
+      "step": 5133
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.401571762428471e-05,
+      "loss": 0.2635,
+      "step": 5134
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399843048959038e-05,
+      "loss": 0.3667,
+      "step": 5135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.39811440827684e-05,
+      "loss": 0.2235,
+      "step": 5136
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.396385840591526e-05,
+      "loss": 0.303,
+      "step": 5137
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.394657346112737e-05,
+      "loss": 0.295,
+      "step": 5138
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.392928925050105e-05,
+      "loss": 0.2513,
+      "step": 5139
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.391200577613256e-05,
+      "loss": 0.286,
+      "step": 5140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3894723040118016e-05,
+      "loss": 0.2266,
+      "step": 5141
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3877441044553486e-05,
+      "loss": 0.2704,
+      "step": 5142
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.386015979153496e-05,
+      "loss": 0.2759,
+      "step": 5143
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.384287928315826e-05,
+      "loss": 0.2644,
+      "step": 5144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.382559952151921e-05,
+      "loss": 0.352,
+      "step": 5145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.38083205087135e-05,
+      "loss": 0.2601,
+      "step": 5146
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3791042246836734e-05,
+      "loss": 0.3298,
+      "step": 5147
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.377376473798441e-05,
+      "loss": 0.2459,
+      "step": 5148
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.375648798425197e-05,
+      "loss": 0.2975,
+      "step": 5149
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.373921198773474e-05,
+      "loss": 0.2372,
+      "step": 5150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3721936750527953e-05,
+      "loss": 0.1896,
+      "step": 5151
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.370466227472675e-05,
+      "loss": 0.316,
+      "step": 5152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.36873885624262e-05,
+      "loss": 0.2247,
+      "step": 5153
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.367011561572126e-05,
+      "loss": 0.2174,
+      "step": 5154
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.365284343670678e-05,
+      "loss": 0.2574,
+      "step": 5155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3635572027477576e-05,
+      "loss": 0.2303,
+      "step": 5156
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3618301390128296e-05,
+      "loss": 0.218,
+      "step": 5157
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3601031526753543e-05,
+      "loss": 0.2682,
+      "step": 5158
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.358376243944782e-05,
+      "loss": 0.2309,
+      "step": 5159
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.35664941303055e-05,
+      "loss": 0.3113,
+      "step": 5160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3549226601420895e-05,
+      "loss": 0.2386,
+      "step": 5161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.353195985488824e-05,
+      "loss": 0.2083,
+      "step": 5162
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3514693892801633e-05,
+      "loss": 0.2252,
+      "step": 5163
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.34974287172551e-05,
+      "loss": 0.2131,
+      "step": 5164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3480164330342564e-05,
+      "loss": 0.3026,
+      "step": 5165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3462900734157865e-05,
+      "loss": 0.2665,
+      "step": 5166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.344563793079474e-05,
+      "loss": 0.2008,
+      "step": 5167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.34283759223468e-05,
+      "loss": 0.2868,
+      "step": 5168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.341111471090762e-05,
+      "loss": 0.2702,
+      "step": 5169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.339385429857063e-05,
+      "loss": 0.3172,
+      "step": 5170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3376594687429176e-05,
+      "loss": 0.234,
+      "step": 5171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3359335879576515e-05,
+      "loss": 0.1603,
+      "step": 5172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3342077877105805e-05,
+      "loss": 0.2393,
+      "step": 5173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.332482068211009e-05,
+      "loss": 0.218,
+      "step": 5174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3307564296682356e-05,
+      "loss": 0.2045,
+      "step": 5175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.329030872291543e-05,
+      "loss": 0.3077,
+      "step": 5176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3273053962902076e-05,
+      "loss": 0.19,
+      "step": 5177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.325580001873498e-05,
+      "loss": 0.2749,
+      "step": 5178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3238546892506686e-05,
+      "loss": 0.2121,
+      "step": 5179
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3221294586309676e-05,
+      "loss": 0.293,
+      "step": 5180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3204043102236287e-05,
+      "loss": 0.3473,
+      "step": 5181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3186792442378816e-05,
+      "loss": 0.2739,
+      "step": 5182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3169542608829416e-05,
+      "loss": 0.2192,
+      "step": 5183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.315229360368014e-05,
+      "loss": 0.3512,
+      "step": 5184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.313504542902298e-05,
+      "loss": 0.2968,
+      "step": 5185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.311779808694979e-05,
+      "loss": 0.254,
+      "step": 5186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.310055157955232e-05,
+      "loss": 0.2072,
+      "step": 5187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.308330590892224e-05,
+      "loss": 0.3118,
+      "step": 5188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.306606107715112e-05,
+      "loss": 0.2242,
+      "step": 5189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3048817086330404e-05,
+      "loss": 0.281,
+      "step": 5190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3031573938551465e-05,
+      "loss": 0.3941,
+      "step": 5191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.301433163590556e-05,
+      "loss": 0.326,
+      "step": 5192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.299709018048381e-05,
+      "loss": 0.2689,
+      "step": 5193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297984957437728e-05,
+      "loss": 0.2506,
+      "step": 5194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.296260981967693e-05,
+      "loss": 0.2656,
+      "step": 5195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.29453709184736e-05,
+      "loss": 0.239,
+      "step": 5196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.292813287285801e-05,
+      "loss": 0.2967,
+      "step": 5197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.291089568492082e-05,
+      "loss": 0.2316,
+      "step": 5198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2893659356752544e-05,
+      "loss": 0.2898,
+      "step": 5199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.287642389044362e-05,
+      "loss": 0.27,
+      "step": 5200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.285918928808436e-05,
+      "loss": 0.3342,
+      "step": 5201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2841955551765e-05,
+      "loss": 0.3307,
+      "step": 5202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.282472268357565e-05,
+      "loss": 0.2063,
+      "step": 5203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.280749068560629e-05,
+      "loss": 0.2249,
+      "step": 5204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.279025955994686e-05,
+      "loss": 0.2908,
+      "step": 5205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.277302930868714e-05,
+      "loss": 0.1975,
+      "step": 5206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.275579993391682e-05,
+      "loss": 0.2741,
+      "step": 5207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.27385714377255e-05,
+      "loss": 0.2418,
+      "step": 5208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.272134382220263e-05,
+      "loss": 0.2719,
+      "step": 5209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.270411708943758e-05,
+      "loss": 0.2396,
+      "step": 5210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.268689124151964e-05,
+      "loss": 0.2531,
+      "step": 5211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2669666280537965e-05,
+      "loss": 0.3275,
+      "step": 5212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2652442208581585e-05,
+      "loss": 0.2669,
+      "step": 5213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2635219027739435e-05,
+      "loss": 0.2156,
+      "step": 5214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.261799674010038e-05,
+      "loss": 0.2774,
+      "step": 5215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.260077534775312e-05,
+      "loss": 0.2414,
+      "step": 5216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.258355485278628e-05,
+      "loss": 0.2515,
+      "step": 5217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.256633525728837e-05,
+      "loss": 0.275,
+      "step": 5218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.254911656334778e-05,
+      "loss": 0.2491,
+      "step": 5219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.253189877305281e-05,
+      "loss": 0.2683,
+      "step": 5220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2514681888491615e-05,
+      "loss": 0.3143,
+      "step": 5221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.24974659117523e-05,
+      "loss": 0.2644,
+      "step": 5222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2480250844922795e-05,
+      "loss": 0.1859,
+      "step": 5223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2463036690090964e-05,
+      "loss": 0.2329,
+      "step": 5224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.244582344934457e-05,
+      "loss": 0.3408,
+      "step": 5225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2428611124771184e-05,
+      "loss": 0.2351,
+      "step": 5226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.241139971845834e-05,
+      "loss": 0.251,
+      "step": 5227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.239418923249347e-05,
+      "loss": 0.2924,
+      "step": 5228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2376979668963854e-05,
+      "loss": 0.2535,
+      "step": 5229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.235977102995665e-05,
+      "loss": 0.2491,
+      "step": 5230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.234256331755897e-05,
+      "loss": 0.2765,
+      "step": 5231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.232535653385774e-05,
+      "loss": 0.2257,
+      "step": 5232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2308150680939814e-05,
+      "loss": 0.2484,
+      "step": 5233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.229094576089192e-05,
+      "loss": 0.2477,
+      "step": 5234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.227374177580069e-05,
+      "loss": 0.2555,
+      "step": 5235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.225653872775262e-05,
+      "loss": 0.2639,
+      "step": 5236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.223933661883408e-05,
+      "loss": 0.2741,
+      "step": 5237
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.222213545113139e-05,
+      "loss": 0.3193,
+      "step": 5238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.220493522673067e-05,
+      "loss": 0.2942,
+      "step": 5239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.218773594771801e-05,
+      "loss": 0.2745,
+      "step": 5240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.217053761617933e-05,
+      "loss": 0.2363,
+      "step": 5241
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.215334023420043e-05,
+      "loss": 0.2216,
+      "step": 5242
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.213614380386702e-05,
+      "loss": 0.2873,
+      "step": 5243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2118948327264703e-05,
+      "loss": 0.2628,
+      "step": 5244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.210175380647894e-05,
+      "loss": 0.2412,
+      "step": 5245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2084560243595106e-05,
+      "loss": 0.2166,
+      "step": 5246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.206736764069841e-05,
+      "loss": 0.3005,
+      "step": 5247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2050175999874016e-05,
+      "loss": 0.2801,
+      "step": 5248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.20329853232069e-05,
+      "loss": 0.1856,
+      "step": 5249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2015795612781965e-05,
+      "loss": 0.252,
+      "step": 5250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1998606870683995e-05,
+      "loss": 0.2376,
+      "step": 5251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.198141909899763e-05,
+      "loss": 0.2211,
+      "step": 5252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1964232299807415e-05,
+      "loss": 0.2627,
+      "step": 5253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1947046475197765e-05,
+      "loss": 0.2995,
+      "step": 5254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1929861627252997e-05,
+      "loss": 0.285,
+      "step": 5255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1912677758057286e-05,
+      "loss": 0.3466,
+      "step": 5256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1895494869694706e-05,
+      "loss": 0.295,
+      "step": 5257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.187831296424918e-05,
+      "loss": 0.2721,
+      "step": 5258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.186113204380455e-05,
+      "loss": 0.198,
+      "step": 5259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.184395211044453e-05,
+      "loss": 0.2285,
+      "step": 5260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1826773166252696e-05,
+      "loss": 0.3633,
+      "step": 5261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.180959521331253e-05,
+      "loss": 0.2832,
+      "step": 5262
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.179241825370735e-05,
+      "loss": 0.3123,
+      "step": 5263
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1775242289520436e-05,
+      "loss": 0.3217,
+      "step": 5264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.175806732283485e-05,
+      "loss": 0.2385,
+      "step": 5265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.174089335573359e-05,
+      "loss": 0.3016,
+      "step": 5266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.172372039029952e-05,
+      "loss": 0.2452,
+      "step": 5267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1706548428615384e-05,
+      "loss": 0.2334,
+      "step": 5268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.168937747276381e-05,
+      "loss": 0.297,
+      "step": 5269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1672207524827275e-05,
+      "loss": 0.2171,
+      "step": 5270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.165503858688818e-05,
+      "loss": 0.3003,
+      "step": 5271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.163787066102878e-05,
+      "loss": 0.2902,
+      "step": 5272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.16207037493312e-05,
+      "loss": 0.2713,
+      "step": 5273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.160353785387746e-05,
+      "loss": 0.2987,
+      "step": 5274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.158637297674941e-05,
+      "loss": 0.243,
+      "step": 5275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.156920912002883e-05,
+      "loss": 0.2627,
+      "step": 5276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.155204628579738e-05,
+      "loss": 0.2291,
+      "step": 5277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.153488447613654e-05,
+      "loss": 0.2818,
+      "step": 5278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.151772369312772e-05,
+      "loss": 0.2147,
+      "step": 5279
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1500563938852185e-05,
+      "loss": 0.2418,
+      "step": 5280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1483405215391083e-05,
+      "loss": 0.2494,
+      "step": 5281
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.146624752482541e-05,
+      "loss": 0.2649,
+      "step": 5282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.144909086923606e-05,
+      "loss": 0.2507,
+      "step": 5283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.143193525070382e-05,
+      "loss": 0.2615,
+      "step": 5284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.141478067130931e-05,
+      "loss": 0.2219,
+      "step": 5285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.139762713313304e-05,
+      "loss": 0.2216,
+      "step": 5286
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1380474638255396e-05,
+      "loss": 0.3001,
+      "step": 5287
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.136332318875665e-05,
+      "loss": 0.2988,
+      "step": 5288
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.134617278671694e-05,
+      "loss": 0.2465,
+      "step": 5289
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1329023434216275e-05,
+      "loss": 0.1573,
+      "step": 5290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1311875133334504e-05,
+      "loss": 0.2803,
+      "step": 5291
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1294727886151385e-05,
+      "loss": 0.2232,
+      "step": 5292
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.127758169474657e-05,
+      "loss": 0.2652,
+      "step": 5293
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.126043656119954e-05,
+      "loss": 0.2694,
+      "step": 5294
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1243292487589654e-05,
+      "loss": 0.3287,
+      "step": 5295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.122614947599616e-05,
+      "loss": 0.1994,
+      "step": 5296
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120900752849816e-05,
+      "loss": 0.1861,
+      "step": 5297
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.119186664717465e-05,
+      "loss": 0.2327,
+      "step": 5298
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.117472683410446e-05,
+      "loss": 0.2789,
+      "step": 5299
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.115758809136634e-05,
+      "loss": 0.1833,
+      "step": 5300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.114045042103887e-05,
+      "loss": 0.2452,
+      "step": 5301
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.11233138252005e-05,
+      "loss": 0.2459,
+      "step": 5302
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.110617830592957e-05,
+      "loss": 0.323,
+      "step": 5303
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.108904386530429e-05,
+      "loss": 0.2168,
+      "step": 5304
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.107191050540272e-05,
+      "loss": 0.2679,
+      "step": 5305
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.105477822830281e-05,
+      "loss": 0.2614,
+      "step": 5306
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1037647036082374e-05,
+      "loss": 0.2544,
+      "step": 5307
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1020516930819056e-05,
+      "loss": 0.2395,
+      "step": 5308
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.100338791459042e-05,
+      "loss": 0.3263,
+      "step": 5309
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.098625998947388e-05,
+      "loss": 0.2905,
+      "step": 5310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.096913315754673e-05,
+      "loss": 0.2185,
+      "step": 5311
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.095200742088609e-05,
+      "loss": 0.2372,
+      "step": 5312
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0934882781568994e-05,
+      "loss": 0.2331,
+      "step": 5313
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.091775924167232e-05,
+      "loss": 0.212,
+      "step": 5314
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.090063680327282e-05,
+      "loss": 0.2236,
+      "step": 5315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.088351546844709e-05,
+      "loss": 0.2813,
+      "step": 5316
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.086639523927164e-05,
+      "loss": 0.2212,
+      "step": 5317
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.084927611782281e-05,
+      "loss": 0.2574,
+      "step": 5318
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.083215810617678e-05,
+      "loss": 0.2621,
+      "step": 5319
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.081504120640968e-05,
+      "loss": 0.2148,
+      "step": 5320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0797925420597415e-05,
+      "loss": 0.2129,
+      "step": 5321
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.078081075081581e-05,
+      "loss": 0.2665,
+      "step": 5322
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.076369719914055e-05,
+      "loss": 0.282,
+      "step": 5323
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.074658476764714e-05,
+      "loss": 0.2854,
+      "step": 5324
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.072947345841098e-05,
+      "loss": 0.2087,
+      "step": 5325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.071236327350737e-05,
+      "loss": 0.2856,
+      "step": 5326
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0695254215011417e-05,
+      "loss": 0.2068,
+      "step": 5327
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0678146284998125e-05,
+      "loss": 0.2445,
+      "step": 5328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066103948554232e-05,
+      "loss": 0.2341,
+      "step": 5329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.064393381871876e-05,
+      "loss": 0.1904,
+      "step": 5330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062682928660201e-05,
+      "loss": 0.1808,
+      "step": 5331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.06097258912665e-05,
+      "loss": 0.1891,
+      "step": 5332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059262363478655e-05,
+      "loss": 0.211,
+      "step": 5333
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057552251923633e-05,
+      "loss": 0.3315,
+      "step": 5334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0558422546689865e-05,
+      "loss": 0.2118,
+      "step": 5335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054132371922102e-05,
+      "loss": 0.2624,
+      "step": 5336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05242260389036e-05,
+      "loss": 0.28,
+      "step": 5337
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0507129507811176e-05,
+      "loss": 0.2452,
+      "step": 5338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049003412801724e-05,
+      "loss": 0.1864,
+      "step": 5339
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047293990159511e-05,
+      "loss": 0.311,
+      "step": 5340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0455846830617997e-05,
+      "loss": 0.2125,
+      "step": 5341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.043875491715892e-05,
+      "loss": 0.2121,
+      "step": 5342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.042166416329083e-05,
+      "loss": 0.2634,
+      "step": 5343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0404574571086484e-05,
+      "loss": 0.2521,
+      "step": 5344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.038748614261851e-05,
+      "loss": 0.279,
+      "step": 5345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.037039887995941e-05,
+      "loss": 0.2265,
+      "step": 5346
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.035331278518153e-05,
+      "loss": 0.249,
+      "step": 5347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033622786035707e-05,
+      "loss": 0.2048,
+      "step": 5348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0319144107558085e-05,
+      "loss": 0.2134,
+      "step": 5349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030206152885653e-05,
+      "loss": 0.2521,
+      "step": 5350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028498012632416e-05,
+      "loss": 0.3204,
+      "step": 5351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.026789990203261e-05,
+      "loss": 0.1844,
+      "step": 5352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.025082085805341e-05,
+      "loss": 0.2128,
+      "step": 5353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.023374299645787e-05,
+      "loss": 0.2745,
+      "step": 5354
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.021666631931722e-05,
+      "loss": 0.2571,
+      "step": 5355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0199590828702544e-05,
+      "loss": 0.232,
+      "step": 5356
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.018251652668473e-05,
+      "loss": 0.2654,
+      "step": 5357
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0165443415334546e-05,
+      "loss": 0.2347,
+      "step": 5358
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0148371496722655e-05,
+      "loss": 0.2031,
+      "step": 5359
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.013130077291954e-05,
+      "loss": 0.2954,
+      "step": 5360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.011423124599554e-05,
+      "loss": 0.2309,
+      "step": 5361
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0097162918020836e-05,
+      "loss": 0.274,
+      "step": 5362
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0080095791065505e-05,
+      "loss": 0.2795,
+      "step": 5363
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.006302986719945e-05,
+      "loss": 0.2263,
+      "step": 5364
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.004596514849242e-05,
+      "loss": 0.3072,
+      "step": 5365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0028901637014043e-05,
+      "loss": 0.2148,
+      "step": 5366
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.001183933483378e-05,
+      "loss": 0.3654,
+      "step": 5367
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9994778244020955e-05,
+      "loss": 0.2815,
+      "step": 5368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.997771836664473e-05,
+      "loss": 0.2726,
+      "step": 5369
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.996065970477415e-05,
+      "loss": 0.2261,
+      "step": 5370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9943602260478095e-05,
+      "loss": 0.2027,
+      "step": 5371
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.99265460358253e-05,
+      "loss": 0.2501,
+      "step": 5372
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.990949103288433e-05,
+      "loss": 0.2963,
+      "step": 5373
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.989243725372362e-05,
+      "loss": 0.2853,
+      "step": 5374
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.987538470041149e-05,
+      "loss": 0.2283,
+      "step": 5375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.985833337501605e-05,
+      "loss": 0.2398,
+      "step": 5376
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9841283279605306e-05,
+      "loss": 0.2844,
+      "step": 5377
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.982423441624708e-05,
+      "loss": 0.2819,
+      "step": 5378
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9807186787009085e-05,
+      "loss": 0.2671,
+      "step": 5379
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9790140393958866e-05,
+      "loss": 0.2069,
+      "step": 5380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.977309523916379e-05,
+      "loss": 0.2649,
+      "step": 5381
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.975605132469111e-05,
+      "loss": 0.2124,
+      "step": 5382
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.973900865260794e-05,
+      "loss": 0.3036,
+      "step": 5383
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9721967224981194e-05,
+      "loss": 0.3252,
+      "step": 5384
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.970492704387766e-05,
+      "loss": 0.2526,
+      "step": 5385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9687888111364e-05,
+      "loss": 0.2161,
+      "step": 5386
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.967085042950668e-05,
+      "loss": 0.2797,
+      "step": 5387
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.965381400037205e-05,
+      "loss": 0.2825,
+      "step": 5388
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.96367788260263e-05,
+      "loss": 0.2327,
+      "step": 5389
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.961974490853544e-05,
+      "loss": 0.2678,
+      "step": 5390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.960271224996534e-05,
+      "loss": 0.2302,
+      "step": 5391
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9585680852381765e-05,
+      "loss": 0.2778,
+      "step": 5392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.956865071785026e-05,
+      "loss": 0.2244,
+      "step": 5393
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955162184843625e-05,
+      "loss": 0.2204,
+      "step": 5394
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9534594246205016e-05,
+      "loss": 0.2807,
+      "step": 5395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.951756791322165e-05,
+      "loss": 0.2128,
+      "step": 5396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.950054285155114e-05,
+      "loss": 0.2708,
+      "step": 5397
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.948351906325825e-05,
+      "loss": 0.3059,
+      "step": 5398
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9466496550407675e-05,
+      "loss": 0.223,
+      "step": 5399
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.944947531506388e-05,
+      "loss": 0.3259,
+      "step": 5400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.943245535929122e-05,
+      "loss": 0.2342,
+      "step": 5401
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9415436685153866e-05,
+      "loss": 0.2372,
+      "step": 5402
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.939841929471587e-05,
+      "loss": 0.2283,
+      "step": 5403
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.93814031900411e-05,
+      "loss": 0.3108,
+      "step": 5404
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9364388373193276e-05,
+      "loss": 0.2547,
+      "step": 5405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9347374846235955e-05,
+      "loss": 0.2489,
+      "step": 5406
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.933036261123252e-05,
+      "loss": 0.2182,
+      "step": 5407
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.931335167024627e-05,
+      "loss": 0.2293,
+      "step": 5408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.929634202534026e-05,
+      "loss": 0.2108,
+      "step": 5409
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.927933367857744e-05,
+      "loss": 0.2416,
+      "step": 5410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9262326632020575e-05,
+      "loss": 0.2824,
+      "step": 5411
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.924532088773231e-05,
+      "loss": 0.2871,
+      "step": 5412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9228316447775086e-05,
+      "loss": 0.2182,
+      "step": 5413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9211313314211226e-05,
+      "loss": 0.2229,
+      "step": 5414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.919431148910284e-05,
+      "loss": 0.3058,
+      "step": 5415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.917731097451196e-05,
+      "loss": 0.2349,
+      "step": 5416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9160311772500386e-05,
+      "loss": 0.2312,
+      "step": 5417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.914331388512978e-05,
+      "loss": 0.2575,
+      "step": 5418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.912631731446168e-05,
+      "loss": 0.2431,
+      "step": 5419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9109322062557424e-05,
+      "loss": 0.3202,
+      "step": 5420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9092328131478204e-05,
+      "loss": 0.231,
+      "step": 5421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.907533552328504e-05,
+      "loss": 0.3028,
+      "step": 5422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9058344240038795e-05,
+      "loss": 0.2661,
+      "step": 5423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.904135428380019e-05,
+      "loss": 0.2258,
+      "step": 5424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.902436565662977e-05,
+      "loss": 0.2691,
+      "step": 5425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.900737836058792e-05,
+      "loss": 0.2095,
+      "step": 5426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.899039239773487e-05,
+      "loss": 0.2133,
+      "step": 5427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.897340777013068e-05,
+      "loss": 0.2276,
+      "step": 5428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8956424479835245e-05,
+      "loss": 0.2458,
+      "step": 5429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.893944252890832e-05,
+      "loss": 0.2408,
+      "step": 5430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.892246191940946e-05,
+      "loss": 0.2713,
+      "step": 5431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89054826533981e-05,
+      "loss": 0.1702,
+      "step": 5432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.888850473293348e-05,
+      "loss": 0.31,
+      "step": 5433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.887152816007469e-05,
+      "loss": 0.2084,
+      "step": 5434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.885455293688064e-05,
+      "loss": 0.1824,
+      "step": 5435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8837579065410125e-05,
+      "loss": 0.3207,
+      "step": 5436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.882060654772171e-05,
+      "loss": 0.2142,
+      "step": 5437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.880363538587385e-05,
+      "loss": 0.2614,
+      "step": 5438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8786665581924805e-05,
+      "loss": 0.3259,
+      "step": 5439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8769697137932655e-05,
+      "loss": 0.1836,
+      "step": 5440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8752730055955364e-05,
+      "loss": 0.2628,
+      "step": 5441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.873576433805071e-05,
+      "loss": 0.2964,
+      "step": 5442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.871879998627629e-05,
+      "loss": 0.3248,
+      "step": 5443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.870183700268954e-05,
+      "loss": 0.3096,
+      "step": 5444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.868487538934774e-05,
+      "loss": 0.2709,
+      "step": 5445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.866791514830801e-05,
+      "loss": 0.288,
+      "step": 5446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.865095628162728e-05,
+      "loss": 0.2206,
+      "step": 5447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.863399879136235e-05,
+      "loss": 0.22,
+      "step": 5448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8617042679569805e-05,
+      "loss": 0.3011,
+      "step": 5449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8600087948306105e-05,
+      "loss": 0.3122,
+      "step": 5450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8583134599627505e-05,
+      "loss": 0.2468,
+      "step": 5451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.856618263559013e-05,
+      "loss": 0.3182,
+      "step": 5452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8549232058249915e-05,
+      "loss": 0.2338,
+      "step": 5453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.853228286966265e-05,
+      "loss": 0.254,
+      "step": 5454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.851533507188391e-05,
+      "loss": 0.2407,
+      "step": 5455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.849838866696913e-05,
+      "loss": 0.2579,
+      "step": 5456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.848144365697358e-05,
+      "loss": 0.2527,
+      "step": 5457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.846450004395238e-05,
+      "loss": 0.2732,
+      "step": 5458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8447557829960426e-05,
+      "loss": 0.2601,
+      "step": 5459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8430617017052485e-05,
+      "loss": 0.2734,
+      "step": 5460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8413677607283156e-05,
+      "loss": 0.2307,
+      "step": 5461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.839673960270685e-05,
+      "loss": 0.2999,
+      "step": 5462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.837980300537781e-05,
+      "loss": 0.1955,
+      "step": 5463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.83628678173501e-05,
+      "loss": 0.2156,
+      "step": 5464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.834593404067765e-05,
+      "loss": 0.2147,
+      "step": 5465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.832900167741418e-05,
+      "loss": 0.2042,
+      "step": 5466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8312070729613256e-05,
+      "loss": 0.2754,
+      "step": 5467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.829514119932826e-05,
+      "loss": 0.2968,
+      "step": 5468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8278213088612444e-05,
+      "loss": 0.1975,
+      "step": 5469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.826128639951881e-05,
+      "loss": 0.2548,
+      "step": 5470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8244361134100273e-05,
+      "loss": 0.2409,
+      "step": 5471
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.822743729440951e-05,
+      "loss": 0.2789,
+      "step": 5472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.821051488249903e-05,
+      "loss": 0.2066,
+      "step": 5473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.819359390042123e-05,
+      "loss": 0.3355,
+      "step": 5474
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.817667435022828e-05,
+      "loss": 0.2678,
+      "step": 5475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.815975623397218e-05,
+      "loss": 0.1956,
+      "step": 5476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.814283955370476e-05,
+      "loss": 0.2334,
+      "step": 5477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.81259243114777e-05,
+      "loss": 0.2044,
+      "step": 5478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.810901050934247e-05,
+      "loss": 0.3067,
+      "step": 5479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.809209814935038e-05,
+      "loss": 0.219,
+      "step": 5480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8075187233552586e-05,
+      "loss": 0.2333,
+      "step": 5481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.805827776400004e-05,
+      "loss": 0.1663,
+      "step": 5482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.804136974274352e-05,
+      "loss": 0.2514,
+      "step": 5483
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.802446317183364e-05,
+      "loss": 0.2577,
+      "step": 5484
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8007558053320836e-05,
+      "loss": 0.247,
+      "step": 5485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.799065438925537e-05,
+      "loss": 0.27,
+      "step": 5486
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7973752181687335e-05,
+      "loss": 0.2639,
+      "step": 5487
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7956851432666606e-05,
+      "loss": 0.2614,
+      "step": 5488
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7939952144242916e-05,
+      "loss": 0.1985,
+      "step": 5489
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7923054318465845e-05,
+      "loss": 0.1541,
+      "step": 5490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.790615795738475e-05,
+      "loss": 0.1806,
+      "step": 5491
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7889263063048825e-05,
+      "loss": 0.2694,
+      "step": 5492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.787236963750708e-05,
+      "loss": 0.2036,
+      "step": 5493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.785547768280837e-05,
+      "loss": 0.1766,
+      "step": 5494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7838587201001364e-05,
+      "loss": 0.2975,
+      "step": 5495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.782169819413452e-05,
+      "loss": 0.3586,
+      "step": 5496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7804810664256154e-05,
+      "loss": 0.2397,
+      "step": 5497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7787924613414406e-05,
+      "loss": 0.2009,
+      "step": 5498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.777104004365721e-05,
+      "loss": 0.2685,
+      "step": 5499
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7754156957032317e-05,
+      "loss": 0.2367,
+      "step": 5500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.773727535558734e-05,
+      "loss": 0.1966,
+      "step": 5501
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.772039524136967e-05,
+      "loss": 0.2649,
+      "step": 5502
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7703516616426546e-05,
+      "loss": 0.2057,
+      "step": 5503
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7686639482805e-05,
+      "loss": 0.2618,
+      "step": 5504
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.766976384255189e-05,
+      "loss": 0.3028,
+      "step": 5505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.76528896977139e-05,
+      "loss": 0.2757,
+      "step": 5506
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.763601705033755e-05,
+      "loss": 0.2546,
+      "step": 5507
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7619145902469157e-05,
+      "loss": 0.2622,
+      "step": 5508
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.760227625615485e-05,
+      "loss": 0.2497,
+      "step": 5509
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.758540811344058e-05,
+      "loss": 0.1842,
+      "step": 5510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.756854147637213e-05,
+      "loss": 0.2183,
+      "step": 5511
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.75516763469951e-05,
+      "loss": 0.2061,
+      "step": 5512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.753481272735488e-05,
+      "loss": 0.245,
+      "step": 5513
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.751795061949671e-05,
+      "loss": 0.224,
+      "step": 5514
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.750109002546564e-05,
+      "loss": 0.2327,
+      "step": 5515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7484230947306495e-05,
+      "loss": 0.1851,
+      "step": 5516
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.746737338706397e-05,
+      "loss": 0.2057,
+      "step": 5517
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.745051734678256e-05,
+      "loss": 0.2815,
+      "step": 5518
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.743366282850658e-05,
+      "loss": 0.3687,
+      "step": 5519
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.741680983428013e-05,
+      "loss": 0.1779,
+      "step": 5520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.739995836614716e-05,
+      "loss": 0.3109,
+      "step": 5521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.738310842615139e-05,
+      "loss": 0.2182,
+      "step": 5522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.736626001633643e-05,
+      "loss": 0.3323,
+      "step": 5523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.734941313874565e-05,
+      "loss": 0.2811,
+      "step": 5524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.733256779542223e-05,
+      "loss": 0.2449,
+      "step": 5525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.731572398840918e-05,
+      "loss": 0.2947,
+      "step": 5526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7298881719749345e-05,
+      "loss": 0.2733,
+      "step": 5527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.728204099148533e-05,
+      "loss": 0.2358,
+      "step": 5528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7265201805659616e-05,
+      "loss": 0.2714,
+      "step": 5529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.724836416431443e-05,
+      "loss": 0.2512,
+      "step": 5530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7231528069491885e-05,
+      "loss": 0.2837,
+      "step": 5531
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7214693523233836e-05,
+      "loss": 0.3265,
+      "step": 5532
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.719786052758198e-05,
+      "loss": 0.2167,
+      "step": 5533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.718102908457787e-05,
+      "loss": 0.2856,
+      "step": 5534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7164199196262795e-05,
+      "loss": 0.2792,
+      "step": 5535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.714737086467791e-05,
+      "loss": 0.2978,
+      "step": 5536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7130544091864125e-05,
+      "loss": 0.2084,
+      "step": 5537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7113718879862236e-05,
+      "loss": 0.2639,
+      "step": 5538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.709689523071277e-05,
+      "loss": 0.3133,
+      "step": 5539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.708007314645614e-05,
+      "loss": 0.1786,
+      "step": 5540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7063252629132526e-05,
+      "loss": 0.2835,
+      "step": 5541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.704643368078191e-05,
+      "loss": 0.261,
+      "step": 5542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7029616303444125e-05,
+      "loss": 0.2925,
+      "step": 5543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7012800499158765e-05,
+      "loss": 0.2276,
+      "step": 5544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.699598626996527e-05,
+      "loss": 0.1942,
+      "step": 5545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6979173617902864e-05,
+      "loss": 0.2846,
+      "step": 5546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.696236254501061e-05,
+      "loss": 0.227,
+      "step": 5547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.694555305332734e-05,
+      "loss": 0.2332,
+      "step": 5548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.692874514489173e-05,
+      "loss": 0.3201,
+      "step": 5549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.691193882174222e-05,
+      "loss": 0.2888,
+      "step": 5550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.689513408591714e-05,
+      "loss": 0.3015,
+      "step": 5551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6878330939454526e-05,
+      "loss": 0.2353,
+      "step": 5552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.68615293843923e-05,
+      "loss": 0.1821,
+      "step": 5553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.684472942276813e-05,
+      "loss": 0.2214,
+      "step": 5554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6827931056619536e-05,
+      "loss": 0.3053,
+      "step": 5555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.681113428798384e-05,
+      "loss": 0.1905,
+      "step": 5556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.679433911889814e-05,
+      "loss": 0.235,
+      "step": 5557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.677754555139938e-05,
+      "loss": 0.2923,
+      "step": 5558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.676075358752426e-05,
+      "loss": 0.1503,
+      "step": 5559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.674396322930935e-05,
+      "loss": 0.2525,
+      "step": 5560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.672717447879097e-05,
+      "loss": 0.2832,
+      "step": 5561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6710387338005256e-05,
+      "loss": 0.2401,
+      "step": 5562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6693601808988184e-05,
+      "loss": 0.1403,
+      "step": 5563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.66768178937755e-05,
+      "loss": 0.1653,
+      "step": 5564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6660035594402756e-05,
+      "loss": 0.2597,
+      "step": 5565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.664325491290531e-05,
+      "loss": 0.2548,
+      "step": 5566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.662647585131834e-05,
+      "loss": 0.2172,
+      "step": 5567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.660969841167682e-05,
+      "loss": 0.2935,
+      "step": 5568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.659292259601552e-05,
+      "loss": 0.2136,
+      "step": 5569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6576148406369e-05,
+      "loss": 0.3103,
+      "step": 5570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.655937584477165e-05,
+      "loss": 0.2494,
+      "step": 5571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.654260491325765e-05,
+      "loss": 0.2771,
+      "step": 5572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.652583561386099e-05,
+      "loss": 0.3003,
+      "step": 5573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.650906794861546e-05,
+      "loss": 0.2381,
+      "step": 5574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6492301919554626e-05,
+      "loss": 0.1786,
+      "step": 5575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.647553752871191e-05,
+      "loss": 0.3069,
+      "step": 5576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.645877477812048e-05,
+      "loss": 0.2582,
+      "step": 5577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6442013669813336e-05,
+      "loss": 0.2706,
+      "step": 5578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.642525420582326e-05,
+      "loss": 0.1291,
+      "step": 5579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.640849638818286e-05,
+      "loss": 0.1888,
+      "step": 5580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6391740218924524e-05,
+      "loss": 0.1739,
+      "step": 5581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.637498570008044e-05,
+      "loss": 0.1854,
+      "step": 5582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.635823283368262e-05,
+      "loss": 0.2985,
+      "step": 5583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6341481621762826e-05,
+      "loss": 0.1867,
+      "step": 5584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6324732066352694e-05,
+      "loss": 0.206,
+      "step": 5585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6307984169483576e-05,
+      "loss": 0.1839,
+      "step": 5586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6291237933186685e-05,
+      "loss": 0.219,
+      "step": 5587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.627449335949298e-05,
+      "loss": 0.2615,
+      "step": 5588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6257750450433284e-05,
+      "loss": 0.2797,
+      "step": 5589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.624100920803817e-05,
+      "loss": 0.2763,
+      "step": 5590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.622426963433802e-05,
+      "loss": 0.2818,
+      "step": 5591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.620753173136301e-05,
+      "loss": 0.2907,
+      "step": 5592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6190795501143125e-05,
+      "loss": 0.1933,
+      "step": 5593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6174060945708146e-05,
+      "loss": 0.2689,
+      "step": 5594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.615732806708763e-05,
+      "loss": 0.2263,
+      "step": 5595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.614059686731096e-05,
+      "loss": 0.2092,
+      "step": 5596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6123867348407295e-05,
+      "loss": 0.3718,
+      "step": 5597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.61071395124056e-05,
+      "loss": 0.1703,
+      "step": 5598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.609041336133462e-05,
+      "loss": 0.3583,
+      "step": 5599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6073688897222936e-05,
+      "loss": 0.1959,
+      "step": 5600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.605696612209887e-05,
+      "loss": 0.2169,
+      "step": 5601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.604024503799059e-05,
+      "loss": 0.1506,
+      "step": 5602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6023525646926e-05,
+      "loss": 0.2801,
+      "step": 5603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.600680795093286e-05,
+      "loss": 0.2791,
+      "step": 5604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.599009195203868e-05,
+      "loss": 0.2841,
+      "step": 5605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5973377652270807e-05,
+      "loss": 0.2693,
+      "step": 5606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.595666505365634e-05,
+      "loss": 0.263,
+      "step": 5607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5939954158222176e-05,
+      "loss": 0.2337,
+      "step": 5608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.592324496799505e-05,
+      "loss": 0.2346,
+      "step": 5609
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.590653748500144e-05,
+      "loss": 0.2723,
+      "step": 5610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.588983171126762e-05,
+      "loss": 0.2427,
+      "step": 5611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.587312764881969e-05,
+      "loss": 0.2435,
+      "step": 5612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5856425299683536e-05,
+      "loss": 0.2533,
+      "step": 5613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.58397246658848e-05,
+      "loss": 0.2481,
+      "step": 5614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5823025749448945e-05,
+      "loss": 0.2091,
+      "step": 5615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.580632855240124e-05,
+      "loss": 0.3425,
+      "step": 5616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.578963307676671e-05,
+      "loss": 0.2341,
+      "step": 5617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5772939324570196e-05,
+      "loss": 0.2433,
+      "step": 5618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5756247297836324e-05,
+      "loss": 0.2461,
+      "step": 5619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.573955699858949e-05,
+      "loss": 0.2547,
+      "step": 5620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5722868428853895e-05,
+      "loss": 0.2065,
+      "step": 5621
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5706181590653565e-05,
+      "loss": 0.2522,
+      "step": 5622
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.568949648601227e-05,
+      "loss": 0.2861,
+      "step": 5623
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5672813116953584e-05,
+      "loss": 0.2809,
+      "step": 5624
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5656131485500865e-05,
+      "loss": 0.2306,
+      "step": 5625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5639451593677284e-05,
+      "loss": 0.2042,
+      "step": 5626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5622773443505774e-05,
+      "loss": 0.2537,
+      "step": 5627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.560609703700906e-05,
+      "loss": 0.2298,
+      "step": 5628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.558942237620968e-05,
+      "loss": 0.3047,
+      "step": 5629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.557274946312992e-05,
+      "loss": 0.2325,
+      "step": 5630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.55560782997919e-05,
+      "loss": 0.2643,
+      "step": 5631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5539408888217474e-05,
+      "loss": 0.2656,
+      "step": 5632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5522741230428344e-05,
+      "loss": 0.1993,
+      "step": 5633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.550607532844596e-05,
+      "loss": 0.2466,
+      "step": 5634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.548941118429158e-05,
+      "loss": 0.2274,
+      "step": 5635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.54727487999862e-05,
+      "loss": 0.2203,
+      "step": 5636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.545608817755065e-05,
+      "loss": 0.228,
+      "step": 5637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.543942931900557e-05,
+      "loss": 0.2424,
+      "step": 5638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5422772226371316e-05,
+      "loss": 0.2104,
+      "step": 5639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.540611690166809e-05,
+      "loss": 0.1967,
+      "step": 5640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.538946334691582e-05,
+      "loss": 0.2261,
+      "step": 5641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.53728115641343e-05,
+      "loss": 0.2992,
+      "step": 5642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5356161555343036e-05,
+      "loss": 0.2137,
+      "step": 5643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.533951332256136e-05,
+      "loss": 0.2197,
+      "step": 5644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.532286686780835e-05,
+      "loss": 0.305,
+      "step": 5645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.530622219310293e-05,
+      "loss": 0.1953,
+      "step": 5646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5289579300463746e-05,
+      "loss": 0.2676,
+      "step": 5647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5272938191909256e-05,
+      "loss": 0.1994,
+      "step": 5648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5256298869457715e-05,
+      "loss": 0.2,
+      "step": 5649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.523966133512713e-05,
+      "loss": 0.2352,
+      "step": 5650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.522302559093532e-05,
+      "loss": 0.2996,
+      "step": 5651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.520639163889986e-05,
+      "loss": 0.2934,
+      "step": 5652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.518975948103813e-05,
+      "loss": 0.2149,
+      "step": 5653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.517312911936727e-05,
+      "loss": 0.2281,
+      "step": 5654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.515650055590422e-05,
+      "loss": 0.2234,
+      "step": 5655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.513987379266571e-05,
+      "loss": 0.2341,
+      "step": 5656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.512324883166823e-05,
+      "loss": 0.2824,
+      "step": 5657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5106625674928065e-05,
+      "loss": 0.22,
+      "step": 5658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.509000432446128e-05,
+      "loss": 0.1884,
+      "step": 5659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5073384782283714e-05,
+      "loss": 0.284,
+      "step": 5660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.505676705041098e-05,
+      "loss": 0.2447,
+      "step": 5661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.50401511308585e-05,
+      "loss": 0.2134,
+      "step": 5662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.502353702564145e-05,
+      "loss": 0.2366,
+      "step": 5663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5006924736774795e-05,
+      "loss": 0.1994,
+      "step": 5664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.499031426627326e-05,
+      "loss": 0.2226,
+      "step": 5665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497370561615141e-05,
+      "loss": 0.2739,
+      "step": 5666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4957098788423525e-05,
+      "loss": 0.2369,
+      "step": 5667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.494049378510368e-05,
+      "loss": 0.2318,
+      "step": 5668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.492389060820574e-05,
+      "loss": 0.2447,
+      "step": 5669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.490728925974334e-05,
+      "loss": 0.2847,
+      "step": 5670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.48906897417299e-05,
+      "loss": 0.1912,
+      "step": 5671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.487409205617862e-05,
+      "loss": 0.3026,
+      "step": 5672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4857496205102474e-05,
+      "loss": 0.2871,
+      "step": 5673
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4840902190514196e-05,
+      "loss": 0.2034,
+      "step": 5674
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.482431001442633e-05,
+      "loss": 0.2162,
+      "step": 5675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.480771967885119e-05,
+      "loss": 0.2168,
+      "step": 5676
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.479113118580082e-05,
+      "loss": 0.2919,
+      "step": 5677
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4774544537287124e-05,
+      "loss": 0.2474,
+      "step": 5678
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.47579597353217e-05,
+      "loss": 0.2391,
+      "step": 5679
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.474137678191598e-05,
+      "loss": 0.2495,
+      "step": 5680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4724795679081137e-05,
+      "loss": 0.2577,
+      "step": 5681
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4708216428828134e-05,
+      "loss": 0.2765,
+      "step": 5682
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.469163903316772e-05,
+      "loss": 0.2854,
+      "step": 5683
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.467506349411041e-05,
+      "loss": 0.2716,
+      "step": 5684
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.465848981366647e-05,
+      "loss": 0.2362,
+      "step": 5685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.464191799384597e-05,
+      "loss": 0.2703,
+      "step": 5686
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.462534803665873e-05,
+      "loss": 0.2662,
+      "step": 5687
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.46087799441144e-05,
+      "loss": 0.2574,
+      "step": 5688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4592213718222336e-05,
+      "loss": 0.2641,
+      "step": 5689
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.457564936099168e-05,
+      "loss": 0.2487,
+      "step": 5690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.45590868744314e-05,
+      "loss": 0.2633,
+      "step": 5691
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.454252626055018e-05,
+      "loss": 0.213,
+      "step": 5692
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.452596752135651e-05,
+      "loss": 0.3339,
+      "step": 5693
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.450941065885861e-05,
+      "loss": 0.3341,
+      "step": 5694
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4492855675064524e-05,
+      "loss": 0.2236,
+      "step": 5695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.447630257198205e-05,
+      "loss": 0.1883,
+      "step": 5696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.445975135161873e-05,
+      "loss": 0.2107,
+      "step": 5697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4443202015981934e-05,
+      "loss": 0.3271,
+      "step": 5698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4426654567078754e-05,
+      "loss": 0.27,
+      "step": 5699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.441010900691608e-05,
+      "loss": 0.2743,
+      "step": 5700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.439356533750054e-05,
+      "loss": 0.3028,
+      "step": 5701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4377023560838576e-05,
+      "loss": 0.1915,
+      "step": 5702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.436048367893636e-05,
+      "loss": 0.2747,
+      "step": 5703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.434394569379988e-05,
+      "loss": 0.177,
+      "step": 5704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4327409607434856e-05,
+      "loss": 0.2149,
+      "step": 5705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.431087542184679e-05,
+      "loss": 0.2817,
+      "step": 5706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.429434313904094e-05,
+      "loss": 0.2034,
+      "step": 5707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4277812761022374e-05,
+      "loss": 0.2425,
+      "step": 5708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.426128428979588e-05,
+      "loss": 0.2549,
+      "step": 5709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.424475772736605e-05,
+      "loss": 0.2163,
+      "step": 5710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.422823307573722e-05,
+      "loss": 0.3019,
+      "step": 5711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4211710336913525e-05,
+      "loss": 0.3226,
+      "step": 5712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.419518951289883e-05,
+      "loss": 0.2541,
+      "step": 5713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.417867060569677e-05,
+      "loss": 0.2045,
+      "step": 5714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.416215361731081e-05,
+      "loss": 0.243,
+      "step": 5715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.41456385497441e-05,
+      "loss": 0.2334,
+      "step": 5716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4129125404999615e-05,
+      "loss": 0.2478,
+      "step": 5717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.411261418508005e-05,
+      "loss": 0.2255,
+      "step": 5718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4096104891987904e-05,
+      "loss": 0.248,
+      "step": 5719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4079597527725416e-05,
+      "loss": 0.3085,
+      "step": 5720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.406309209429463e-05,
+      "loss": 0.2387,
+      "step": 5721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.404658859369731e-05,
+      "loss": 0.2656,
+      "step": 5722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.403008702793502e-05,
+      "loss": 0.2261,
+      "step": 5723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.401358739900907e-05,
+      "loss": 0.2994,
+      "step": 5724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.399708970892054e-05,
+      "loss": 0.3295,
+      "step": 5725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.398059395967028e-05,
+      "loss": 0.2504,
+      "step": 5726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3964100153258884e-05,
+      "loss": 0.2563,
+      "step": 5727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.394760829168674e-05,
+      "loss": 0.2698,
+      "step": 5728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.393111837695399e-05,
+      "loss": 0.2699,
+      "step": 5729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.391463041106051e-05,
+      "loss": 0.2484,
+      "step": 5730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.389814439600601e-05,
+      "loss": 0.2163,
+      "step": 5731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.388166033378989e-05,
+      "loss": 0.3068,
+      "step": 5732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.386517822641135e-05,
+      "loss": 0.1677,
+      "step": 5733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3848698075869345e-05,
+      "loss": 0.1608,
+      "step": 5734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3832219884162585e-05,
+      "loss": 0.233,
+      "step": 5735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3815743653289556e-05,
+      "loss": 0.2162,
+      "step": 5736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.37992693852485e-05,
+      "loss": 0.2668,
+      "step": 5737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3782797082037434e-05,
+      "loss": 0.2032,
+      "step": 5738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.376632674565411e-05,
+      "loss": 0.3354,
+      "step": 5739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3749858378096045e-05,
+      "loss": 0.3004,
+      "step": 5740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.373339198136055e-05,
+      "loss": 0.157,
+      "step": 5741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.371692755744468e-05,
+      "loss": 0.1548,
+      "step": 5742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3700465108345215e-05,
+      "loss": 0.2429,
+      "step": 5743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.368400463605876e-05,
+      "loss": 0.2489,
+      "step": 5744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.366754614258163e-05,
+      "loss": 0.2158,
+      "step": 5745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.365108962990991e-05,
+      "loss": 0.1917,
+      "step": 5746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.363463510003946e-05,
+      "loss": 0.1964,
+      "step": 5747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.361818255496589e-05,
+      "loss": 0.2294,
+      "step": 5748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.360173199668458e-05,
+      "loss": 0.2583,
+      "step": 5749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3585283427190664e-05,
+      "loss": 0.2362,
+      "step": 5750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.356883684847899e-05,
+      "loss": 0.2552,
+      "step": 5751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.355239226254423e-05,
+      "loss": 0.2299,
+      "step": 5752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.35359496713808e-05,
+      "loss": 0.2304,
+      "step": 5753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.351950907698285e-05,
+      "loss": 0.2412,
+      "step": 5754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.350307048134429e-05,
+      "loss": 0.3178,
+      "step": 5755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3486633886458817e-05,
+      "loss": 0.2829,
+      "step": 5756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.347019929431986e-05,
+      "loss": 0.2505,
+      "step": 5757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.345376670692062e-05,
+      "loss": 0.2989,
+      "step": 5758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.343733612625404e-05,
+      "loss": 0.2271,
+      "step": 5759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.342090755431281e-05,
+      "loss": 0.2334,
+      "step": 5760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3404480993089414e-05,
+      "loss": 0.3182,
+      "step": 5761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3388056444576063e-05,
+      "loss": 0.3405,
+      "step": 5762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.337163391076473e-05,
+      "loss": 0.2817,
+      "step": 5763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.335521339364716e-05,
+      "loss": 0.2752,
+      "step": 5764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.333879489521483e-05,
+      "loss": 0.2471,
+      "step": 5765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.332237841745898e-05,
+      "loss": 0.2614,
+      "step": 5766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.33059639623706e-05,
+      "loss": 0.296,
+      "step": 5767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3289551531940445e-05,
+      "loss": 0.3039,
+      "step": 5768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3273141128159e-05,
+      "loss": 0.1769,
+      "step": 5769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.325673275301656e-05,
+      "loss": 0.1935,
+      "step": 5770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.324032640850313e-05,
+      "loss": 0.3635,
+      "step": 5771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.322392209660845e-05,
+      "loss": 0.2638,
+      "step": 5772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.320751981932207e-05,
+      "loss": 0.2597,
+      "step": 5773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.319111957863326e-05,
+      "loss": 0.16,
+      "step": 5774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3174721376531024e-05,
+      "loss": 0.2451,
+      "step": 5775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.315832521500416e-05,
+      "loss": 0.2097,
+      "step": 5776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3141931096041203e-05,
+      "loss": 0.2301,
+      "step": 5777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3125539021630436e-05,
+      "loss": 0.2548,
+      "step": 5778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3109148993759886e-05,
+      "loss": 0.2864,
+      "step": 5779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.309276101441734e-05,
+      "loss": 0.3778,
+      "step": 5780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.307637508559034e-05,
+      "loss": 0.2248,
+      "step": 5781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.30599912092662e-05,
+      "loss": 0.2802,
+      "step": 5782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.304360938743193e-05,
+      "loss": 0.2478,
+      "step": 5783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.302722962207433e-05,
+      "loss": 0.2553,
+      "step": 5784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3010851915179945e-05,
+      "loss": 0.1964,
+      "step": 5785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.299447626873508e-05,
+      "loss": 0.2356,
+      "step": 5786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.297810268472577e-05,
+      "loss": 0.3286,
+      "step": 5787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.296173116513781e-05,
+      "loss": 0.2064,
+      "step": 5788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.294536171195673e-05,
+      "loss": 0.2067,
+      "step": 5789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.292899432716785e-05,
+      "loss": 0.2692,
+      "step": 5790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2912629012756194e-05,
+      "loss": 0.2473,
+      "step": 5791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.289626577070655e-05,
+      "loss": 0.2209,
+      "step": 5792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.287990460300346e-05,
+      "loss": 0.3158,
+      "step": 5793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.286354551163122e-05,
+      "loss": 0.2195,
+      "step": 5794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2847188498573866e-05,
+      "loss": 0.2424,
+      "step": 5795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2830833565815164e-05,
+      "loss": 0.2225,
+      "step": 5796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.281448071533867e-05,
+      "loss": 0.2193,
+      "step": 5797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.279812994912765e-05,
+      "loss": 0.2589,
+      "step": 5798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.278178126916515e-05,
+      "loss": 0.1995,
+      "step": 5799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.27654346774339e-05,
+      "loss": 0.2519,
+      "step": 5800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2749090175916465e-05,
+      "loss": 0.2588,
+      "step": 5801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2732747766595074e-05,
+      "loss": 0.2735,
+      "step": 5802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.271640745145176e-05,
+      "loss": 0.2904,
+      "step": 5803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.270006923246829e-05,
+      "loss": 0.2701,
+      "step": 5804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.268373311162614e-05,
+      "loss": 0.2843,
+      "step": 5805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2667399090906594e-05,
+      "loss": 0.2072,
+      "step": 5806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.265106717229063e-05,
+      "loss": 0.2557,
+      "step": 5807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.263473735775899e-05,
+      "loss": 0.2478,
+      "step": 5808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2618409649292154e-05,
+      "loss": 0.1523,
+      "step": 5809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2602084048870366e-05,
+      "loss": 0.2692,
+      "step": 5810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.258576055847359e-05,
+      "loss": 0.2045,
+      "step": 5811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.256943918008154e-05,
+      "loss": 0.2155,
+      "step": 5812
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.255311991567369e-05,
+      "loss": 0.2083,
+      "step": 5813
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.253680276722924e-05,
+      "loss": 0.277,
+      "step": 5814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.252048773672715e-05,
+      "loss": 0.2959,
+      "step": 5815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.250417482614609e-05,
+      "loss": 0.2451,
+      "step": 5816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.248786403746452e-05,
+      "loss": 0.2431,
+      "step": 5817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.247155537266058e-05,
+      "loss": 0.1807,
+      "step": 5818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.245524883371223e-05,
+      "loss": 0.2447,
+      "step": 5819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.243894442259712e-05,
+      "loss": 0.2811,
+      "step": 5820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.242264214129265e-05,
+      "loss": 0.2964,
+      "step": 5821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.240634199177595e-05,
+      "loss": 0.2455,
+      "step": 5822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.239004397602394e-05,
+      "loss": 0.2047,
+      "step": 5823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.237374809601325e-05,
+      "loss": 0.1584,
+      "step": 5824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2357454353720204e-05,
+      "loss": 0.1903,
+      "step": 5825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.234116275112096e-05,
+      "loss": 0.2796,
+      "step": 5826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2324873290191345e-05,
+      "loss": 0.2764,
+      "step": 5827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2308585972906966e-05,
+      "loss": 0.233,
+      "step": 5828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.229230080124313e-05,
+      "loss": 0.2689,
+      "step": 5829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2276017777174935e-05,
+      "loss": 0.3087,
+      "step": 5830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.225973690267718e-05,
+      "loss": 0.2802,
+      "step": 5831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.224345817972442e-05,
+      "loss": 0.2154,
+      "step": 5832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2227181610290935e-05,
+      "loss": 0.2257,
+      "step": 5833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.221090719635075e-05,
+      "loss": 0.2478,
+      "step": 5834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.219463493987763e-05,
+      "loss": 0.2742,
+      "step": 5835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.21783648428451e-05,
+      "loss": 0.3226,
+      "step": 5836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.216209690722638e-05,
+      "loss": 0.3015,
+      "step": 5837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.214583113499446e-05,
+      "loss": 0.2946,
+      "step": 5838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.212956752812206e-05,
+      "loss": 0.2397,
+      "step": 5839
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.211330608858164e-05,
+      "loss": 0.2618,
+      "step": 5840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.209704681834539e-05,
+      "loss": 0.2997,
+      "step": 5841
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.208078971938522e-05,
+      "loss": 0.1802,
+      "step": 5842
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.206453479367283e-05,
+      "loss": 0.2503,
+      "step": 5843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.20482820431796e-05,
+      "loss": 0.271,
+      "step": 5844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2032031469876666e-05,
+      "loss": 0.2835,
+      "step": 5845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.201578307573492e-05,
+      "loss": 0.2031,
+      "step": 5846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1999536862724974e-05,
+      "loss": 0.2571,
+      "step": 5847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.198329283281716e-05,
+      "loss": 0.2424,
+      "step": 5848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1967050987981565e-05,
+      "loss": 0.206,
+      "step": 5849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1950811330188e-05,
+      "loss": 0.2497,
+      "step": 5850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.193457386140601e-05,
+      "loss": 0.2161,
+      "step": 5851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.19183385836049e-05,
+      "loss": 0.2449,
+      "step": 5852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.190210549875368e-05,
+      "loss": 0.2024,
+      "step": 5853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.188587460882111e-05,
+      "loss": 0.2116,
+      "step": 5854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.186964591577566e-05,
+      "loss": 0.3273,
+      "step": 5855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1853419421585565e-05,
+      "loss": 0.274,
+      "step": 5856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1837195128218786e-05,
+      "loss": 0.3269,
+      "step": 5857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.182097303764299e-05,
+      "loss": 0.2395,
+      "step": 5858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.180475315182563e-05,
+      "loss": 0.3141,
+      "step": 5859
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.178853547273383e-05,
+      "loss": 0.2373,
+      "step": 5860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.177232000233448e-05,
+      "loss": 0.2303,
+      "step": 5861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.175610674259421e-05,
+      "loss": 0.206,
+      "step": 5862
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.173989569547935e-05,
+      "loss": 0.2267,
+      "step": 5863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.172368686295601e-05,
+      "loss": 0.3087,
+      "step": 5864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.170748024698998e-05,
+      "loss": 0.2086,
+      "step": 5865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1691275849546817e-05,
+      "loss": 0.1971,
+      "step": 5866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1675073672591774e-05,
+      "loss": 0.2746,
+      "step": 5867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1658873718089876e-05,
+      "loss": 0.2773,
+      "step": 5868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.164267598800585e-05,
+      "loss": 0.2498,
+      "step": 5869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.162648048430418e-05,
+      "loss": 0.2524,
+      "step": 5870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1610287208949024e-05,
+      "loss": 0.2321,
+      "step": 5871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1594096163904343e-05,
+      "loss": 0.2592,
+      "step": 5872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.157790735113378e-05,
+      "loss": 0.2295,
+      "step": 5873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1561720772600735e-05,
+      "loss": 0.2278,
+      "step": 5874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1545536430268285e-05,
+      "loss": 0.1985,
+      "step": 5875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.15293543260993e-05,
+      "loss": 0.2658,
+      "step": 5876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.151317446205635e-05,
+      "loss": 0.1821,
+      "step": 5877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.149699684010172e-05,
+      "loss": 0.1972,
+      "step": 5878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148082146219746e-05,
+      "loss": 0.1721,
+      "step": 5879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.146464833030531e-05,
+      "loss": 0.2074,
+      "step": 5880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.144847744638676e-05,
+      "loss": 0.1954,
+      "step": 5881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.143230881240301e-05,
+      "loss": 0.2093,
+      "step": 5882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.141614243031501e-05,
+      "loss": 0.193,
+      "step": 5883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1399978302083395e-05,
+      "loss": 0.2085,
+      "step": 5884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.138381642966859e-05,
+      "loss": 0.2381,
+      "step": 5885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.13676568150307e-05,
+      "loss": 0.1767,
+      "step": 5886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.135149946012957e-05,
+      "loss": 0.203,
+      "step": 5887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.133534436692476e-05,
+      "loss": 0.1483,
+      "step": 5888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1319191537375576e-05,
+      "loss": 0.2547,
+      "step": 5889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.130304097344103e-05,
+      "loss": 0.2597,
+      "step": 5890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.128689267707988e-05,
+      "loss": 0.2004,
+      "step": 5891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.127074665025059e-05,
+      "loss": 0.2626,
+      "step": 5892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.125460289491136e-05,
+      "loss": 0.2124,
+      "step": 5893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.123846141302009e-05,
+      "loss": 0.161,
+      "step": 5894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.122232220653444e-05,
+      "loss": 0.1645,
+      "step": 5895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1206185277411786e-05,
+      "loss": 0.2549,
+      "step": 5896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1190050627609226e-05,
+      "loss": 0.2996,
+      "step": 5897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1173918259083545e-05,
+      "loss": 0.2153,
+      "step": 5898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.11577881737913e-05,
+      "loss": 0.1957,
+      "step": 5899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.114166037368874e-05,
+      "loss": 0.2644,
+      "step": 5900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1125534860731874e-05,
+      "loss": 0.1651,
+      "step": 5901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.11094116368764e-05,
+      "loss": 0.3241,
+      "step": 5902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.109329070407774e-05,
+      "loss": 0.2512,
+      "step": 5903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.107717206429105e-05,
+      "loss": 0.2217,
+      "step": 5904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.10610557194712e-05,
+      "loss": 0.2229,
+      "step": 5905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.10449416715728e-05,
+      "loss": 0.3017,
+      "step": 5906
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.102882992255016e-05,
+      "loss": 0.1756,
+      "step": 5907
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.10127204743573e-05,
+      "loss": 0.2114,
+      "step": 5908
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0996613328948006e-05,
+      "loss": 0.1925,
+      "step": 5909
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.098050848827575e-05,
+      "loss": 0.2678,
+      "step": 5910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.096440595429372e-05,
+      "loss": 0.2263,
+      "step": 5911
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.094830572895485e-05,
+      "loss": 0.2406,
+      "step": 5912
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.093220781421178e-05,
+      "loss": 0.2552,
+      "step": 5913
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.091611221201688e-05,
+      "loss": 0.2537,
+      "step": 5914
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0900018924322205e-05,
+      "loss": 0.2809,
+      "step": 5915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.088392795307956e-05,
+      "loss": 0.2196,
+      "step": 5916
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.086783930024047e-05,
+      "loss": 0.2301,
+      "step": 5917
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.085175296775617e-05,
+      "loss": 0.246,
+      "step": 5918
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.083566895757764e-05,
+      "loss": 0.1726,
+      "step": 5919
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0819587271655504e-05,
+      "loss": 0.2684,
+      "step": 5920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.080350791194019e-05,
+      "loss": 0.2734,
+      "step": 5921
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0787430880381805e-05,
+      "loss": 0.2726,
+      "step": 5922
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0771356178930175e-05,
+      "loss": 0.2843,
+      "step": 5923
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.075528380953483e-05,
+      "loss": 0.224,
+      "step": 5924
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.073921377414505e-05,
+      "loss": 0.2285,
+      "step": 5925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.072314607470982e-05,
+      "loss": 0.2667,
+      "step": 5926
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.070708071317781e-05,
+      "loss": 0.2799,
+      "step": 5927
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0691017691497445e-05,
+      "loss": 0.326,
+      "step": 5928
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.067495701161686e-05,
+      "loss": 0.2223,
+      "step": 5929
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.065889867548391e-05,
+      "loss": 0.2496,
+      "step": 5930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.064284268504613e-05,
+      "loss": 0.2423,
+      "step": 5931
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.06267890422508e-05,
+      "loss": 0.2283,
+      "step": 5932
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.061073774904491e-05,
+      "loss": 0.2741,
+      "step": 5933
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.059468880737519e-05,
+      "loss": 0.2212,
+      "step": 5934
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.057864221918804e-05,
+      "loss": 0.2251,
+      "step": 5935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0562597986429596e-05,
+      "loss": 0.2549,
+      "step": 5936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0546556111045704e-05,
+      "loss": 0.2107,
+      "step": 5937
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.053051659498195e-05,
+      "loss": 0.2573,
+      "step": 5938
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.051447944018359e-05,
+      "loss": 0.2172,
+      "step": 5939
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0498444648595614e-05,
+      "loss": 0.189,
+      "step": 5940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.048241222216275e-05,
+      "loss": 0.2791,
+      "step": 5941
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.04663821628294e-05,
+      "loss": 0.2432,
+      "step": 5942
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0450354472539698e-05,
+      "loss": 0.2841,
+      "step": 5943
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0434329153237473e-05,
+      "loss": 0.2513,
+      "step": 5944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.041830620686631e-05,
+      "loss": 0.2533,
+      "step": 5945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0402285635369464e-05,
+      "loss": 0.2478,
+      "step": 5946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0386267440689908e-05,
+      "loss": 0.2355,
+      "step": 5947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0370251624770342e-05,
+      "loss": 0.1554,
+      "step": 5948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.035423818955316e-05,
+      "loss": 0.2723,
+      "step": 5949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0338227136980475e-05,
+      "loss": 0.2301,
+      "step": 5950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.032221846899413e-05,
+      "loss": 0.2814,
+      "step": 5951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0306212187535653e-05,
+      "loss": 0.2732,
+      "step": 5952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0290208294546273e-05,
+      "loss": 0.2095,
+      "step": 5953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0274206791966985e-05,
+      "loss": 0.2131,
+      "step": 5954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0258207681738432e-05,
+      "loss": 0.2446,
+      "step": 5955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0242210965800997e-05,
+      "loss": 0.2315,
+      "step": 5956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0226216646094756e-05,
+      "loss": 0.259,
+      "step": 5957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.021022472455952e-05,
+      "loss": 0.2284,
+      "step": 5958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0194235203134803e-05,
+      "loss": 0.2044,
+      "step": 5959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0178248083759785e-05,
+      "loss": 0.2479,
+      "step": 5960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0162263368373422e-05,
+      "loss": 0.1742,
+      "step": 5961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0146281058914334e-05,
+      "loss": 0.2177,
+      "step": 5962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0130301157320874e-05,
+      "loss": 0.2377,
+      "step": 5963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0114323665531063e-05,
+      "loss": 0.2544,
+      "step": 5964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0098348585482665e-05,
+      "loss": 0.2456,
+      "step": 5965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0082375919113147e-05,
+      "loss": 0.1398,
+      "step": 5966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0066405668359686e-05,
+      "loss": 0.2218,
+      "step": 5967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0050437835159152e-05,
+      "loss": 0.2223,
+      "step": 5968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0034472421448133e-05,
+      "loss": 0.2333,
+      "step": 5969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0018509429162912e-05,
+      "loss": 0.2287,
+      "step": 5970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0002548860239493e-05,
+      "loss": 0.2744,
+      "step": 5971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.998659071661358e-05,
+      "loss": 0.2728,
+      "step": 5972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.997063500022057e-05,
+      "loss": 0.18,
+      "step": 5973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.99546817129956e-05,
+      "loss": 0.1948,
+      "step": 5974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.993873085687347e-05,
+      "loss": 0.2598,
+      "step": 5975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9922782433788715e-05,
+      "loss": 0.2294,
+      "step": 5976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.990683644567556e-05,
+      "loss": 0.2195,
+      "step": 5977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9890892894467938e-05,
+      "loss": 0.2819,
+      "step": 5978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9874951782099507e-05,
+      "loss": 0.2344,
+      "step": 5979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9859013110503585e-05,
+      "loss": 0.2242,
+      "step": 5980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9843076881613237e-05,
+      "loss": 0.2289,
+      "step": 5981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9827143097361198e-05,
+      "loss": 0.1089,
+      "step": 5982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9811211759679924e-05,
+      "loss": 0.2871,
+      "step": 5983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9795282870501594e-05,
+      "loss": 0.2165,
+      "step": 5984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9779356431758055e-05,
+      "loss": 0.2017,
+      "step": 5985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9763432445380855e-05,
+      "loss": 0.2697,
+      "step": 5986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.974751091330129e-05,
+      "loss": 0.1778,
+      "step": 5987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.973159183745031e-05,
+      "loss": 0.2003,
+      "step": 5988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9715675219758598e-05,
+      "loss": 0.2314,
+      "step": 5989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9699761062156507e-05,
+      "loss": 0.2813,
+      "step": 5990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9683849366574134e-05,
+      "loss": 0.1815,
+      "step": 5991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9667940134941246e-05,
+      "loss": 0.2673,
+      "step": 5992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9652033369187303e-05,
+      "loss": 0.2518,
+      "step": 5993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9636129071241514e-05,
+      "loss": 0.192,
+      "step": 5994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9620227243032738e-05,
+      "loss": 0.1439,
+      "step": 5995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9604327886489576e-05,
+      "loss": 0.2532,
+      "step": 5996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9588431003540273e-05,
+      "loss": 0.2267,
+      "step": 5997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.957253659611282e-05,
+      "loss": 0.2859,
+      "step": 5998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9556644666134904e-05,
+      "loss": 0.2242,
+      "step": 5999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.95407552155339e-05,
+      "loss": 0.1181,
+      "step": 6000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.952486824623688e-05,
+      "loss": 0.132,
+      "step": 6001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.950898376017064e-05,
+      "loss": 0.2445,
+      "step": 6002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.949310175926162e-05,
+      "loss": 0.2759,
+      "step": 6003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9477222245436033e-05,
+      "loss": 0.2572,
+      "step": 6004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9461345220619735e-05,
+      "loss": 0.2362,
+      "step": 6005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9445470686738287e-05,
+      "loss": 0.2534,
+      "step": 6006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.942959864571697e-05,
+      "loss": 0.2367,
+      "step": 6007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.941372909948076e-05,
+      "loss": 0.2798,
+      "step": 6008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9397862049954305e-05,
+      "loss": 0.2654,
+      "step": 6009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9381997499061963e-05,
+      "loss": 0.2937,
+      "step": 6010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9366135448727805e-05,
+      "loss": 0.2917,
+      "step": 6011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.935027590087559e-05,
+      "loss": 0.2288,
+      "step": 6012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9334418857428746e-05,
+      "loss": 0.1714,
+      "step": 6013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9318564320310444e-05,
+      "loss": 0.3183,
+      "step": 6014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.93027122914435e-05,
+      "loss": 0.2042,
+      "step": 6015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9286862772750477e-05,
+      "loss": 0.3127,
+      "step": 6016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.927101576615361e-05,
+      "loss": 0.247,
+      "step": 6017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.925517127357482e-05,
+      "loss": 0.2164,
+      "step": 6018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9239329296935723e-05,
+      "loss": 0.2384,
+      "step": 6019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.922348983815767e-05,
+      "loss": 0.2765,
+      "step": 6020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9207652899161642e-05,
+      "loss": 0.1758,
+      "step": 6021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.919181848186837e-05,
+      "loss": 0.303,
+      "step": 6022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9175986588198244e-05,
+      "loss": 0.2654,
+      "step": 6023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9160157220071365e-05,
+      "loss": 0.2372,
+      "step": 6024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9144330379407548e-05,
+      "loss": 0.2348,
+      "step": 6025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9128506068126237e-05,
+      "loss": 0.2927,
+      "step": 6026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9112684288146642e-05,
+      "loss": 0.265,
+      "step": 6027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.909686504138763e-05,
+      "loss": 0.2708,
+      "step": 6028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.908104832976773e-05,
+      "loss": 0.3721,
+      "step": 6029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9065234155205222e-05,
+      "loss": 0.221,
+      "step": 6030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9049422519618074e-05,
+      "loss": 0.1949,
+      "step": 6031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9033613424923895e-05,
+      "loss": 0.2128,
+      "step": 6032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9017806873040042e-05,
+      "loss": 0.2539,
+      "step": 6033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9002002865883516e-05,
+      "loss": 0.2647,
+      "step": 6034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.898620140537104e-05,
+      "loss": 0.3068,
+      "step": 6035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8970402493419042e-05,
+      "loss": 0.2098,
+      "step": 6036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8954606131943578e-05,
+      "loss": 0.2745,
+      "step": 6037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8938812322860466e-05,
+      "loss": 0.235,
+      "step": 6038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8923021068085187e-05,
+      "loss": 0.27,
+      "step": 6039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.890723236953289e-05,
+      "loss": 0.176,
+      "step": 6040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8891446229118436e-05,
+      "loss": 0.2894,
+      "step": 6041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8875662648756407e-05,
+      "loss": 0.2209,
+      "step": 6042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8859881630360997e-05,
+      "loss": 0.3006,
+      "step": 6043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.884410317584614e-05,
+      "loss": 0.202,
+      "step": 6044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.882832728712551e-05,
+      "loss": 0.2426,
+      "step": 6045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8812553966112323e-05,
+      "loss": 0.2708,
+      "step": 6046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8796783214719615e-05,
+      "loss": 0.2163,
+      "step": 6047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8781015034860094e-05,
+      "loss": 0.1976,
+      "step": 6048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8765249428446074e-05,
+      "loss": 0.236,
+      "step": 6049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8749486397389646e-05,
+      "loss": 0.1741,
+      "step": 6050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8733725943602575e-05,
+      "loss": 0.1709,
+      "step": 6051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.871796806899625e-05,
+      "loss": 0.3003,
+      "step": 6052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.870221277548183e-05,
+      "loss": 0.2067,
+      "step": 6053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8686460064970078e-05,
+      "loss": 0.1772,
+      "step": 6054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8670709939371527e-05,
+      "loss": 0.212,
+      "step": 6055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8654962400596364e-05,
+      "loss": 0.2285,
+      "step": 6056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.863921745055441e-05,
+      "loss": 0.2023,
+      "step": 6057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.862347509115525e-05,
+      "loss": 0.3728,
+      "step": 6058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.860773532430814e-05,
+      "loss": 0.3094,
+      "step": 6059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.859199815192195e-05,
+      "loss": 0.1999,
+      "step": 6060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.857626357590535e-05,
+      "loss": 0.221,
+      "step": 6061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8560531598166607e-05,
+      "loss": 0.2472,
+      "step": 6062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8544802220613675e-05,
+      "loss": 0.2019,
+      "step": 6063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.852907544515424e-05,
+      "loss": 0.2296,
+      "step": 6064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8513351273695678e-05,
+      "loss": 0.2897,
+      "step": 6065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8497629708144967e-05,
+      "loss": 0.2531,
+      "step": 6066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8481910750408856e-05,
+      "loss": 0.2064,
+      "step": 6067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8466194402393775e-05,
+      "loss": 0.2871,
+      "step": 6068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.845048066600574e-05,
+      "loss": 0.244,
+      "step": 6069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8434769543150563e-05,
+      "loss": 0.204,
+      "step": 6070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.84190610357337e-05,
+      "loss": 0.2351,
+      "step": 6071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8403355145660248e-05,
+      "loss": 0.1623,
+      "step": 6072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8387651874835064e-05,
+      "loss": 0.203,
+      "step": 6073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8371951225162606e-05,
+      "loss": 0.2279,
+      "step": 6074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8356253198547077e-05,
+      "loss": 0.2456,
+      "step": 6075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8340557796892354e-05,
+      "loss": 0.2528,
+      "step": 6076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.832486502210193e-05,
+      "loss": 0.3341,
+      "step": 6077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.830917487607909e-05,
+      "loss": 0.2413,
+      "step": 6078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.82934873607267e-05,
+      "loss": 0.2145,
+      "step": 6079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8277802477947367e-05,
+      "loss": 0.1905,
+      "step": 6080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8262120229643334e-05,
+      "loss": 0.246,
+      "step": 6081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8246440617716585e-05,
+      "loss": 0.2433,
+      "step": 6082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8230763644068708e-05,
+      "loss": 0.1969,
+      "step": 6083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.821508931060104e-05,
+      "loss": 0.2318,
+      "step": 6084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8199417619214576e-05,
+      "loss": 0.2516,
+      "step": 6085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8183748571809944e-05,
+      "loss": 0.2147,
+      "step": 6086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8168082170287524e-05,
+      "loss": 0.2112,
+      "step": 6087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8152418416547355e-05,
+      "loss": 0.1941,
+      "step": 6088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8136757312489103e-05,
+      "loss": 0.2684,
+      "step": 6089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8121098860012174e-05,
+      "loss": 0.2764,
+      "step": 6090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8105443061015647e-05,
+      "loss": 0.2219,
+      "step": 6091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8089789917398224e-05,
+      "loss": 0.1694,
+      "step": 6092
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8074139431058356e-05,
+      "loss": 0.2613,
+      "step": 6093
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8058491603894133e-05,
+      "loss": 0.3044,
+      "step": 6094
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8042846437803305e-05,
+      "loss": 0.2342,
+      "step": 6095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8027203934683334e-05,
+      "loss": 0.2067,
+      "step": 6096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8011564096431377e-05,
+      "loss": 0.2407,
+      "step": 6097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7995926924944188e-05,
+      "loss": 0.1943,
+      "step": 6098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.798029242211828e-05,
+      "loss": 0.2572,
+      "step": 6099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7964660589849835e-05,
+      "loss": 0.2538,
+      "step": 6100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7949031430034627e-05,
+      "loss": 0.2639,
+      "step": 6101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.793340494456822e-05,
+      "loss": 0.2561,
+      "step": 6102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7917781135345755e-05,
+      "loss": 0.1855,
+      "step": 6103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7902160004262123e-05,
+      "loss": 0.2669,
+      "step": 6104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7886541553211865e-05,
+      "loss": 0.2112,
+      "step": 6105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7870925784089153e-05,
+      "loss": 0.2078,
+      "step": 6106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7855312698787904e-05,
+      "loss": 0.2265,
+      "step": 6107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7839702299201693e-05,
+      "loss": 0.2967,
+      "step": 6108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7824094587223707e-05,
+      "loss": 0.2098,
+      "step": 6109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7808489564746908e-05,
+      "loss": 0.2754,
+      "step": 6110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7792887233663844e-05,
+      "loss": 0.1535,
+      "step": 6111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7777287595866764e-05,
+      "loss": 0.22,
+      "step": 6112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7761690653247606e-05,
+      "loss": 0.2663,
+      "step": 6113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7746096407698002e-05,
+      "loss": 0.2585,
+      "step": 6114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7730504861109187e-05,
+      "loss": 0.2992,
+      "step": 6115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7714916015372118e-05,
+      "loss": 0.266,
+      "step": 6116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7699329872377437e-05,
+      "loss": 0.2322,
+      "step": 6117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7683746434015413e-05,
+      "loss": 0.2158,
+      "step": 6118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7668165702176007e-05,
+      "loss": 0.2291,
+      "step": 6119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7652587678748898e-05,
+      "loss": 0.3179,
+      "step": 6120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.763701236562334e-05,
+      "loss": 0.2142,
+      "step": 6121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7621439764688358e-05,
+      "loss": 0.216,
+      "step": 6122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7605869877832557e-05,
+      "loss": 0.1813,
+      "step": 6123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7590302706944284e-05,
+      "loss": 0.2155,
+      "step": 6124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7574738253911536e-05,
+      "loss": 0.2088,
+      "step": 6125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.755917652062196e-05,
+      "loss": 0.1931,
+      "step": 6126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.754361750896291e-05,
+      "loss": 0.23,
+      "step": 6127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7528061220821372e-05,
+      "loss": 0.2677,
+      "step": 6128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7512507658083997e-05,
+      "loss": 0.3621,
+      "step": 6129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7496956822637156e-05,
+      "loss": 0.1703,
+      "step": 6130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7481408716366863e-05,
+      "loss": 0.2747,
+      "step": 6131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7465863341158772e-05,
+      "loss": 0.2589,
+      "step": 6132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7450320698898237e-05,
+      "loss": 0.236,
+      "step": 6133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7434780791470306e-05,
+      "loss": 0.216,
+      "step": 6134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7419243620759623e-05,
+      "loss": 0.2122,
+      "step": 6135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7403709188650563e-05,
+      "loss": 0.2972,
+      "step": 6136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7388177497027157e-05,
+      "loss": 0.2138,
+      "step": 6137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.737264854777306e-05,
+      "loss": 0.2986,
+      "step": 6138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7357122342771645e-05,
+      "loss": 0.1978,
+      "step": 6139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7341598883905962e-05,
+      "loss": 0.2213,
+      "step": 6140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7326078173058662e-05,
+      "loss": 0.1924,
+      "step": 6141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7310560212112134e-05,
+      "loss": 0.2217,
+      "step": 6142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.729504500294838e-05,
+      "loss": 0.1759,
+      "step": 6143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7279532547449082e-05,
+      "loss": 0.2621,
+      "step": 6144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.726402284749561e-05,
+      "loss": 0.1815,
+      "step": 6145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7248515904968997e-05,
+      "loss": 0.2496,
+      "step": 6146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.72330117217499e-05,
+      "loss": 0.2094,
+      "step": 6147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7217510299718708e-05,
+      "loss": 0.2341,
+      "step": 6148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7202011640755398e-05,
+      "loss": 0.2001,
+      "step": 6149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.718651574673968e-05,
+      "loss": 0.2467,
+      "step": 6150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.717102261955091e-05,
+      "loss": 0.2223,
+      "step": 6151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7155532261068062e-05,
+      "loss": 0.214,
+      "step": 6152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.714004467316984e-05,
+      "loss": 0.236,
+      "step": 6153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.71245598577346e-05,
+      "loss": 0.2789,
+      "step": 6154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7109077816640295e-05,
+      "loss": 0.3451,
+      "step": 6155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7093598551764626e-05,
+      "loss": 0.3061,
+      "step": 6156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7078122064984935e-05,
+      "loss": 0.206,
+      "step": 6157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7062648358178177e-05,
+      "loss": 0.2604,
+      "step": 6158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.704717743322104e-05,
+      "loss": 0.2395,
+      "step": 6159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.703170929198986e-05,
+      "loss": 0.1952,
+      "step": 6160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7016243936360562e-05,
+      "loss": 0.2432,
+      "step": 6161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7000781368208816e-05,
+      "loss": 0.2582,
+      "step": 6162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.698532158940994e-05,
+      "loss": 0.1924,
+      "step": 6163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.696986460183888e-05,
+      "loss": 0.3118,
+      "step": 6164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.695441040737028e-05,
+      "loss": 0.1434,
+      "step": 6165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.693895900787844e-05,
+      "loss": 0.2439,
+      "step": 6166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6923510405237274e-05,
+      "loss": 0.2092,
+      "step": 6167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6908064601320438e-05,
+      "loss": 0.2281,
+      "step": 6168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6892621598001156e-05,
+      "loss": 0.1915,
+      "step": 6169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6877181397152396e-05,
+      "loss": 0.2586,
+      "step": 6170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6861744000646755e-05,
+      "loss": 0.2417,
+      "step": 6171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.684630941035645e-05,
+      "loss": 0.2171,
+      "step": 6172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6830877628153428e-05,
+      "loss": 0.1852,
+      "step": 6173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.681544865590926e-05,
+      "loss": 0.2236,
+      "step": 6174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6800022495495142e-05,
+      "loss": 0.2501,
+      "step": 6175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.678459914878201e-05,
+      "loss": 0.2467,
+      "step": 6176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6769178617640383e-05,
+      "loss": 0.238,
+      "step": 6177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.675376090394046e-05,
+      "loss": 0.238,
+      "step": 6178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6738346009552116e-05,
+      "loss": 0.2369,
+      "step": 6179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6722933936344906e-05,
+      "loss": 0.2253,
+      "step": 6180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6707524686187964e-05,
+      "loss": 0.1866,
+      "step": 6181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6692118260950143e-05,
+      "loss": 0.1456,
+      "step": 6182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.667671466249997e-05,
+      "loss": 0.1799,
+      "step": 6183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6661313892705554e-05,
+      "loss": 0.1561,
+      "step": 6184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6645915953434732e-05,
+      "loss": 0.2191,
+      "step": 6185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6630520846554973e-05,
+      "loss": 0.2061,
+      "step": 6186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.661512857393338e-05,
+      "loss": 0.1987,
+      "step": 6187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6599739137436763e-05,
+      "loss": 0.2549,
+      "step": 6188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6584352538931524e-05,
+      "loss": 0.1943,
+      "step": 6189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6568968780283764e-05,
+      "loss": 0.1962,
+      "step": 6190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6553587863359264e-05,
+      "loss": 0.1476,
+      "step": 6191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6538209790023406e-05,
+      "loss": 0.2623,
+      "step": 6192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6522834562141207e-05,
+      "loss": 0.3237,
+      "step": 6193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6507462181577426e-05,
+      "loss": 0.2395,
+      "step": 6194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6492092650196437e-05,
+      "loss": 0.2056,
+      "step": 6195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6476725969862227e-05,
+      "loss": 0.1814,
+      "step": 6196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6461362142438507e-05,
+      "loss": 0.205,
+      "step": 6197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.644600116978857e-05,
+      "loss": 0.2868,
+      "step": 6198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6430643053775416e-05,
+      "loss": 0.2224,
+      "step": 6199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6415287796261706e-05,
+      "loss": 0.2839,
+      "step": 6200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6399935399109687e-05,
+      "loss": 0.2518,
+      "step": 6201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6384585864181333e-05,
+      "loss": 0.2545,
+      "step": 6202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6369239193338257e-05,
+      "loss": 0.2384,
+      "step": 6203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6353895388441657e-05,
+      "loss": 0.203,
+      "step": 6204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.633855445135247e-05,
+      "loss": 0.2219,
+      "step": 6205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.632321638393126e-05,
+      "loss": 0.2304,
+      "step": 6206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.630788118803819e-05,
+      "loss": 0.2048,
+      "step": 6207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6292548865533174e-05,
+      "loss": 0.2713,
+      "step": 6208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.627721941827568e-05,
+      "loss": 0.2164,
+      "step": 6209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6261892848124862e-05,
+      "loss": 0.2581,
+      "step": 6210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6246569156939547e-05,
+      "loss": 0.2249,
+      "step": 6211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.623124834657823e-05,
+      "loss": 0.1728,
+      "step": 6212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.621593041889896e-05,
+      "loss": 0.2438,
+      "step": 6213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6200615375759542e-05,
+      "loss": 0.2481,
+      "step": 6214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6185303219017397e-05,
+      "loss": 0.2204,
+      "step": 6215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6169993950529548e-05,
+      "loss": 0.2883,
+      "step": 6216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6154687572152758e-05,
+      "loss": 0.2127,
+      "step": 6217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6139384085743345e-05,
+      "loss": 0.2341,
+      "step": 6218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6124083493157337e-05,
+      "loss": 0.1333,
+      "step": 6219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6108785796250417e-05,
+      "loss": 0.2272,
+      "step": 6220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6093490996877856e-05,
+      "loss": 0.1558,
+      "step": 6221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6078199096894634e-05,
+      "loss": 0.2803,
+      "step": 6222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.606291009815537e-05,
+      "loss": 0.1974,
+      "step": 6223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6047624002514292e-05,
+      "loss": 0.1873,
+      "step": 6224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.603234081182533e-05,
+      "loss": 0.258,
+      "step": 6225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6017060527942022e-05,
+      "loss": 0.2348,
+      "step": 6226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.600178315271754e-05,
+      "loss": 0.2459,
+      "step": 6227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5986508688004756e-05,
+      "loss": 0.2484,
+      "step": 6228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.597123713565618e-05,
+      "loss": 0.1555,
+      "step": 6229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5955968497523907e-05,
+      "loss": 0.3809,
+      "step": 6230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5940702775459747e-05,
+      "loss": 0.2483,
+      "step": 6231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5925439971315145e-05,
+      "loss": 0.255,
+      "step": 6232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.591018008694115e-05,
+      "loss": 0.2506,
+      "step": 6233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5894923124188496e-05,
+      "loss": 0.177,
+      "step": 6234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5879669084907583e-05,
+      "loss": 0.2548,
+      "step": 6235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5864417970948374e-05,
+      "loss": 0.1633,
+      "step": 6236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5849169784160577e-05,
+      "loss": 0.275,
+      "step": 6237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.583392452639345e-05,
+      "loss": 0.2424,
+      "step": 6238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.581868219949597e-05,
+      "loss": 0.244,
+      "step": 6239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5803442805316747e-05,
+      "loss": 0.2689,
+      "step": 6240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.578820634570398e-05,
+      "loss": 0.2576,
+      "step": 6241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5772972822505603e-05,
+      "loss": 0.2471,
+      "step": 6242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5757742237569103e-05,
+      "loss": 0.2634,
+      "step": 6243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5742514592741647e-05,
+      "loss": 0.2533,
+      "step": 6244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.572728988987006e-05,
+      "loss": 0.313,
+      "step": 6245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.571206813080082e-05,
+      "loss": 0.2728,
+      "step": 6246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.569684931737999e-05,
+      "loss": 0.2189,
+      "step": 6247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5681633451453336e-05,
+      "loss": 0.2822,
+      "step": 6248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5666420534866253e-05,
+      "loss": 0.2194,
+      "step": 6249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5651210569463733e-05,
+      "loss": 0.2448,
+      "step": 6250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.563600355709047e-05,
+      "loss": 0.2435,
+      "step": 6251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.562079949959079e-05,
+      "loss": 0.2078,
+      "step": 6252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5605598398808613e-05,
+      "loss": 0.2334,
+      "step": 6253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5590400256587542e-05,
+      "loss": 0.2227,
+      "step": 6254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5575205074770846e-05,
+      "loss": 0.2295,
+      "step": 6255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5560012855201356e-05,
+      "loss": 0.2255,
+      "step": 6256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5544823599721623e-05,
+      "loss": 0.2795,
+      "step": 6257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5529637310173794e-05,
+      "loss": 0.1374,
+      "step": 6258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.551445398839964e-05,
+      "loss": 0.2607,
+      "step": 6259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5499273636240627e-05,
+      "loss": 0.2229,
+      "step": 6260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5484096255537847e-05,
+      "loss": 0.2464,
+      "step": 6261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5468921848131983e-05,
+      "loss": 0.1935,
+      "step": 6262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5453750415863426e-05,
+      "loss": 0.2057,
+      "step": 6263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.543858196057214e-05,
+      "loss": 0.2208,
+      "step": 6264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5423416484097773e-05,
+      "loss": 0.3126,
+      "step": 6265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5408253988279618e-05,
+      "loss": 0.2196,
+      "step": 6266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5393094474956557e-05,
+      "loss": 0.2451,
+      "step": 6267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5377937945967155e-05,
+      "loss": 0.2959,
+      "step": 6268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.536278440314962e-05,
+      "loss": 0.2095,
+      "step": 6269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.534763384834174e-05,
+      "loss": 0.1893,
+      "step": 6270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5332486283380995e-05,
+      "loss": 0.3067,
+      "step": 6271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.531734171010452e-05,
+      "loss": 0.2798,
+      "step": 6272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5302200130348995e-05,
+      "loss": 0.1794,
+      "step": 6273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.528706154595085e-05,
+      "loss": 0.2,
+      "step": 6274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.527192595874608e-05,
+      "loss": 0.2424,
+      "step": 6275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5256793370570297e-05,
+      "loss": 0.2258,
+      "step": 6276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5241663783258818e-05,
+      "loss": 0.2019,
+      "step": 6277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.522653719864658e-05,
+      "loss": 0.2021,
+      "step": 6278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5211413618568113e-05,
+      "loss": 0.1962,
+      "step": 6279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.519629304485761e-05,
+      "loss": 0.2606,
+      "step": 6280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.518117547934894e-05,
+      "loss": 0.2369,
+      "step": 6281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5166060923875502e-05,
+      "loss": 0.2195,
+      "step": 6282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5150949380270455e-05,
+      "loss": 0.1726,
+      "step": 6283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5135840850366483e-05,
+      "loss": 0.2834,
+      "step": 6284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5120735335995975e-05,
+      "loss": 0.2854,
+      "step": 6285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5105632838990955e-05,
+      "loss": 0.1817,
+      "step": 6286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.509053336118301e-05,
+      "loss": 0.2674,
+      "step": 6287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.507543690440345e-05,
+      "loss": 0.3492,
+      "step": 6288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5060343470483172e-05,
+      "loss": 0.2809,
+      "step": 6289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5045253061252686e-05,
+      "loss": 0.2804,
+      "step": 6290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5030165678542204e-05,
+      "loss": 0.2165,
+      "step": 6291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.501508132418151e-05,
+      "loss": 0.1713,
+      "step": 6292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.500000000000001e-05,
+      "loss": 0.2294,
+      "step": 6293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4984921707826802e-05,
+      "loss": 0.2923,
+      "step": 6294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.49698464494906e-05,
+      "loss": 0.2797,
+      "step": 6295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4954774226819695e-05,
+      "loss": 0.2965,
+      "step": 6296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4939705041642076e-05,
+      "loss": 0.2563,
+      "step": 6297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4924638895785357e-05,
+      "loss": 0.1817,
+      "step": 6298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.490957579107673e-05,
+      "loss": 0.2004,
+      "step": 6299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.489451572934306e-05,
+      "loss": 0.2162,
+      "step": 6300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.487945871241087e-05,
+      "loss": 0.2089,
+      "step": 6301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4864404742106236e-05,
+      "loss": 0.2423,
+      "step": 6302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4849353820254944e-05,
+      "loss": 0.2517,
+      "step": 6303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.483430594868234e-05,
+      "loss": 0.3099,
+      "step": 6304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.481926112921345e-05,
+      "loss": 0.2032,
+      "step": 6305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4804219363672943e-05,
+      "loss": 0.2777,
+      "step": 6306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.478918065388506e-05,
+      "loss": 0.1914,
+      "step": 6307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4774145001673683e-05,
+      "loss": 0.1555,
+      "step": 6308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4759112408862368e-05,
+      "loss": 0.2724,
+      "step": 6309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4744082877274283e-05,
+      "loss": 0.2068,
+      "step": 6310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4729056408732177e-05,
+      "loss": 0.1762,
+      "step": 6311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4714033005058512e-05,
+      "loss": 0.2306,
+      "step": 6312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4699012668075283e-05,
+      "loss": 0.2365,
+      "step": 6313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4683995399604187e-05,
+      "loss": 0.2403,
+      "step": 6314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.466898120146654e-05,
+      "loss": 0.2693,
+      "step": 6315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4653970075483235e-05,
+      "loss": 0.2478,
+      "step": 6316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.463896202347484e-05,
+      "loss": 0.1958,
+      "step": 6317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4623957047261547e-05,
+      "loss": 0.1945,
+      "step": 6318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.460895514866315e-05,
+      "loss": 0.1765,
+      "step": 6319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.459395632949908e-05,
+      "loss": 0.1446,
+      "step": 6320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4578960591588434e-05,
+      "loss": 0.1982,
+      "step": 6321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4563967936749847e-05,
+      "loss": 0.2032,
+      "step": 6322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.454897836680169e-05,
+      "loss": 0.2665,
+      "step": 6323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4533991883561868e-05,
+      "loss": 0.1749,
+      "step": 6324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4519008488847934e-05,
+      "loss": 0.2339,
+      "step": 6325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4504028184477096e-05,
+      "loss": 0.2126,
+      "step": 6326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.448905097226619e-05,
+      "loss": 0.2367,
+      "step": 6327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4474076854031626e-05,
+      "loss": 0.2455,
+      "step": 6328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4459105831589478e-05,
+      "loss": 0.1977,
+      "step": 6329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4444137906755465e-05,
+      "loss": 0.287,
+      "step": 6330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4429173081344852e-05,
+      "loss": 0.2108,
+      "step": 6331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.441421135717263e-05,
+      "loss": 0.1721,
+      "step": 6332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4399252736053312e-05,
+      "loss": 0.3428,
+      "step": 6333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4384297219801118e-05,
+      "loss": 0.2534,
+      "step": 6334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4369344810229855e-05,
+      "loss": 0.2686,
+      "step": 6335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.435439550915294e-05,
+      "loss": 0.1952,
+      "step": 6336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4339449318383432e-05,
+      "loss": 0.2713,
+      "step": 6337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4324506239734045e-05,
+      "loss": 0.1821,
+      "step": 6338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4309566275017028e-05,
+      "loss": 0.2555,
+      "step": 6339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.429462942604435e-05,
+      "loss": 0.2017,
+      "step": 6340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4279695694627536e-05,
+      "loss": 0.2182,
+      "step": 6341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4264765082577733e-05,
+      "loss": 0.2505,
+      "step": 6342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.424983759170576e-05,
+      "loss": 0.1543,
+      "step": 6343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4234913223822032e-05,
+      "loss": 0.1911,
+      "step": 6344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.421999198073655e-05,
+      "loss": 0.2505,
+      "step": 6345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.420507386425899e-05,
+      "loss": 0.2007,
+      "step": 6346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4190158876198642e-05,
+      "loss": 0.2426,
+      "step": 6347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4175247018364363e-05,
+      "loss": 0.2185,
+      "step": 6348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4160338292564682e-05,
+      "loss": 0.2694,
+      "step": 6349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4145432700607756e-05,
+      "loss": 0.1847,
+      "step": 6350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4130530244301308e-05,
+      "loss": 0.211,
+      "step": 6351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4115630925452747e-05,
+      "loss": 0.2779,
+      "step": 6352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4100734745869032e-05,
+      "loss": 0.181,
+      "step": 6353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.408584170735679e-05,
+      "loss": 0.2977,
+      "step": 6354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.407095181172227e-05,
+      "loss": 0.1847,
+      "step": 6355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4056065060771317e-05,
+      "loss": 0.1648,
+      "step": 6356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.404118145630937e-05,
+      "loss": 0.1565,
+      "step": 6357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.402630100014156e-05,
+      "loss": 0.2515,
+      "step": 6358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.401142369407256e-05,
+      "loss": 0.2784,
+      "step": 6359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.399654953990671e-05,
+      "loss": 0.2025,
+      "step": 6360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3981678539447967e-05,
+      "loss": 0.2056,
+      "step": 6361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3966810694499857e-05,
+      "loss": 0.2297,
+      "step": 6362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3951946006865577e-05,
+      "loss": 0.2082,
+      "step": 6363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3937084478347936e-05,
+      "loss": 0.2213,
+      "step": 6364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3922226110749318e-05,
+      "loss": 0.186,
+      "step": 6365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3907370905871758e-05,
+      "loss": 0.2413,
+      "step": 6366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3892518865516922e-05,
+      "loss": 0.2364,
+      "step": 6367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3877669991486036e-05,
+      "loss": 0.1551,
+      "step": 6368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.386282428558001e-05,
+      "loss": 0.2496,
+      "step": 6369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.38479817495993e-05,
+      "loss": 0.1886,
+      "step": 6370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.383314238534403e-05,
+      "loss": 0.2076,
+      "step": 6371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3818306194613953e-05,
+      "loss": 0.188,
+      "step": 6372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3803473179208373e-05,
+      "loss": 0.1985,
+      "step": 6373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.378864334092623e-05,
+      "loss": 0.2288,
+      "step": 6374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3773816681566118e-05,
+      "loss": 0.1982,
+      "step": 6375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3758993202926226e-05,
+      "loss": 0.2392,
+      "step": 6376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3744172906804324e-05,
+      "loss": 0.1743,
+      "step": 6377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3729355794997847e-05,
+      "loss": 0.2329,
+      "step": 6378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.37145418693038e-05,
+      "loss": 0.254,
+      "step": 6379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.369973113151882e-05,
+      "loss": 0.1742,
+      "step": 6380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3684923583439188e-05,
+      "loss": 0.1666,
+      "step": 6381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3670119226860727e-05,
+      "loss": 0.1952,
+      "step": 6382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3655318063578936e-05,
+      "loss": 0.2306,
+      "step": 6383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3640520095388917e-05,
+      "loss": 0.255,
+      "step": 6384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.362572532408535e-05,
+      "loss": 0.2451,
+      "step": 6385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3610933751462553e-05,
+      "loss": 0.2964,
+      "step": 6386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.359614537931448e-05,
+      "loss": 0.1836,
+      "step": 6387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3581360209434627e-05,
+      "loss": 0.209,
+      "step": 6388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.356657824361618e-05,
+      "loss": 0.2306,
+      "step": 6389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3551799483651894e-05,
+      "loss": 0.257,
+      "step": 6390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3537023931334113e-05,
+      "loss": 0.2266,
+      "step": 6391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.352225158845484e-05,
+      "loss": 0.1905,
+      "step": 6392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3507482456805697e-05,
+      "loss": 0.1926,
+      "step": 6393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.349271653817784e-05,
+      "loss": 0.2834,
+      "step": 6394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3477953834362103e-05,
+      "loss": 0.201,
+      "step": 6395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.346319434714894e-05,
+      "loss": 0.2639,
+      "step": 6396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.344843807832835e-05,
+      "loss": 0.2949,
+      "step": 6397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3433685029690007e-05,
+      "loss": 0.2757,
+      "step": 6398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3418935203023128e-05,
+      "loss": 0.2005,
+      "step": 6399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3404188600116605e-05,
+      "loss": 0.3078,
+      "step": 6400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.338944522275892e-05,
+      "loss": 0.1309,
+      "step": 6401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3374705072738123e-05,
+      "loss": 0.1844,
+      "step": 6402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.335996815184192e-05,
+      "loss": 0.1951,
+      "step": 6403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3345234461857628e-05,
+      "loss": 0.1705,
+      "step": 6404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.333050400457212e-05,
+      "loss": 0.1606,
+      "step": 6405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3315776781771948e-05,
+      "loss": 0.2293,
+      "step": 6406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.330105279524321e-05,
+      "loss": 0.2552,
+      "step": 6407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.328633204677162e-05,
+      "loss": 0.2367,
+      "step": 6408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.327161453814254e-05,
+      "loss": 0.2248,
+      "step": 6409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3256900271140926e-05,
+      "loss": 0.1944,
+      "step": 6410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3242189247551287e-05,
+      "loss": 0.1772,
+      "step": 6411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3227481469157813e-05,
+      "loss": 0.2694,
+      "step": 6412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.321277693774428e-05,
+      "loss": 0.3102,
+      "step": 6413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.319807565509402e-05,
+      "loss": 0.1768,
+      "step": 6414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.318337762299003e-05,
+      "loss": 0.182,
+      "step": 6415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3168682843214907e-05,
+      "loss": 0.1938,
+      "step": 6416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.315399131755081e-05,
+      "loss": 0.2526,
+      "step": 6417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.313930304777956e-05,
+      "loss": 0.1811,
+      "step": 6418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.312461803568252e-05,
+      "loss": 0.1595,
+      "step": 6419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3109936283040724e-05,
+      "loss": 0.1787,
+      "step": 6420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3095257791634778e-05,
+      "loss": 0.2906,
+      "step": 6421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.308058256324489e-05,
+      "loss": 0.2381,
+      "step": 6422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3065910599650852e-05,
+      "loss": 0.1926,
+      "step": 6423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3051241902632108e-05,
+      "loss": 0.2264,
+      "step": 6424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3036576473967697e-05,
+      "loss": 0.3091,
+      "step": 6425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.302191431543621e-05,
+      "loss": 0.2561,
+      "step": 6426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3007255428815914e-05,
+      "loss": 0.2773,
+      "step": 6427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2992599815884613e-05,
+      "loss": 0.2497,
+      "step": 6428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2977947478419758e-05,
+      "loss": 0.1943,
+      "step": 6429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.296329841819841e-05,
+      "loss": 0.3011,
+      "step": 6430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2948652636997177e-05,
+      "loss": 0.1381,
+      "step": 6431
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2934010136592315e-05,
+      "loss": 0.2583,
+      "step": 6432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2919370918759698e-05,
+      "loss": 0.2262,
+      "step": 6433
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.290473498527473e-05,
+      "loss": 0.2258,
+      "step": 6434
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.289010233791249e-05,
+      "loss": 0.2694,
+      "step": 6435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2875472978447637e-05,
+      "loss": 0.2225,
+      "step": 6436
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.28608469086544e-05,
+      "loss": 0.1776,
+      "step": 6437
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2846224130306665e-05,
+      "loss": 0.2007,
+      "step": 6438
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.283160464517787e-05,
+      "loss": 0.2828,
+      "step": 6439
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2816988455041043e-05,
+      "loss": 0.2821,
+      "step": 6440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2802375561668878e-05,
+      "loss": 0.2224,
+      "step": 6441
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2787765966833628e-05,
+      "loss": 0.2368,
+      "step": 6442
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2773159672307132e-05,
+      "loss": 0.1481,
+      "step": 6443
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.275855667986086e-05,
+      "loss": 0.193,
+      "step": 6444
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2743956991265874e-05,
+      "loss": 0.2271,
+      "step": 6445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2729360608292805e-05,
+      "loss": 0.2894,
+      "step": 6446
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.271476753271194e-05,
+      "loss": 0.2303,
+      "step": 6447
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2700177766293096e-05,
+      "loss": 0.2846,
+      "step": 6448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2685591310805744e-05,
+      "loss": 0.1991,
+      "step": 6449
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.267100816801895e-05,
+      "loss": 0.2012,
+      "step": 6450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2656428339701326e-05,
+      "loss": 0.2851,
+      "step": 6451
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.264185182762113e-05,
+      "loss": 0.2231,
+      "step": 6452
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.262727863354624e-05,
+      "loss": 0.3096,
+      "step": 6453
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.261270875924405e-05,
+      "loss": 0.2395,
+      "step": 6454
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2598142206481643e-05,
+      "loss": 0.142,
+      "step": 6455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.258357897702563e-05,
+      "loss": 0.2172,
+      "step": 6456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.256901907264223e-05,
+      "loss": 0.2147,
+      "step": 6457
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.255446249509729e-05,
+      "loss": 0.2322,
+      "step": 6458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2539909246156254e-05,
+      "loss": 0.2234,
+      "step": 6459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2525359327584104e-05,
+      "loss": 0.197,
+      "step": 6460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.251081274114549e-05,
+      "loss": 0.2488,
+      "step": 6461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.249626948860464e-05,
+      "loss": 0.2226,
+      "step": 6462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.248172957172532e-05,
+      "loss": 0.2135,
+      "step": 6463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2467192992270982e-05,
+      "loss": 0.2204,
+      "step": 6464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.245265975200458e-05,
+      "loss": 0.2743,
+      "step": 6465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.243812985268874e-05,
+      "loss": 0.1914,
+      "step": 6466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2423603296085667e-05,
+      "loss": 0.2185,
+      "step": 6467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2409080083957107e-05,
+      "loss": 0.201,
+      "step": 6468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2394560218064463e-05,
+      "loss": 0.1861,
+      "step": 6469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2380043700168724e-05,
+      "loss": 0.1355,
+      "step": 6470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2365530532030444e-05,
+      "loss": 0.2171,
+      "step": 6471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2351020715409758e-05,
+      "loss": 0.1732,
+      "step": 6472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2336514252066465e-05,
+      "loss": 0.2545,
+      "step": 6473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.232201114375988e-05,
+      "loss": 0.1679,
+      "step": 6474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2307511392248953e-05,
+      "loss": 0.2408,
+      "step": 6475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2293014999292246e-05,
+      "loss": 0.2946,
+      "step": 6476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2278521966647848e-05,
+      "loss": 0.1669,
+      "step": 6477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2264032296073495e-05,
+      "loss": 0.2196,
+      "step": 6478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2249545989326514e-05,
+      "loss": 0.2586,
+      "step": 6479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.223506304816378e-05,
+      "loss": 0.1545,
+      "step": 6480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.22205834743418e-05,
+      "loss": 0.2036,
+      "step": 6481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.220610726961669e-05,
+      "loss": 0.2641,
+      "step": 6482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.219163443574408e-05,
+      "loss": 0.2071,
+      "step": 6483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.217716497447928e-05,
+      "loss": 0.2542,
+      "step": 6484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2162698887577123e-05,
+      "loss": 0.2232,
+      "step": 6485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.214823617679207e-05,
+      "loss": 0.2199,
+      "step": 6486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2133776843878186e-05,
+      "loss": 0.2301,
+      "step": 6487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.211932089058908e-05,
+      "loss": 0.1847,
+      "step": 6488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2104868318677964e-05,
+      "loss": 0.1866,
+      "step": 6489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2090419129897666e-05,
+      "loss": 0.2035,
+      "step": 6490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2075973326000606e-05,
+      "loss": 0.2458,
+      "step": 6491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2061530908738736e-05,
+      "loss": 0.2615,
+      "step": 6492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.204709187986368e-05,
+      "loss": 0.2226,
+      "step": 6493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.203265624112657e-05,
+      "loss": 0.2394,
+      "step": 6494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.201822399427818e-05,
+      "loss": 0.244,
+      "step": 6495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2003795141068883e-05,
+      "loss": 0.1948,
+      "step": 6496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1989369683248568e-05,
+      "loss": 0.1732,
+      "step": 6497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1974947622566787e-05,
+      "loss": 0.1549,
+      "step": 6498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1960528960772664e-05,
+      "loss": 0.2784,
+      "step": 6499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1946113699614868e-05,
+      "loss": 0.1812,
+      "step": 6500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.19317018408417e-05,
+      "loss": 0.2488,
+      "step": 6501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1917293386201056e-05,
+      "loss": 0.1759,
+      "step": 6502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1902888337440363e-05,
+      "loss": 0.2549,
+      "step": 6503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1888486696306704e-05,
+      "loss": 0.2327,
+      "step": 6504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1874088464546694e-05,
+      "loss": 0.2517,
+      "step": 6505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.185969364390655e-05,
+      "loss": 0.2419,
+      "step": 6506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1845302236132082e-05,
+      "loss": 0.1809,
+      "step": 6507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1830914242968707e-05,
+      "loss": 0.2165,
+      "step": 6508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1816529666161377e-05,
+      "loss": 0.2455,
+      "step": 6509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.180214850745467e-05,
+      "loss": 0.2154,
+      "step": 6510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1787770768592763e-05,
+      "loss": 0.1702,
+      "step": 6511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1773396451319344e-05,
+      "loss": 0.1575,
+      "step": 6512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.175902555737778e-05,
+      "loss": 0.2317,
+      "step": 6513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1744658088510945e-05,
+      "loss": 0.2108,
+      "step": 6514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.173029404646134e-05,
+      "loss": 0.2733,
+      "step": 6515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1715933432971062e-05,
+      "loss": 0.2073,
+      "step": 6516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1701576249781742e-05,
+      "loss": 0.1939,
+      "step": 6517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.168722249863463e-05,
+      "loss": 0.1624,
+      "step": 6518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1672872181270577e-05,
+      "loss": 0.2365,
+      "step": 6519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1658525299429977e-05,
+      "loss": 0.175,
+      "step": 6520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1644181854852803e-05,
+      "loss": 0.1614,
+      "step": 6521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1629841849278675e-05,
+      "loss": 0.2978,
+      "step": 6522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1615505284446712e-05,
+      "loss": 0.2075,
+      "step": 6523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1601172162095678e-05,
+      "loss": 0.1675,
+      "step": 6524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.158684248396392e-05,
+      "loss": 0.2759,
+      "step": 6525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1572516251789303e-05,
+      "loss": 0.1789,
+      "step": 6526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.155819346730934e-05,
+      "loss": 0.2573,
+      "step": 6527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1543874132261116e-05,
+      "loss": 0.2917,
+      "step": 6528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1529558248381254e-05,
+      "loss": 0.2117,
+      "step": 6529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.151524581740601e-05,
+      "loss": 0.2061,
+      "step": 6530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1500936841071212e-05,
+      "loss": 0.1953,
+      "step": 6531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1486631321112223e-05,
+      "loss": 0.2619,
+      "step": 6532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1472329259264053e-05,
+      "loss": 0.1793,
+      "step": 6533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1458030657261235e-05,
+      "loss": 0.1697,
+      "step": 6534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1443735516837914e-05,
+      "loss": 0.2228,
+      "step": 6535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1429443839727843e-05,
+      "loss": 0.1768,
+      "step": 6536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1415155627664284e-05,
+      "loss": 0.1346,
+      "step": 6537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1400870882380118e-05,
+      "loss": 0.1868,
+      "step": 6538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1386589605607825e-05,
+      "loss": 0.221,
+      "step": 6539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1372311799079416e-05,
+      "loss": 0.2595,
+      "step": 6540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1358037464526515e-05,
+      "loss": 0.1808,
+      "step": 6541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.134376660368035e-05,
+      "loss": 0.1794,
+      "step": 6542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.132949921827165e-05,
+      "loss": 0.2156,
+      "step": 6543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1315235310030784e-05,
+      "loss": 0.2631,
+      "step": 6544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.130097488068771e-05,
+      "loss": 0.2721,
+      "step": 6545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.12867179319719e-05,
+      "loss": 0.1577,
+      "step": 6546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1272464465612458e-05,
+      "loss": 0.207,
+      "step": 6547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1258214483338068e-05,
+      "loss": 0.2163,
+      "step": 6548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1243967986876933e-05,
+      "loss": 0.2069,
+      "step": 6549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1229724977956895e-05,
+      "loss": 0.1889,
+      "step": 6550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1215485458305363e-05,
+      "loss": 0.1795,
+      "step": 6551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1201249429649284e-05,
+      "loss": 0.1495,
+      "step": 6552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1187016893715233e-05,
+      "loss": 0.1949,
+      "step": 6553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1172787852229324e-05,
+      "loss": 0.1689,
+      "step": 6554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1158562306917252e-05,
+      "loss": 0.1823,
+      "step": 6555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1144340259504298e-05,
+      "loss": 0.2166,
+      "step": 6556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1130121711715338e-05,
+      "loss": 0.2378,
+      "step": 6557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1115906665274776e-05,
+      "loss": 0.2144,
+      "step": 6558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.110169512190664e-05,
+      "loss": 0.2493,
+      "step": 6559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1087487083334485e-05,
+      "loss": 0.1902,
+      "step": 6560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.107328255128148e-05,
+      "loss": 0.1921,
+      "step": 6561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1059081527470375e-05,
+      "loss": 0.2034,
+      "step": 6562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.104488401362344e-05,
+      "loss": 0.2337,
+      "step": 6563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1030690011462566e-05,
+      "loss": 0.2359,
+      "step": 6564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.101649952270923e-05,
+      "loss": 0.1269,
+      "step": 6565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1002312549084426e-05,
+      "loss": 0.2694,
+      "step": 6566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0988129092308765e-05,
+      "loss": 0.1956,
+      "step": 6567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0973949154102446e-05,
+      "loss": 0.2526,
+      "step": 6568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0959772736185173e-05,
+      "loss": 0.1762,
+      "step": 6569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.094559984027631e-05,
+      "loss": 0.2346,
+      "step": 6570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0931430468094733e-05,
+      "loss": 0.2355,
+      "step": 6571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.091726462135888e-05,
+      "loss": 0.1581,
+      "step": 6572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.090310230178682e-05,
+      "loss": 0.2326,
+      "step": 6573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0888943511096177e-05,
+      "loss": 0.2107,
+      "step": 6574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.087478825100409e-05,
+      "loss": 0.2044,
+      "step": 6575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0860636523227346e-05,
+      "loss": 0.2559,
+      "step": 6576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.084648832948228e-05,
+      "loss": 0.2262,
+      "step": 6577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0832343671484754e-05,
+      "loss": 0.2054,
+      "step": 6578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.081820255095028e-05,
+      "loss": 0.2175,
+      "step": 6579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0804064969593852e-05,
+      "loss": 0.221,
+      "step": 6580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0789930929130102e-05,
+      "loss": 0.2639,
+      "step": 6581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0775800431273234e-05,
+      "loss": 0.2399,
+      "step": 6582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0761673477736954e-05,
+      "loss": 0.2378,
+      "step": 6583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.074755007023461e-05,
+      "loss": 0.2097,
+      "step": 6584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.073343021047911e-05,
+      "loss": 0.2101,
+      "step": 6585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0719313900182895e-05,
+      "loss": 0.2157,
+      "step": 6586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0705201141057978e-05,
+      "loss": 0.1832,
+      "step": 6587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0691091934815997e-05,
+      "loss": 0.2369,
+      "step": 6588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0676986283168082e-05,
+      "loss": 0.2456,
+      "step": 6589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0662884187824993e-05,
+      "loss": 0.2616,
+      "step": 6590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0648785650497055e-05,
+      "loss": 0.2089,
+      "step": 6591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0634690672894103e-05,
+      "loss": 0.3138,
+      "step": 6592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0620599256725597e-05,
+      "loss": 0.1831,
+      "step": 6593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0606511403700575e-05,
+      "loss": 0.1616,
+      "step": 6594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0592427115527573e-05,
+      "loss": 0.1884,
+      "step": 6595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.057834639391476e-05,
+      "loss": 0.19,
+      "step": 6596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0564269240569874e-05,
+      "loss": 0.1871,
+      "step": 6597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0550195657200143e-05,
+      "loss": 0.2738,
+      "step": 6598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0536125645512472e-05,
+      "loss": 0.2096,
+      "step": 6599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.052205920721323e-05,
+      "loss": 0.2035,
+      "step": 6600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0507996344008418e-05,
+      "loss": 0.2251,
+      "step": 6601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0493937057603602e-05,
+      "loss": 0.3054,
+      "step": 6602
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0479881349703883e-05,
+      "loss": 0.2701,
+      "step": 6603
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.046582922201392e-05,
+      "loss": 0.1574,
+      "step": 6604
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.045178067623798e-05,
+      "loss": 0.2253,
+      "step": 6605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0437735714079896e-05,
+      "loss": 0.2749,
+      "step": 6606
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0423694337243006e-05,
+      "loss": 0.1924,
+      "step": 6607
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.040965654743029e-05,
+      "loss": 0.2444,
+      "step": 6608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0395622346344212e-05,
+      "loss": 0.2307,
+      "step": 6609
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.038159173568688e-05,
+      "loss": 0.2177,
+      "step": 6610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0367564717159933e-05,
+      "loss": 0.1812,
+      "step": 6611
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.035354129246454e-05,
+      "loss": 0.2193,
+      "step": 6612
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0339521463301487e-05,
+      "loss": 0.2167,
+      "step": 6613
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0325505231371123e-05,
+      "loss": 0.1736,
+      "step": 6614
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0311492598373305e-05,
+      "loss": 0.1851,
+      "step": 6615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0297483566007507e-05,
+      "loss": 0.2204,
+      "step": 6616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0283478135972766e-05,
+      "loss": 0.2442,
+      "step": 6617
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.026947630996763e-05,
+      "loss": 0.1917,
+      "step": 6618
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.025547808969028e-05,
+      "loss": 0.2386,
+      "step": 6619
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0241483476838403e-05,
+      "loss": 0.2586,
+      "step": 6620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.022749247310926e-05,
+      "loss": 0.2238,
+      "step": 6621
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.02135050801997e-05,
+      "loss": 0.2964,
+      "step": 6622
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.019952129980614e-05,
+      "loss": 0.2338,
+      "step": 6623
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0185541133624493e-05,
+      "loss": 0.2237,
+      "step": 6624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.01715645833503e-05,
+      "loss": 0.2737,
+      "step": 6625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0157591650678666e-05,
+      "loss": 0.2942,
+      "step": 6626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0143622337304185e-05,
+      "loss": 0.1688,
+      "step": 6627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.012965664492111e-05,
+      "loss": 0.2059,
+      "step": 6628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.011569457522315e-05,
+      "loss": 0.2556,
+      "step": 6629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0101736129903666e-05,
+      "loss": 0.2228,
+      "step": 6630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.008778131065555e-05,
+      "loss": 0.1968,
+      "step": 6631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0073830119171216e-05,
+      "loss": 0.1773,
+      "step": 6632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.005988255714269e-05,
+      "loss": 0.2538,
+      "step": 6633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0045938626261546e-05,
+      "loss": 0.245,
+      "step": 6634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0031998328218892e-05,
+      "loss": 0.2374,
+      "step": 6635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.00180616647054e-05,
+      "loss": 0.1973,
+      "step": 6636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.000412863741135e-05,
+      "loss": 0.1909,
+      "step": 6637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9990199248026505e-05,
+      "loss": 0.2541,
+      "step": 6638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9976273498240233e-05,
+      "loss": 0.288,
+      "step": 6639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9962351389741492e-05,
+      "loss": 0.1967,
+      "step": 6640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9948432924218714e-05,
+      "loss": 0.1636,
+      "step": 6641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9934518103359944e-05,
+      "loss": 0.1903,
+      "step": 6642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.992060692885281e-05,
+      "loss": 0.2873,
+      "step": 6643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9906699402384422e-05,
+      "loss": 0.2039,
+      "step": 6644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9892795525641505e-05,
+      "loss": 0.2082,
+      "step": 6645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.987889530031034e-05,
+      "loss": 0.1929,
+      "step": 6646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.986499872807673e-05,
+      "loss": 0.2522,
+      "step": 6647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9851105810626076e-05,
+      "loss": 0.2136,
+      "step": 6648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9837216549643285e-05,
+      "loss": 0.2299,
+      "step": 6649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.982333094681287e-05,
+      "loss": 0.2603,
+      "step": 6650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9809449003818897e-05,
+      "loss": 0.2291,
+      "step": 6651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9795570722344963e-05,
+      "loss": 0.2354,
+      "step": 6652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9781696104074203e-05,
+      "loss": 0.2494,
+      "step": 6653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976782515068938e-05,
+      "loss": 0.2022,
+      "step": 6654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9753957863872718e-05,
+      "loss": 0.261,
+      "step": 6655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.974009424530608e-05,
+      "loss": 0.2438,
+      "step": 6656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9726234296670854e-05,
+      "loss": 0.2246,
+      "step": 6657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9712378019647953e-05,
+      "loss": 0.1904,
+      "step": 6658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.969852541591789e-05,
+      "loss": 0.1975,
+      "step": 6659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9684676487160726e-05,
+      "loss": 0.222,
+      "step": 6660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9670831235056027e-05,
+      "loss": 0.2377,
+      "step": 6661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.965698966128297e-05,
+      "loss": 0.1716,
+      "step": 6662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9643151767520285e-05,
+      "loss": 0.1801,
+      "step": 6663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9629317555446197e-05,
+      "loss": 0.2199,
+      "step": 6664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9615487026738543e-05,
+      "loss": 0.177,
+      "step": 6665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9601660183074716e-05,
+      "loss": 0.1781,
+      "step": 6666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.95878370261316e-05,
+      "loss": 0.274,
+      "step": 6667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.957401755758571e-05,
+      "loss": 0.1819,
+      "step": 6668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9560201779113057e-05,
+      "loss": 0.2405,
+      "step": 6669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.95463896923892e-05,
+      "loss": 0.2556,
+      "step": 6670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9532581299089304e-05,
+      "loss": 0.219,
+      "step": 6671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.951877660088806e-05,
+      "loss": 0.2577,
+      "step": 6672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.950497559945968e-05,
+      "loss": 0.2363,
+      "step": 6673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9491178296477985e-05,
+      "loss": 0.2371,
+      "step": 6674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.947738469361628e-05,
+      "loss": 0.1908,
+      "step": 6675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.946359479254748e-05,
+      "loss": 0.156,
+      "step": 6676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.944980859494404e-05,
+      "loss": 0.2533,
+      "step": 6677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9436026102477928e-05,
+      "loss": 0.2617,
+      "step": 6678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.94222473168207e-05,
+      "loss": 0.1869,
+      "step": 6679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.940847223964348e-05,
+      "loss": 0.2594,
+      "step": 6680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9394700872616855e-05,
+      "loss": 0.1863,
+      "step": 6681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9380933217411063e-05,
+      "loss": 0.1907,
+      "step": 6682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9367169275695856e-05,
+      "loss": 0.2306,
+      "step": 6683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9353409049140513e-05,
+      "loss": 0.2736,
+      "step": 6684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9339652539413867e-05,
+      "loss": 0.1489,
+      "step": 6685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932589974818434e-05,
+      "loss": 0.2133,
+      "step": 6686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931215067711984e-05,
+      "loss": 0.2304,
+      "step": 6687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9298405327887882e-05,
+      "loss": 0.2124,
+      "step": 6688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.928466370215552e-05,
+      "loss": 0.2026,
+      "step": 6689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9270925801589308e-05,
+      "loss": 0.1845,
+      "step": 6690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9257191627855398e-05,
+      "loss": 0.178,
+      "step": 6691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9243461182619503e-05,
+      "loss": 0.2347,
+      "step": 6692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.922973446754681e-05,
+      "loss": 0.2339,
+      "step": 6693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9216011484302137e-05,
+      "loss": 0.2319,
+      "step": 6694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9202292234549785e-05,
+      "loss": 0.2224,
+      "step": 6695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9188576719953633e-05,
+      "loss": 0.2131,
+      "step": 6696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9174864942177136e-05,
+      "loss": 0.1575,
+      "step": 6697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9161156902883208e-05,
+      "loss": 0.2043,
+      "step": 6698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.91474526037344e-05,
+      "loss": 0.1397,
+      "step": 6699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.913375204639279e-05,
+      "loss": 0.2338,
+      "step": 6700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9120055232519957e-05,
+      "loss": 0.2012,
+      "step": 6701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9106362163777046e-05,
+      "loss": 0.2137,
+      "step": 6702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9092672841824793e-05,
+      "loss": 0.3263,
+      "step": 6703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9078987268323405e-05,
+      "loss": 0.1891,
+      "step": 6704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.906530544493269e-05,
+      "loss": 0.1465,
+      "step": 6705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.905162737331201e-05,
+      "loss": 0.2554,
+      "step": 6706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9037953055120196e-05,
+      "loss": 0.1333,
+      "step": 6707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.90242824920157e-05,
+      "loss": 0.1495,
+      "step": 6708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9010615685656513e-05,
+      "loss": 0.2096,
+      "step": 6709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.899695263770011e-05,
+      "loss": 0.201,
+      "step": 6710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8983293349803566e-05,
+      "loss": 0.2033,
+      "step": 6711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8969637823623504e-05,
+      "loss": 0.2318,
+      "step": 6712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.895598606081604e-05,
+      "loss": 0.2701,
+      "step": 6713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.894233806303689e-05,
+      "loss": 0.2752,
+      "step": 6714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8928693831941264e-05,
+      "loss": 0.1979,
+      "step": 6715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8915053369183944e-05,
+      "loss": 0.1784,
+      "step": 6716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8901416676419277e-05,
+      "loss": 0.1824,
+      "step": 6717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8887783755301104e-05,
+      "loss": 0.2005,
+      "step": 6718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8874154607482815e-05,
+      "loss": 0.2131,
+      "step": 6719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.886052923461737e-05,
+      "loss": 0.1753,
+      "step": 6720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.884690763835728e-05,
+      "loss": 0.2623,
+      "step": 6721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8833289820354538e-05,
+      "loss": 0.1823,
+      "step": 6722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8819675782260765e-05,
+      "loss": 0.2177,
+      "step": 6723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8806065525727025e-05,
+      "loss": 0.2534,
+      "step": 6724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8792459052404e-05,
+      "loss": 0.2174,
+      "step": 6725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8778856363941906e-05,
+      "loss": 0.229,
+      "step": 6726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8765257461990442e-05,
+      "loss": 0.18,
+      "step": 6727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8751662348198905e-05,
+      "loss": 0.1992,
+      "step": 6728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.873807102421614e-05,
+      "loss": 0.2625,
+      "step": 6729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8724483491690463e-05,
+      "loss": 0.1857,
+      "step": 6730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.87108997522698e-05,
+      "loss": 0.2763,
+      "step": 6731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8697319807601605e-05,
+      "loss": 0.1964,
+      "step": 6732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.868374365933282e-05,
+      "loss": 0.197,
+      "step": 6733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.867017130911e-05,
+      "loss": 0.1786,
+      "step": 6734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8656602758579196e-05,
+      "loss": 0.1891,
+      "step": 6735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8643038009385977e-05,
+      "loss": 0.203,
+      "step": 6736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.862947706317551e-05,
+      "loss": 0.2499,
+      "step": 6737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.861591992159247e-05,
+      "loss": 0.2892,
+      "step": 6738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.860236658628106e-05,
+      "loss": 0.2477,
+      "step": 6739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8588817058885033e-05,
+      "loss": 0.2,
+      "step": 6740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8575271341047705e-05,
+      "loss": 0.1823,
+      "step": 6741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8561729434411867e-05,
+      "loss": 0.1703,
+      "step": 6742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8548191340619926e-05,
+      "loss": 0.2068,
+      "step": 6743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8534657061313743e-05,
+      "loss": 0.2037,
+      "step": 6744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8521126598134785e-05,
+      "loss": 0.1534,
+      "step": 6745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.850759995272404e-05,
+      "loss": 0.2421,
+      "step": 6746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8494077126722004e-05,
+      "loss": 0.2528,
+      "step": 6747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8480558121768737e-05,
+      "loss": 0.2756,
+      "step": 6748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8467042939503842e-05,
+      "loss": 0.1771,
+      "step": 6749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8453531581566435e-05,
+      "loss": 0.2069,
+      "step": 6750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8440024049595157e-05,
+      "loss": 0.1966,
+      "step": 6751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8426520345228242e-05,
+      "loss": 0.2592,
+      "step": 6752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8413020470103394e-05,
+      "loss": 0.1952,
+      "step": 6753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.839952442585789e-05,
+      "loss": 0.2214,
+      "step": 6754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.838603221412855e-05,
+      "loss": 0.2241,
+      "step": 6755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.83725438365517e-05,
+      "loss": 0.206,
+      "step": 6756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8359059294763203e-05,
+      "loss": 0.2403,
+      "step": 6757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.834557859039851e-05,
+      "loss": 0.2581,
+      "step": 6758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8332101725092522e-05,
+      "loss": 0.1519,
+      "step": 6759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8318628700479728e-05,
+      "loss": 0.2014,
+      "step": 6760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.830515951819417e-05,
+      "loss": 0.2145,
+      "step": 6761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8291694179869357e-05,
+      "loss": 0.2558,
+      "step": 6762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8278232687138397e-05,
+      "loss": 0.166,
+      "step": 6763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8264775041633882e-05,
+      "loss": 0.2465,
+      "step": 6764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8251321244987963e-05,
+      "loss": 0.3064,
+      "step": 6765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8237871298832347e-05,
+      "loss": 0.2218,
+      "step": 6766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8224425204798234e-05,
+      "loss": 0.169,
+      "step": 6767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8210982964516342e-05,
+      "loss": 0.1916,
+      "step": 6768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8197544579617e-05,
+      "loss": 0.2612,
+      "step": 6769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8184110051729967e-05,
+      "loss": 0.2361,
+      "step": 6770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8170679382484617e-05,
+      "loss": 0.1924,
+      "step": 6771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8157252573509842e-05,
+      "loss": 0.1979,
+      "step": 6772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.814382962643401e-05,
+      "loss": 0.2427,
+      "step": 6773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8130410542885085e-05,
+      "loss": 0.2004,
+      "step": 6774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.811699532449055e-05,
+      "loss": 0.2419,
+      "step": 6775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8103583972877375e-05,
+      "loss": 0.1916,
+      "step": 6776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8090176489672104e-05,
+      "loss": 0.3248,
+      "step": 6777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.807677287650083e-05,
+      "loss": 0.1729,
+      "step": 6778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8063373134989103e-05,
+      "loss": 0.2376,
+      "step": 6779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8049977266762075e-05,
+      "loss": 0.1609,
+      "step": 6780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8036585273444406e-05,
+      "loss": 0.1436,
+      "step": 6781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.802319715666026e-05,
+      "loss": 0.2132,
+      "step": 6782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8009812918033374e-05,
+      "loss": 0.2028,
+      "step": 6783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.799643255918698e-05,
+      "loss": 0.1618,
+      "step": 6784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7983056081743843e-05,
+      "loss": 0.185,
+      "step": 6785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7969683487326276e-05,
+      "loss": 0.2468,
+      "step": 6786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.795631477755613e-05,
+      "loss": 0.2362,
+      "step": 6787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7942949954054738e-05,
+      "loss": 0.1878,
+      "step": 6788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7929589018443016e-05,
+      "loss": 0.1694,
+      "step": 6789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7916231972341353e-05,
+      "loss": 0.2175,
+      "step": 6790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7902878817369712e-05,
+      "loss": 0.1988,
+      "step": 6791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7889529555147584e-05,
+      "loss": 0.1799,
+      "step": 6792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7876184187293943e-05,
+      "loss": 0.2425,
+      "step": 6793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7862842715427323e-05,
+      "loss": 0.1811,
+      "step": 6794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.784950514116582e-05,
+      "loss": 0.2087,
+      "step": 6795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7836171466126965e-05,
+      "loss": 0.2426,
+      "step": 6796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7822841691927906e-05,
+      "loss": 0.2096,
+      "step": 6797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7809515820185286e-05,
+      "loss": 0.2249,
+      "step": 6798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7796193852515258e-05,
+      "loss": 0.282,
+      "step": 6799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77828757905335e-05,
+      "loss": 0.1738,
+      "step": 6800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.776956163585526e-05,
+      "loss": 0.1341,
+      "step": 6801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.775625139009525e-05,
+      "loss": 0.232,
+      "step": 6802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7742945054867764e-05,
+      "loss": 0.1845,
+      "step": 6803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7729642631786614e-05,
+      "loss": 0.2085,
+      "step": 6804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7716344122465083e-05,
+      "loss": 0.2416,
+      "step": 6805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770304952851604e-05,
+      "loss": 0.2394,
+      "step": 6806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7689758851551875e-05,
+      "loss": 0.2262,
+      "step": 6807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767647209318445e-05,
+      "loss": 0.3106,
+      "step": 6808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.766318925502522e-05,
+      "loss": 0.1791,
+      "step": 6809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76499103386851e-05,
+      "loss": 0.2503,
+      "step": 6810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7636635345774576e-05,
+      "loss": 0.1337,
+      "step": 6811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7623364277903666e-05,
+      "loss": 0.2225,
+      "step": 6812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7610097136681852e-05,
+      "loss": 0.2385,
+      "step": 6813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7596833923718192e-05,
+      "loss": 0.2412,
+      "step": 6814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7583574640621275e-05,
+      "loss": 0.159,
+      "step": 6815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.757031928899917e-05,
+      "loss": 0.1812,
+      "step": 6816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7557067870459474e-05,
+      "loss": 0.1754,
+      "step": 6817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7543820386609362e-05,
+      "loss": 0.2113,
+      "step": 6818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.753057683905545e-05,
+      "loss": 0.1882,
+      "step": 6819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7517337229403946e-05,
+      "loss": 0.2094,
+      "step": 6820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.750410155926057e-05,
+      "loss": 0.1898,
+      "step": 6821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.749086983023051e-05,
+      "loss": 0.197,
+      "step": 6822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7477642043918535e-05,
+      "loss": 0.203,
+      "step": 6823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.746441820192893e-05,
+      "loss": 0.1918,
+      "step": 6824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.745119830586545e-05,
+      "loss": 0.1723,
+      "step": 6825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.743798235733144e-05,
+      "loss": 0.2348,
+      "step": 6826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.742477035792973e-05,
+      "loss": 0.2104,
+      "step": 6827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7411562309262656e-05,
+      "loss": 0.2844,
+      "step": 6828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.739835821293213e-05,
+      "loss": 0.2093,
+      "step": 6829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7385158070539504e-05,
+      "loss": 0.2111,
+      "step": 6830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7371961883685722e-05,
+      "loss": 0.2136,
+      "step": 6831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7358769653971234e-05,
+      "loss": 0.2461,
+      "step": 6832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7345581382995986e-05,
+      "loss": 0.1691,
+      "step": 6833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7332397072359435e-05,
+      "loss": 0.1921,
+      "step": 6834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.73192167236606e-05,
+      "loss": 0.2329,
+      "step": 6835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7306040338498004e-05,
+      "loss": 0.1854,
+      "step": 6836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7292867918469658e-05,
+      "loss": 0.1615,
+      "step": 6837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7279699465173148e-05,
+      "loss": 0.2458,
+      "step": 6838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.726653498020552e-05,
+      "loss": 0.3103,
+      "step": 6839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7253374465163385e-05,
+      "loss": 0.2718,
+      "step": 6840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7240217921642866e-05,
+      "loss": 0.2064,
+      "step": 6841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7227065351239553e-05,
+      "loss": 0.1849,
+      "step": 6842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7213916755548627e-05,
+      "loss": 0.2112,
+      "step": 6843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7200772136164754e-05,
+      "loss": 0.2115,
+      "step": 6844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.71876314946821e-05,
+      "loss": 0.1901,
+      "step": 6845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.717449483269437e-05,
+      "loss": 0.1526,
+      "step": 6846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7161362151794807e-05,
+      "loss": 0.2174,
+      "step": 6847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7148233453576128e-05,
+      "loss": 0.2344,
+      "step": 6848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7135108739630572e-05,
+      "loss": 0.2591,
+      "step": 6849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7121988011549938e-05,
+      "loss": 0.1782,
+      "step": 6850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.710887127092548e-05,
+      "loss": 0.1913,
+      "step": 6851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7095758519348014e-05,
+      "loss": 0.1736,
+      "step": 6852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.708264975840789e-05,
+      "loss": 0.2514,
+      "step": 6853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7069544989694885e-05,
+      "loss": 0.2569,
+      "step": 6854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.705644421479839e-05,
+      "loss": 0.1843,
+      "step": 6855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7043347435307272e-05,
+      "loss": 0.1761,
+      "step": 6856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.703025465280989e-05,
+      "loss": 0.2279,
+      "step": 6857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.701716586889417e-05,
+      "loss": 0.1524,
+      "step": 6858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7004081085147482e-05,
+      "loss": 0.2892,
+      "step": 6859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.699100030315679e-05,
+      "loss": 0.1137,
+      "step": 6860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6977923524508532e-05,
+      "loss": 0.2051,
+      "step": 6861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.696485075078864e-05,
+      "loss": 0.2497,
+      "step": 6862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6951781983582604e-05,
+      "loss": 0.243,
+      "step": 6863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.693871722447542e-05,
+      "loss": 0.1458,
+      "step": 6864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6925656475051567e-05,
+      "loss": 0.2956,
+      "step": 6865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6912599736895047e-05,
+      "loss": 0.1424,
+      "step": 6866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6899547011589418e-05,
+      "loss": 0.1898,
+      "step": 6867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.688649830071768e-05,
+      "loss": 0.2241,
+      "step": 6868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6873453605862405e-05,
+      "loss": 0.2251,
+      "step": 6869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6860412928605678e-05,
+      "loss": 0.2043,
+      "step": 6870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6847376270529037e-05,
+      "loss": 0.2407,
+      "step": 6871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6834343633213595e-05,
+      "loss": 0.2678,
+      "step": 6872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6821315018239962e-05,
+      "loss": 0.3281,
+      "step": 6873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6808290427188235e-05,
+      "loss": 0.2328,
+      "step": 6874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6795269861638042e-05,
+      "loss": 0.1767,
+      "step": 6875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6782253323168546e-05,
+      "loss": 0.1916,
+      "step": 6876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.676924081335836e-05,
+      "loss": 0.2711,
+      "step": 6877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.675623233378568e-05,
+      "loss": 0.1959,
+      "step": 6878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6743227886028152e-05,
+      "loss": 0.2259,
+      "step": 6879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6730227471662966e-05,
+      "loss": 0.2125,
+      "step": 6880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.671723109226684e-05,
+      "loss": 0.1921,
+      "step": 6881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6704238749415957e-05,
+      "loss": 0.1504,
+      "step": 6882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6691250444686023e-05,
+      "loss": 0.2183,
+      "step": 6883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.667826617965229e-05,
+      "loss": 0.2184,
+      "step": 6884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6665285955889465e-05,
+      "loss": 0.2505,
+      "step": 6885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6652309774971813e-05,
+      "loss": 0.199,
+      "step": 6886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6639337638473103e-05,
+      "loss": 0.2558,
+      "step": 6887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6626369547966565e-05,
+      "loss": 0.2099,
+      "step": 6888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6613405505024987e-05,
+      "loss": 0.2198,
+      "step": 6889
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.660044551122068e-05,
+      "loss": 0.262,
+      "step": 6890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.658748956812539e-05,
+      "loss": 0.2193,
+      "step": 6891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6574537677310448e-05,
+      "loss": 0.2165,
+      "step": 6892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.656158984034667e-05,
+      "loss": 0.2076,
+      "step": 6893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6548646058804346e-05,
+      "loss": 0.1924,
+      "step": 6894
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6535706334253314e-05,
+      "loss": 0.2642,
+      "step": 6895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6522770668262932e-05,
+      "loss": 0.1888,
+      "step": 6896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6509839062402004e-05,
+      "loss": 0.1466,
+      "step": 6897
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6496911518238912e-05,
+      "loss": 0.2423,
+      "step": 6898
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6483988037341495e-05,
+      "loss": 0.2209,
+      "step": 6899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6471068621277103e-05,
+      "loss": 0.1965,
+      "step": 6900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6458153271612627e-05,
+      "loss": 0.18,
+      "step": 6901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6445241989914455e-05,
+      "loss": 0.1991,
+      "step": 6902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6432334777748442e-05,
+      "loss": 0.1806,
+      "step": 6903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.641943163668001e-05,
+      "loss": 0.2152,
+      "step": 6904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.640653256827403e-05,
+      "loss": 0.2681,
+      "step": 6905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6393637574094913e-05,
+      "loss": 0.1936,
+      "step": 6906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6380746655706585e-05,
+      "loss": 0.2238,
+      "step": 6907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.636785981467243e-05,
+      "loss": 0.2498,
+      "step": 6908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.635497705255539e-05,
+      "loss": 0.2251,
+      "step": 6909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.63420983709179e-05,
+      "loss": 0.2061,
+      "step": 6910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6329223771321865e-05,
+      "loss": 0.2003,
+      "step": 6911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.631635325532873e-05,
+      "loss": 0.2631,
+      "step": 6912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6303486824499458e-05,
+      "loss": 0.2582,
+      "step": 6913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6290624480394474e-05,
+      "loss": 0.1945,
+      "step": 6914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6277766224573705e-05,
+      "loss": 0.2166,
+      "step": 6915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.626491205859665e-05,
+      "loss": 0.2526,
+      "step": 6916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.625206198402222e-05,
+      "loss": 0.1674,
+      "step": 6917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6239216002408907e-05,
+      "loss": 0.2949,
+      "step": 6918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.622637411531468e-05,
+      "loss": 0.1806,
+      "step": 6919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6213536324296984e-05,
+      "loss": 0.1683,
+      "step": 6920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.62007026309128e-05,
+      "loss": 0.198,
+      "step": 6921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6187873036718625e-05,
+      "loss": 0.164,
+      "step": 6922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.61750475432704e-05,
+      "loss": 0.22,
+      "step": 6923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.616222615212363e-05,
+      "loss": 0.1609,
+      "step": 6924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.614940886483328e-05,
+      "loss": 0.2332,
+      "step": 6925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6136595682953846e-05,
+      "loss": 0.176,
+      "step": 6926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6123786608039333e-05,
+      "loss": 0.1851,
+      "step": 6927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6110981641643185e-05,
+      "loss": 0.2098,
+      "step": 6928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6098180785318425e-05,
+      "loss": 0.275,
+      "step": 6929
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6085384040617547e-05,
+      "loss": 0.2586,
+      "step": 6930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6072591409092537e-05,
+      "loss": 0.1998,
+      "step": 6931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.605980289229487e-05,
+      "loss": 0.1881,
+      "step": 6932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6047018491775582e-05,
+      "loss": 0.2751,
+      "step": 6933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.603423820908513e-05,
+      "loss": 0.206,
+      "step": 6934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6021462045773523e-05,
+      "loss": 0.2121,
+      "step": 6935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.600869000339028e-05,
+      "loss": 0.1643,
+      "step": 6936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5995922083484368e-05,
+      "loss": 0.1668,
+      "step": 6937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5983158287604292e-05,
+      "loss": 0.1755,
+      "step": 6938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5970398617298076e-05,
+      "loss": 0.1281,
+      "step": 6939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5957643074113183e-05,
+      "loss": 0.1813,
+      "step": 6940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5944891659596617e-05,
+      "loss": 0.1943,
+      "step": 6941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5932144375294898e-05,
+      "loss": 0.2295,
+      "step": 6942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5919401222753986e-05,
+      "loss": 0.2326,
+      "step": 6943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5906662203519412e-05,
+      "loss": 0.1939,
+      "step": 6944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5893927319136127e-05,
+      "loss": 0.191,
+      "step": 6945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5881196571148643e-05,
+      "loss": 0.2066,
+      "step": 6946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5868469961100968e-05,
+      "loss": 0.1826,
+      "step": 6947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.585574749053657e-05,
+      "loss": 0.2159,
+      "step": 6948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.584302916099842e-05,
+      "loss": 0.2053,
+      "step": 6949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5830314974029015e-05,
+      "loss": 0.2661,
+      "step": 6950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.581760493117035e-05,
+      "loss": 0.1801,
+      "step": 6951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5804899033963875e-05,
+      "loss": 0.1573,
+      "step": 6952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.57921972839506e-05,
+      "loss": 0.1502,
+      "step": 6953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.577949968267096e-05,
+      "loss": 0.2231,
+      "step": 6954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.576680623166495e-05,
+      "loss": 0.1696,
+      "step": 6955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5754116932472035e-05,
+      "loss": 0.1707,
+      "step": 6956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.574143178663116e-05,
+      "loss": 0.2094,
+      "step": 6957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.572875079568079e-05,
+      "loss": 0.2074,
+      "step": 6958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5716073961158907e-05,
+      "loss": 0.1924,
+      "step": 6959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5703401284602913e-05,
+      "loss": 0.2072,
+      "step": 6960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5690732767549794e-05,
+      "loss": 0.1914,
+      "step": 6961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5678068411535983e-05,
+      "loss": 0.2124,
+      "step": 6962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5665408218097417e-05,
+      "loss": 0.2326,
+      "step": 6963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.565275218876951e-05,
+      "loss": 0.217,
+      "step": 6964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5640100325087216e-05,
+      "loss": 0.246,
+      "step": 6965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5627452628584933e-05,
+      "loss": 0.1328,
+      "step": 6966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5614809100796585e-05,
+      "loss": 0.2546,
+      "step": 6967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5602169743255605e-05,
+      "loss": 0.188,
+      "step": 6968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5589534557494867e-05,
+      "loss": 0.2054,
+      "step": 6969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5576903545046784e-05,
+      "loss": 0.1969,
+      "step": 6970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.556427670744327e-05,
+      "loss": 0.2589,
+      "step": 6971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.555165404621567e-05,
+      "loss": 0.2193,
+      "step": 6972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5539035562894905e-05,
+      "loss": 0.2608,
+      "step": 6973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.552642125901132e-05,
+      "loss": 0.2069,
+      "step": 6974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5513811136094787e-05,
+      "loss": 0.1881,
+      "step": 6975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.550120519567469e-05,
+      "loss": 0.262,
+      "step": 6976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5488603439279842e-05,
+      "loss": 0.1929,
+      "step": 6977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5476005868438614e-05,
+      "loss": 0.206,
+      "step": 6978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5463412484678857e-05,
+      "loss": 0.187,
+      "step": 6979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5450823289527876e-05,
+      "loss": 0.212,
+      "step": 6980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.543823828451249e-05,
+      "loss": 0.3393,
+      "step": 6981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5425657471159037e-05,
+      "loss": 0.2681,
+      "step": 6982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.541308085099329e-05,
+      "loss": 0.2047,
+      "step": 6983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5400508425540562e-05,
+      "loss": 0.1866,
+      "step": 6984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5387940196325657e-05,
+      "loss": 0.259,
+      "step": 6985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5375376164872825e-05,
+      "loss": 0.1831,
+      "step": 6986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.536281633270585e-05,
+      "loss": 0.2722,
+      "step": 6987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.535026070134801e-05,
+      "loss": 0.1729,
+      "step": 6988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5337709272322016e-05,
+      "loss": 0.2231,
+      "step": 6989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.532516204715014e-05,
+      "loss": 0.2264,
+      "step": 6990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5312619027354114e-05,
+      "loss": 0.2295,
+      "step": 6991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5300080214455137e-05,
+      "loss": 0.2448,
+      "step": 6992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5287545609973954e-05,
+      "loss": 0.1605,
+      "step": 6993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5275015215430727e-05,
+      "loss": 0.1834,
+      "step": 6994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5262489032345168e-05,
+      "loss": 0.2275,
+      "step": 6995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.524996706223647e-05,
+      "loss": 0.2075,
+      "step": 6996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.523744930662329e-05,
+      "loss": 0.2432,
+      "step": 6997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5224935767023763e-05,
+      "loss": 0.2494,
+      "step": 6998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5212426444955568e-05,
+      "loss": 0.1655,
+      "step": 6999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5199921341935814e-05,
+      "loss": 0.2392,
+      "step": 7000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5187420459481133e-05,
+      "loss": 0.266,
+      "step": 7001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5174923799107654e-05,
+      "loss": 0.1211,
+      "step": 7002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5162431362330948e-05,
+      "loss": 0.168,
+      "step": 7003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5149943150666112e-05,
+      "loss": 0.2647,
+      "step": 7004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5137459165627738e-05,
+      "loss": 0.1764,
+      "step": 7005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5124979408729861e-05,
+      "loss": 0.1769,
+      "step": 7006
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5112503881486039e-05,
+      "loss": 0.2133,
+      "step": 7007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5100032585409336e-05,
+      "loss": 0.2487,
+      "step": 7008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5087565522012226e-05,
+      "loss": 0.2316,
+      "step": 7009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5075102692806741e-05,
+      "loss": 0.2246,
+      "step": 7010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5062644099304402e-05,
+      "loss": 0.2818,
+      "step": 7011
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5050189743016163e-05,
+      "loss": 0.2218,
+      "step": 7012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5037739625452484e-05,
+      "loss": 0.211,
+      "step": 7013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5025293748123354e-05,
+      "loss": 0.2209,
+      "step": 7014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5012852112538166e-05,
+      "loss": 0.1776,
+      "step": 7015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.500041472020588e-05,
+      "loss": 0.1892,
+      "step": 7016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4987981572634912e-05,
+      "loss": 0.267,
+      "step": 7017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4975552671333131e-05,
+      "loss": 0.187,
+      "step": 7018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4963128017807949e-05,
+      "loss": 0.1396,
+      "step": 7019
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4950707613566201e-05,
+      "loss": 0.2057,
+      "step": 7020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4938291460114246e-05,
+      "loss": 0.2688,
+      "step": 7021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4925879558957945e-05,
+      "loss": 0.2178,
+      "step": 7022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4913471911602583e-05,
+      "loss": 0.2489,
+      "step": 7023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.490106851955298e-05,
+      "loss": 0.2037,
+      "step": 7024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4888669384313436e-05,
+      "loss": 0.2933,
+      "step": 7025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4876274507387695e-05,
+      "loss": 0.2105,
+      "step": 7026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4863883890279018e-05,
+      "loss": 0.2494,
+      "step": 7027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4851497534490172e-05,
+      "loss": 0.2127,
+      "step": 7028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4839115441523355e-05,
+      "loss": 0.2392,
+      "step": 7029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.482673761288026e-05,
+      "loss": 0.1723,
+      "step": 7030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4814364050062102e-05,
+      "loss": 0.2206,
+      "step": 7031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.480199475456952e-05,
+      "loss": 0.1669,
+      "step": 7032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4789629727902682e-05,
+      "loss": 0.2972,
+      "step": 7033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4777268971561237e-05,
+      "loss": 0.2489,
+      "step": 7034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4764912487044274e-05,
+      "loss": 0.2039,
+      "step": 7035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4752560275850402e-05,
+      "loss": 0.1934,
+      "step": 7036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4740212339477721e-05,
+      "loss": 0.257,
+      "step": 7037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4727868679423763e-05,
+      "loss": 0.1832,
+      "step": 7038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4715529297185598e-05,
+      "loss": 0.1952,
+      "step": 7039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4703194194259722e-05,
+      "loss": 0.1437,
+      "step": 7040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4690863372142156e-05,
+      "loss": 0.3028,
+      "step": 7041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.46785368323284e-05,
+      "loss": 0.2105,
+      "step": 7042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4666214576313386e-05,
+      "loss": 0.2281,
+      "step": 7043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4653896605591582e-05,
+      "loss": 0.2191,
+      "step": 7044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4641582921656937e-05,
+      "loss": 0.2226,
+      "step": 7045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4629273526002829e-05,
+      "loss": 0.2006,
+      "step": 7046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.461696842012214e-05,
+      "loss": 0.238,
+      "step": 7047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4604667605507271e-05,
+      "loss": 0.2192,
+      "step": 7048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.459237108365003e-05,
+      "loss": 0.2016,
+      "step": 7049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4580078856041757e-05,
+      "loss": 0.3099,
+      "step": 7050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.456779092417328e-05,
+      "loss": 0.2796,
+      "step": 7051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4555507289534852e-05,
+      "loss": 0.2315,
+      "step": 7052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.454322795361625e-05,
+      "loss": 0.1689,
+      "step": 7053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4530952917906732e-05,
+      "loss": 0.1473,
+      "step": 7054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4518682183894982e-05,
+      "loss": 0.1753,
+      "step": 7055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4506415753069219e-05,
+      "loss": 0.1265,
+      "step": 7056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.449415362691714e-05,
+      "loss": 0.2105,
+      "step": 7057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.448189580692586e-05,
+      "loss": 0.1849,
+      "step": 7058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4469642294582048e-05,
+      "loss": 0.2141,
+      "step": 7059
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.445739309137178e-05,
+      "loss": 0.1614,
+      "step": 7060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4445148198780662e-05,
+      "loss": 0.2095,
+      "step": 7061
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4432907618293768e-05,
+      "loss": 0.2045,
+      "step": 7062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4420671351395631e-05,
+      "loss": 0.2091,
+      "step": 7063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4408439399570257e-05,
+      "loss": 0.1887,
+      "step": 7064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4396211764301143e-05,
+      "loss": 0.1863,
+      "step": 7065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.438398844707129e-05,
+      "loss": 0.152,
+      "step": 7066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4371769449363103e-05,
+      "loss": 0.2798,
+      "step": 7067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4359554772658552e-05,
+      "loss": 0.3209,
+      "step": 7068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.434734441843899e-05,
+      "loss": 0.1708,
+      "step": 7069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4335138388185315e-05,
+      "loss": 0.1584,
+      "step": 7070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4322936683377903e-05,
+      "loss": 0.1705,
+      "step": 7071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4310739305496534e-05,
+      "loss": 0.1874,
+      "step": 7072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4298546256020534e-05,
+      "loss": 0.1677,
+      "step": 7073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4286357536428696e-05,
+      "loss": 0.2157,
+      "step": 7074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4274173148199243e-05,
+      "loss": 0.1435,
+      "step": 7075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.426199309280991e-05,
+      "loss": 0.1903,
+      "step": 7076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4249817371737917e-05,
+      "loss": 0.2043,
+      "step": 7077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4237645986459924e-05,
+      "loss": 0.1869,
+      "step": 7078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4225478938452064e-05,
+      "loss": 0.1523,
+      "step": 7079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4213316229189999e-05,
+      "loss": 0.2487,
+      "step": 7080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4201157860148784e-05,
+      "loss": 0.1955,
+      "step": 7081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4189003832803011e-05,
+      "loss": 0.1496,
+      "step": 7082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4176854148626744e-05,
+      "loss": 0.1967,
+      "step": 7083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4164708809093463e-05,
+      "loss": 0.2308,
+      "step": 7084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4152567815676176e-05,
+      "loss": 0.2507,
+      "step": 7085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4140431169847368e-05,
+      "loss": 0.2422,
+      "step": 7086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4128298873078932e-05,
+      "loss": 0.1376,
+      "step": 7087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4116170926842321e-05,
+      "loss": 0.1561,
+      "step": 7088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4104047332608377e-05,
+      "loss": 0.1402,
+      "step": 7089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4091928091847479e-05,
+      "loss": 0.1633,
+      "step": 7090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4079813206029452e-05,
+      "loss": 0.1749,
+      "step": 7091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4067702676623579e-05,
+      "loss": 0.1742,
+      "step": 7092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4055596505098634e-05,
+      "loss": 0.2143,
+      "step": 7093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4043494692922887e-05,
+      "loss": 0.2692,
+      "step": 7094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4031397241563999e-05,
+      "loss": 0.2774,
+      "step": 7095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.401930415248917e-05,
+      "loss": 0.2483,
+      "step": 7096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4007215427165083e-05,
+      "loss": 0.1927,
+      "step": 7097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.399513106705782e-05,
+      "loss": 0.2531,
+      "step": 7098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3983051073632997e-05,
+      "loss": 0.2567,
+      "step": 7099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.39709754483557e-05,
+      "loss": 0.2395,
+      "step": 7100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3958904192690424e-05,
+      "loss": 0.2047,
+      "step": 7101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3946837308101195e-05,
+      "loss": 0.2147,
+      "step": 7102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3934774796051508e-05,
+      "loss": 0.2047,
+      "step": 7103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.392271665800427e-05,
+      "loss": 0.2935,
+      "step": 7104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3910662895421922e-05,
+      "loss": 0.2305,
+      "step": 7105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.389861350976635e-05,
+      "loss": 0.176,
+      "step": 7106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3886568502498887e-05,
+      "loss": 0.212,
+      "step": 7107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3874527875080378e-05,
+      "loss": 0.2162,
+      "step": 7108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3862491628971097e-05,
+      "loss": 0.2401,
+      "step": 7109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3850459765630803e-05,
+      "loss": 0.226,
+      "step": 7110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.383843228651875e-05,
+      "loss": 0.2064,
+      "step": 7111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.382640919309362e-05,
+      "loss": 0.2026,
+      "step": 7112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3814390486813555e-05,
+      "loss": 0.2019,
+      "step": 7113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3802376169136228e-05,
+      "loss": 0.2511,
+      "step": 7114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3790366241518698e-05,
+      "loss": 0.1953,
+      "step": 7115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3778360705417554e-05,
+      "loss": 0.1977,
+      "step": 7116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.376635956228885e-05,
+      "loss": 0.183,
+      "step": 7117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.375436281358805e-05,
+      "loss": 0.2283,
+      "step": 7118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3742370460770143e-05,
+      "loss": 0.217,
+      "step": 7119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3730382505289586e-05,
+      "loss": 0.2011,
+      "step": 7120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3718398948600241e-05,
+      "loss": 0.1843,
+      "step": 7121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3706419792155505e-05,
+      "loss": 0.1653,
+      "step": 7122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3694445037408221e-05,
+      "loss": 0.1729,
+      "step": 7123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3682474685810658e-05,
+      "loss": 0.174,
+      "step": 7124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3670508738814614e-05,
+      "loss": 0.1699,
+      "step": 7125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3658547197871318e-05,
+      "loss": 0.2531,
+      "step": 7126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3646590064431475e-05,
+      "loss": 0.1824,
+      "step": 7127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3634637339945216e-05,
+      "loss": 0.2116,
+      "step": 7128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3622689025862217e-05,
+      "loss": 0.2098,
+      "step": 7129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3610745123631535e-05,
+      "loss": 0.1136,
+      "step": 7130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3598805634701745e-05,
+      "loss": 0.2156,
+      "step": 7131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3586870560520892e-05,
+      "loss": 0.2433,
+      "step": 7132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3574939902536432e-05,
+      "loss": 0.2222,
+      "step": 7133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3563013662195351e-05,
+      "loss": 0.2131,
+      "step": 7134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3551091840944041e-05,
+      "loss": 0.2234,
+      "step": 7135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3539174440228391e-05,
+      "loss": 0.1403,
+      "step": 7136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3527261461493767e-05,
+      "loss": 0.1773,
+      "step": 7137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3515352906184947e-05,
+      "loss": 0.2023,
+      "step": 7138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3503448775746224e-05,
+      "loss": 0.1532,
+      "step": 7139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3491549071621345e-05,
+      "loss": 0.2609,
+      "step": 7140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3479653795253483e-05,
+      "loss": 0.185,
+      "step": 7141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3467762948085316e-05,
+      "loss": 0.1454,
+      "step": 7142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.345587653155898e-05,
+      "loss": 0.1949,
+      "step": 7143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3443994547116062e-05,
+      "loss": 0.1942,
+      "step": 7144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.343211699619758e-05,
+      "loss": 0.1682,
+      "step": 7145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3420243880244088e-05,
+      "loss": 0.1466,
+      "step": 7146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3408375200695528e-05,
+      "loss": 0.2316,
+      "step": 7147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.339651095899136e-05,
+      "loss": 0.242,
+      "step": 7148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3384651156570483e-05,
+      "loss": 0.1941,
+      "step": 7149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.337279579487124e-05,
+      "loss": 0.1787,
+      "step": 7150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.336094487533147e-05,
+      "loss": 0.1872,
+      "step": 7151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.334909839938846e-05,
+      "loss": 0.1901,
+      "step": 7152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3337256368478934e-05,
+      "loss": 0.2233,
+      "step": 7153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3325418784039123e-05,
+      "loss": 0.2398,
+      "step": 7154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3313585647504666e-05,
+      "loss": 0.2096,
+      "step": 7155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3301756960310702e-05,
+      "loss": 0.2032,
+      "step": 7156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3289932723891835e-05,
+      "loss": 0.1668,
+      "step": 7157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.327811293968208e-05,
+      "loss": 0.242,
+      "step": 7158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3266297609114964e-05,
+      "loss": 0.2389,
+      "step": 7159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3254486733623462e-05,
+      "loss": 0.1033,
+      "step": 7160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3242680314639993e-05,
+      "loss": 0.2487,
+      "step": 7161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3230878353596422e-05,
+      "loss": 0.168,
+      "step": 7162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3219080851924131e-05,
+      "loss": 0.153,
+      "step": 7163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3207287811053892e-05,
+      "loss": 0.2683,
+      "step": 7164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3195499232415986e-05,
+      "loss": 0.1959,
+      "step": 7165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3183715117440142e-05,
+      "loss": 0.1723,
+      "step": 7166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3171935467555523e-05,
+      "loss": 0.2331,
+      "step": 7167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.316016028419078e-05,
+      "loss": 0.262,
+      "step": 7168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3148389568774023e-05,
+      "loss": 0.2274,
+      "step": 7169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3136623322732782e-05,
+      "loss": 0.3001,
+      "step": 7170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3124861547494083e-05,
+      "loss": 0.1771,
+      "step": 7171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3113104244484414e-05,
+      "loss": 0.1938,
+      "step": 7172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.310135141512968e-05,
+      "loss": 0.2246,
+      "step": 7173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3089603060855288e-05,
+      "loss": 0.2002,
+      "step": 7174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3077859183086062e-05,
+      "loss": 0.1954,
+      "step": 7175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3066119783246322e-05,
+      "loss": 0.2089,
+      "step": 7176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.305438486275981e-05,
+      "loss": 0.2137,
+      "step": 7177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.304265442304976e-05,
+      "loss": 0.2508,
+      "step": 7178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3030928465538822e-05,
+      "loss": 0.2247,
+      "step": 7179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.301920699164913e-05,
+      "loss": 0.243,
+      "step": 7180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3007490002802291e-05,
+      "loss": 0.2284,
+      "step": 7181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2995777500419315e-05,
+      "loss": 0.203,
+      "step": 7182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.298406948592072e-05,
+      "loss": 0.2109,
+      "step": 7183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2972365960726435e-05,
+      "loss": 0.2559,
+      "step": 7184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2960666926255883e-05,
+      "loss": 0.1939,
+      "step": 7185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2948972383927937e-05,
+      "loss": 0.2084,
+      "step": 7186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2937282335160895e-05,
+      "loss": 0.1775,
+      "step": 7187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2925596781372528e-05,
+      "loss": 0.1789,
+      "step": 7188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2913915723980091e-05,
+      "loss": 0.199,
+      "step": 7189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2902239164400237e-05,
+      "loss": 0.1823,
+      "step": 7190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2890567104049112e-05,
+      "loss": 0.1845,
+      "step": 7191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2878899544342327e-05,
+      "loss": 0.2111,
+      "step": 7192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.286723648669491e-05,
+      "loss": 0.1811,
+      "step": 7193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2855577932521352e-05,
+      "loss": 0.2018,
+      "step": 7194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2843923883235625e-05,
+      "loss": 0.2395,
+      "step": 7195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.283227434025111e-05,
+      "loss": 0.2188,
+      "step": 7196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2820629304980686e-05,
+      "loss": 0.2307,
+      "step": 7197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2808988778836684e-05,
+      "loss": 0.2312,
+      "step": 7198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.279735276323083e-05,
+      "loss": 0.1968,
+      "step": 7199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2785721259574368e-05,
+      "loss": 0.1623,
+      "step": 7200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2774094269277987e-05,
+      "loss": 0.2542,
+      "step": 7201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2762471793751774e-05,
+      "loss": 0.1898,
+      "step": 7202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2750853834405341e-05,
+      "loss": 0.182,
+      "step": 7203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2739240392647689e-05,
+      "loss": 0.1709,
+      "step": 7204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2727631469887314e-05,
+      "loss": 0.2329,
+      "step": 7205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2716027067532165e-05,
+      "loss": 0.2769,
+      "step": 7206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2704427186989598e-05,
+      "loss": 0.2292,
+      "step": 7207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2692831829666462e-05,
+      "loss": 0.1533,
+      "step": 7208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2681240996969084e-05,
+      "loss": 0.1474,
+      "step": 7209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.266965469030314e-05,
+      "loss": 0.1776,
+      "step": 7210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2658072911073842e-05,
+      "loss": 0.1695,
+      "step": 7211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2646495660685864e-05,
+      "loss": 0.182,
+      "step": 7212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2634922940543254e-05,
+      "loss": 0.2012,
+      "step": 7213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2623354752049582e-05,
+      "loss": 0.228,
+      "step": 7214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2611791096607845e-05,
+      "loss": 0.1599,
+      "step": 7215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2600231975620468e-05,
+      "loss": 0.1955,
+      "step": 7216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.258867739048935e-05,
+      "loss": 0.2922,
+      "step": 7217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2577127342615858e-05,
+      "loss": 0.2399,
+      "step": 7218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2565581833400752e-05,
+      "loss": 0.2276,
+      "step": 7219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2554040864244304e-05,
+      "loss": 0.1797,
+      "step": 7220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2542504436546181e-05,
+      "loss": 0.2596,
+      "step": 7221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.253097255170554e-05,
+      "loss": 0.2351,
+      "step": 7222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2519445211120979e-05,
+      "loss": 0.174,
+      "step": 7223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2507922416190509e-05,
+      "loss": 0.158,
+      "step": 7224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2496404168311642e-05,
+      "loss": 0.1749,
+      "step": 7225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2484890468881328e-05,
+      "loss": 0.1495,
+      "step": 7226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.247338131929593e-05,
+      "loss": 0.1276,
+      "step": 7227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.246187672095127e-05,
+      "loss": 0.1965,
+      "step": 7228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.245037667524266e-05,
+      "loss": 0.1551,
+      "step": 7229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.24388811835648e-05,
+      "loss": 0.2944,
+      "step": 7230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2427390247311882e-05,
+      "loss": 0.2084,
+      "step": 7231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2415903867877548e-05,
+      "loss": 0.1823,
+      "step": 7232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2404422046654835e-05,
+      "loss": 0.2396,
+      "step": 7233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2392944785036276e-05,
+      "loss": 0.1821,
+      "step": 7234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2381472084413859e-05,
+      "loss": 0.1517,
+      "step": 7235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.237000394617896e-05,
+      "loss": 0.274,
+      "step": 7236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.235854037172246e-05,
+      "loss": 0.2088,
+      "step": 7237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2347081362434675e-05,
+      "loss": 0.1827,
+      "step": 7238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2335626919705329e-05,
+      "loss": 0.1477,
+      "step": 7239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2324177044923651e-05,
+      "loss": 0.1774,
+      "step": 7240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2312731739478257e-05,
+      "loss": 0.1612,
+      "step": 7241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2301291004757265e-05,
+      "loss": 0.2367,
+      "step": 7242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.228985484214818e-05,
+      "loss": 0.187,
+      "step": 7243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2278423253038019e-05,
+      "loss": 0.2313,
+      "step": 7244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2266996238813173e-05,
+      "loss": 0.3027,
+      "step": 7245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2255573800859532e-05,
+      "loss": 0.1795,
+      "step": 7246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2244155940562424e-05,
+      "loss": 0.1481,
+      "step": 7247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2232742659306589e-05,
+      "loss": 0.2048,
+      "step": 7248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.222133395847626e-05,
+      "loss": 0.1456,
+      "step": 7249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.220992983945506e-05,
+      "loss": 0.2338,
+      "step": 7250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2198530303626087e-05,
+      "loss": 0.2259,
+      "step": 7251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2187135352371914e-05,
+      "loss": 0.2038,
+      "step": 7252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2175744987074478e-05,
+      "loss": 0.222,
+      "step": 7253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2164359209115234e-05,
+      "loss": 0.215,
+      "step": 7254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.215297801987506e-05,
+      "loss": 0.2767,
+      "step": 7255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2141601420734245e-05,
+      "loss": 0.183,
+      "step": 7256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2130229413072558e-05,
+      "loss": 0.2244,
+      "step": 7257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2118861998269232e-05,
+      "loss": 0.2123,
+      "step": 7258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2107499177702853e-05,
+      "loss": 0.233,
+      "step": 7259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2096140952751528e-05,
+      "loss": 0.2029,
+      "step": 7260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2084787324792813e-05,
+      "loss": 0.205,
+      "step": 7261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2073438295203648e-05,
+      "loss": 0.1894,
+      "step": 7262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2062093865360458e-05,
+      "loss": 0.1859,
+      "step": 7263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2050754036639112e-05,
+      "loss": 0.1866,
+      "step": 7264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2039418810414882e-05,
+      "loss": 0.1414,
+      "step": 7265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2028088188062525e-05,
+      "loss": 0.1995,
+      "step": 7266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.201676217095623e-05,
+      "loss": 0.1879,
+      "step": 7267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.20054407604696e-05,
+      "loss": 0.2049,
+      "step": 7268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1994123957975722e-05,
+      "loss": 0.1437,
+      "step": 7269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.198281176484708e-05,
+      "loss": 0.1759,
+      "step": 7270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1971504182455628e-05,
+      "loss": 0.2202,
+      "step": 7271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1960201212172767e-05,
+      "loss": 0.1774,
+      "step": 7272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1948902855369299e-05,
+      "loss": 0.264,
+      "step": 7273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1937609113415505e-05,
+      "loss": 0.2335,
+      "step": 7274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1926319987681111e-05,
+      "loss": 0.1583,
+      "step": 7275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1915035479535253e-05,
+      "loss": 0.2199,
+      "step": 7276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1903755590346499e-05,
+      "loss": 0.1188,
+      "step": 7277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1892480321482912e-05,
+      "loss": 0.1965,
+      "step": 7278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1881209674311933e-05,
+      "loss": 0.2198,
+      "step": 7279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1869943650200477e-05,
+      "loss": 0.137,
+      "step": 7280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1858682250514913e-05,
+      "loss": 0.1627,
+      "step": 7281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1847425476620989e-05,
+      "loss": 0.1888,
+      "step": 7282
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1836173329883954e-05,
+      "loss": 0.1773,
+      "step": 7283
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1824925811668486e-05,
+      "loss": 0.1663,
+      "step": 7284
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1813682923338653e-05,
+      "loss": 0.1871,
+      "step": 7285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1802444666258016e-05,
+      "loss": 0.2125,
+      "step": 7286
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1791211041789563e-05,
+      "loss": 0.1906,
+      "step": 7287
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.177998205129569e-05,
+      "loss": 0.2003,
+      "step": 7288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1768757696138277e-05,
+      "loss": 0.2027,
+      "step": 7289
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1757537977678589e-05,
+      "loss": 0.1506,
+      "step": 7290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.174632289727739e-05,
+      "loss": 0.2687,
+      "step": 7291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.173511245629481e-05,
+      "loss": 0.2439,
+      "step": 7292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.17239066560905e-05,
+      "loss": 0.1628,
+      "step": 7293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1712705498023458e-05,
+      "loss": 0.2411,
+      "step": 7294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1701508983452186e-05,
+      "loss": 0.1671,
+      "step": 7295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1690317113734618e-05,
+      "loss": 0.2408,
+      "step": 7296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1679129890228069e-05,
+      "loss": 0.2434,
+      "step": 7297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1667947314289374e-05,
+      "loss": 0.2313,
+      "step": 7298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1656769387274712e-05,
+      "loss": 0.1373,
+      "step": 7299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1645596110539775e-05,
+      "loss": 0.2105,
+      "step": 7300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.163442748543967e-05,
+      "loss": 0.1798,
+      "step": 7301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1623263513328903e-05,
+      "loss": 0.2462,
+      "step": 7302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1612104195561457e-05,
+      "loss": 0.1899,
+      "step": 7303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1600949533490758e-05,
+      "loss": 0.1426,
+      "step": 7304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1589799528469613e-05,
+      "loss": 0.1827,
+      "step": 7305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1578654181850312e-05,
+      "loss": 0.1693,
+      "step": 7306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1567513494984583e-05,
+      "loss": 0.1787,
+      "step": 7307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1556377469223561e-05,
+      "loss": 0.1995,
+      "step": 7308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1545246105917806e-05,
+      "loss": 0.1915,
+      "step": 7309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1534119406417366e-05,
+      "loss": 0.1745,
+      "step": 7310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.152299737207166e-05,
+      "loss": 0.2084,
+      "step": 7311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1511880004229592e-05,
+      "loss": 0.1742,
+      "step": 7312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1500767304239484e-05,
+      "loss": 0.2212,
+      "step": 7313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1489659273449072e-05,
+      "loss": 0.2442,
+      "step": 7314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1478555913205562e-05,
+      "loss": 0.1364,
+      "step": 7315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1467457224855544e-05,
+      "loss": 0.2242,
+      "step": 7316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1456363209745092e-05,
+      "loss": 0.1902,
+      "step": 7317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1445273869219698e-05,
+      "loss": 0.2595,
+      "step": 7318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1434189204624252e-05,
+      "loss": 0.2009,
+      "step": 7319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.142310921730313e-05,
+      "loss": 0.1822,
+      "step": 7320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1412033908600122e-05,
+      "loss": 0.2513,
+      "step": 7321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1400963279858417e-05,
+      "loss": 0.2535,
+      "step": 7322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1389897332420685e-05,
+      "loss": 0.2728,
+      "step": 7323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1378836067629029e-05,
+      "loss": 0.1679,
+      "step": 7324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.136777948682491e-05,
+      "loss": 0.1731,
+      "step": 7325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1356727591349303e-05,
+      "loss": 0.2149,
+      "step": 7326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1345680382542596e-05,
+      "loss": 0.2811,
+      "step": 7327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1334637861744574e-05,
+      "loss": 0.1803,
+      "step": 7328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.132360003029449e-05,
+      "loss": 0.1891,
+      "step": 7329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1312566889531029e-05,
+      "loss": 0.1809,
+      "step": 7330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1301538440792264e-05,
+      "loss": 0.1913,
+      "step": 7331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1290514685415749e-05,
+      "loss": 0.2152,
+      "step": 7332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1279495624738456e-05,
+      "loss": 0.1951,
+      "step": 7333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1268481260096752e-05,
+      "loss": 0.2505,
+      "step": 7334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1257471592826502e-05,
+      "loss": 0.222,
+      "step": 7335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1246466624262925e-05,
+      "loss": 0.1396,
+      "step": 7336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.123546635574072e-05,
+      "loss": 0.1913,
+      "step": 7337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1224470788594022e-05,
+      "loss": 0.1944,
+      "step": 7338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1213479924156345e-05,
+      "loss": 0.2469,
+      "step": 7339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.120249376376069e-05,
+      "loss": 0.2002,
+      "step": 7340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1191512308739444e-05,
+      "loss": 0.1751,
+      "step": 7341
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1180535560424465e-05,
+      "loss": 0.2217,
+      "step": 7342
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1169563520146986e-05,
+      "loss": 0.2035,
+      "step": 7343
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.115859618923773e-05,
+      "loss": 0.2003,
+      "step": 7344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1147633569026788e-05,
+      "loss": 0.1855,
+      "step": 7345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1136675660843731e-05,
+      "loss": 0.2004,
+      "step": 7346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1125722466017547e-05,
+      "loss": 0.1687,
+      "step": 7347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.111477398587661e-05,
+      "loss": 0.2282,
+      "step": 7348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1103830221748773e-05,
+      "loss": 0.2407,
+      "step": 7349
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1092891174961324e-05,
+      "loss": 0.1579,
+      "step": 7350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1081956846840908e-05,
+      "loss": 0.1281,
+      "step": 7351
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1071027238713671e-05,
+      "loss": 0.1954,
+      "step": 7352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1060102351905172e-05,
+      "loss": 0.1486,
+      "step": 7353
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1049182187740348e-05,
+      "loss": 0.2094,
+      "step": 7354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1038266747543635e-05,
+      "loss": 0.1417,
+      "step": 7355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1027356032638836e-05,
+      "loss": 0.1548,
+      "step": 7356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.101645004434923e-05,
+      "loss": 0.2402,
+      "step": 7357
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1005548783997471e-05,
+      "loss": 0.1251,
+      "step": 7358
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0994652252905697e-05,
+      "loss": 0.2014,
+      "step": 7359
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0983760452395413e-05,
+      "loss": 0.204,
+      "step": 7360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0972873383787597e-05,
+      "loss": 0.2128,
+      "step": 7361
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0961991048402647e-05,
+      "loss": 0.2146,
+      "step": 7362
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0951113447560352e-05,
+      "loss": 0.2184,
+      "step": 7363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0940240582579974e-05,
+      "loss": 0.2284,
+      "step": 7364
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.092937245478015e-05,
+      "loss": 0.2295,
+      "step": 7365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0918509065478993e-05,
+      "loss": 0.1849,
+      "step": 7366
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.090765041599402e-05,
+      "loss": 0.2295,
+      "step": 7367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0896796507642154e-05,
+      "loss": 0.212,
+      "step": 7368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0885947341739766e-05,
+      "loss": 0.1826,
+      "step": 7369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.087510291960267e-05,
+      "loss": 0.2545,
+      "step": 7370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0864263242546047e-05,
+      "loss": 0.1633,
+      "step": 7371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0853428311884545e-05,
+      "loss": 0.1737,
+      "step": 7372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0842598128932268e-05,
+      "loss": 0.2431,
+      "step": 7373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.083177269500264e-05,
+      "loss": 0.1759,
+      "step": 7374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0820952011408608e-05,
+      "loss": 0.2363,
+      "step": 7375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0810136079462512e-05,
+      "loss": 0.2135,
+      "step": 7376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0799324900476094e-05,
+      "loss": 0.1928,
+      "step": 7377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0788518475760545e-05,
+      "loss": 0.2808,
+      "step": 7378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0777716806626486e-05,
+      "loss": 0.2456,
+      "step": 7379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0766919894383926e-05,
+      "loss": 0.2243,
+      "step": 7380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0756127740342326e-05,
+      "loss": 0.2693,
+      "step": 7381
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0745340345810572e-05,
+      "loss": 0.2084,
+      "step": 7382
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0734557712096943e-05,
+      "loss": 0.1905,
+      "step": 7383
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0723779840509186e-05,
+      "loss": 0.1597,
+      "step": 7384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0713006732354414e-05,
+      "loss": 0.2022,
+      "step": 7385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0702238388939213e-05,
+      "loss": 0.2056,
+      "step": 7386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0691474811569573e-05,
+      "loss": 0.2176,
+      "step": 7387
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0680716001550895e-05,
+      "loss": 0.1818,
+      "step": 7388
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0669961960188007e-05,
+      "loss": 0.1608,
+      "step": 7389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0659212688785203e-05,
+      "loss": 0.1868,
+      "step": 7390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0648468188646093e-05,
+      "loss": 0.2363,
+      "step": 7391
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.063772846107381e-05,
+      "loss": 0.2483,
+      "step": 7392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.062699350737088e-05,
+      "loss": 0.201,
+      "step": 7393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0616263328839216e-05,
+      "loss": 0.1775,
+      "step": 7394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.060553792678019e-05,
+      "loss": 0.2621,
+      "step": 7395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0594817302494597e-05,
+      "loss": 0.1458,
+      "step": 7396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0584101457282614e-05,
+      "loss": 0.1654,
+      "step": 7397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0573390392443872e-05,
+      "loss": 0.1957,
+      "step": 7398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0562684109277426e-05,
+      "loss": 0.1982,
+      "step": 7399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0551982609081706e-05,
+      "loss": 0.1815,
+      "step": 7400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0541285893154617e-05,
+      "loss": 0.1894,
+      "step": 7401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0530593962793472e-05,
+      "loss": 0.1585,
+      "step": 7402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0519906819294956e-05,
+      "loss": 0.198,
+      "step": 7403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0509224463955248e-05,
+      "loss": 0.1784,
+      "step": 7404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0498546898069866e-05,
+      "loss": 0.1794,
+      "step": 7405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0487874122933833e-05,
+      "loss": 0.1981,
+      "step": 7406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0477206139841506e-05,
+      "loss": 0.1727,
+      "step": 7407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0466542950086733e-05,
+      "loss": 0.1333,
+      "step": 7408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0455884554962725e-05,
+      "loss": 0.1997,
+      "step": 7409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0445230955762159e-05,
+      "loss": 0.2348,
+      "step": 7410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0434582153777084e-05,
+      "loss": 0.2417,
+      "step": 7411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0423938150299001e-05,
+      "loss": 0.2073,
+      "step": 7412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0413298946618833e-05,
+      "loss": 0.2232,
+      "step": 7413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0402664544026874e-05,
+      "loss": 0.2133,
+      "step": 7414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.039203494381289e-05,
+      "loss": 0.1663,
+      "step": 7415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0381410147266047e-05,
+      "loss": 0.2587,
+      "step": 7416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0370790155674903e-05,
+      "loss": 0.1283,
+      "step": 7417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0360174970327468e-05,
+      "loss": 0.2358,
+      "step": 7418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0349564592511163e-05,
+      "loss": 0.2014,
+      "step": 7419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0338959023512801e-05,
+      "loss": 0.1573,
+      "step": 7420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0328358264618632e-05,
+      "loss": 0.2375,
+      "step": 7421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0317762317114349e-05,
+      "loss": 0.2447,
+      "step": 7422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0307171182284981e-05,
+      "loss": 0.1561,
+      "step": 7423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0296584861415054e-05,
+      "loss": 0.2402,
+      "step": 7424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0286003355788488e-05,
+      "loss": 0.1628,
+      "step": 7425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0275426666688587e-05,
+      "loss": 0.157,
+      "step": 7426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0264854795398105e-05,
+      "loss": 0.1522,
+      "step": 7427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.025428774319922e-05,
+      "loss": 0.1568,
+      "step": 7428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.024372551137348e-05,
+      "loss": 0.1837,
+      "step": 7429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0233168101201907e-05,
+      "loss": 0.2159,
+      "step": 7430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0222615513964873e-05,
+      "loss": 0.2079,
+      "step": 7431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.021206775094221e-05,
+      "loss": 0.2224,
+      "step": 7432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0201524813413182e-05,
+      "loss": 0.1646,
+      "step": 7433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0190986702656403e-05,
+      "loss": 0.2357,
+      "step": 7434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0180453419949948e-05,
+      "loss": 0.2006,
+      "step": 7435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.016992496657132e-05,
+      "loss": 0.2128,
+      "step": 7436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.015940134379738e-05,
+      "loss": 0.198,
+      "step": 7437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0148882552904454e-05,
+      "loss": 0.1738,
+      "step": 7438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.013836859516829e-05,
+      "loss": 0.205,
+      "step": 7439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.012785947186397e-05,
+      "loss": 0.1525,
+      "step": 7440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0117355184266069e-05,
+      "loss": 0.1634,
+      "step": 7441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0106855733648568e-05,
+      "loss": 0.1704,
+      "step": 7442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0096361121284808e-05,
+      "loss": 0.2543,
+      "step": 7443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0085871348447606e-05,
+      "loss": 0.1656,
+      "step": 7444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0075386416409171e-05,
+      "loss": 0.2497,
+      "step": 7445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0064906326441093e-05,
+      "loss": 0.1436,
+      "step": 7446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0054431079814413e-05,
+      "loss": 0.1533,
+      "step": 7447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0043960677799586e-05,
+      "loss": 0.1743,
+      "step": 7448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0033495121666441e-05,
+      "loss": 0.1721,
+      "step": 7449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0023034412684274e-05,
+      "loss": 0.2183,
+      "step": 7450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0012578552121726e-05,
+      "loss": 0.1619,
+      "step": 7451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0002127541246909e-05,
+      "loss": 0.2184,
+      "step": 7452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.991681381327339e-06,
+      "loss": 0.1983,
+      "step": 7453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.981240073629894e-06,
+      "loss": 0.1866,
+      "step": 7454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.970803619420938e-06,
+      "loss": 0.1855,
+      "step": 7455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.960372019966175e-06,
+      "loss": 0.1952,
+      "step": 7456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.949945276530781e-06,
+      "loss": 0.1341,
+      "step": 7457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.939523390379284e-06,
+      "loss": 0.1736,
+      "step": 7458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.929106362775681e-06,
+      "loss": 0.2478,
+      "step": 7459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.918694194983336e-06,
+      "loss": 0.239,
+      "step": 7460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.908286888265039e-06,
+      "loss": 0.2382,
+      "step": 7461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.897884443883015e-06,
+      "loss": 0.2309,
+      "step": 7462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.887486863098845e-06,
+      "loss": 0.1578,
+      "step": 7463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.877094147173566e-06,
+      "loss": 0.15,
+      "step": 7464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.866706297367623e-06,
+      "loss": 0.2122,
+      "step": 7465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.856323314940835e-06,
+      "loss": 0.1194,
+      "step": 7466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.845945201152457e-06,
+      "loss": 0.1757,
+      "step": 7467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.83557195726117e-06,
+      "loss": 0.2273,
+      "step": 7468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.825203584525023e-06,
+      "loss": 0.153,
+      "step": 7469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.81484008420151e-06,
+      "loss": 0.2083,
+      "step": 7470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.804481457547498e-06,
+      "loss": 0.2047,
+      "step": 7471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.794127705819323e-06,
+      "loss": 0.2753,
+      "step": 7472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.783778830272645e-06,
+      "loss": 0.2462,
+      "step": 7473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.77343483216262e-06,
+      "loss": 0.2052,
+      "step": 7474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.763095712743737e-06,
+      "loss": 0.2188,
+      "step": 7475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.75276147326995e-06,
+      "loss": 0.1699,
+      "step": 7476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.742432114994603e-06,
+      "loss": 0.161,
+      "step": 7477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.732107639170419e-06,
+      "loss": 0.1786,
+      "step": 7478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.721788047049585e-06,
+      "loss": 0.2211,
+      "step": 7479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.711473339883637e-06,
+      "loss": 0.173,
+      "step": 7480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.701163518923556e-06,
+      "loss": 0.1719,
+      "step": 7481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.690858585419738e-06,
+      "loss": 0.1276,
+      "step": 7482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.680558540621937e-06,
+      "loss": 0.2091,
+      "step": 7483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.67026338577936e-06,
+      "loss": 0.1968,
+      "step": 7484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.659973122140625e-06,
+      "loss": 0.1629,
+      "step": 7485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.649687750953712e-06,
+      "loss": 0.1522,
+      "step": 7486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.639407273466044e-06,
+      "loss": 0.1383,
+      "step": 7487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.629131690924464e-06,
+      "loss": 0.2143,
+      "step": 7488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.618861004575152e-06,
+      "loss": 0.1724,
+      "step": 7489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.608595215663763e-06,
+      "loss": 0.1831,
+      "step": 7490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.598334325435348e-06,
+      "loss": 0.2108,
+      "step": 7491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.588078335134325e-06,
+      "loss": 0.1576,
+      "step": 7492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.577827246004562e-06,
+      "loss": 0.1847,
+      "step": 7493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.56758105928932e-06,
+      "loss": 0.2125,
+      "step": 7494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.55733977623124e-06,
+      "loss": 0.2496,
+      "step": 7495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.547103398072399e-06,
+      "loss": 0.1736,
+      "step": 7496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.536871926054275e-06,
+      "loss": 0.2699,
+      "step": 7497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.52664536141773e-06,
+      "loss": 0.1801,
+      "step": 7498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.516423705403066e-06,
+      "loss": 0.2579,
+      "step": 7499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.506206959249941e-06,
+      "loss": 0.187,
+      "step": 7500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.495995124197454e-06,
+      "loss": 0.1441,
+      "step": 7501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.485788201484126e-06,
+      "loss": 0.1711,
+      "step": 7502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.475586192347819e-06,
+      "loss": 0.2437,
+      "step": 7503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.465389098025857e-06,
+      "loss": 0.2121,
+      "step": 7504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.45519691975496e-06,
+      "loss": 0.2548,
+      "step": 7505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.445009658771203e-06,
+      "loss": 0.2429,
+      "step": 7506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.434827316310113e-06,
+      "loss": 0.1598,
+      "step": 7507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.424649893606629e-06,
+      "loss": 0.2857,
+      "step": 7508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.414477391895043e-06,
+      "loss": 0.2193,
+      "step": 7509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.40430981240909e-06,
+      "loss": 0.1041,
+      "step": 7510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.394147156381917e-06,
+      "loss": 0.1539,
+      "step": 7511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.383989425046026e-06,
+      "loss": 0.1416,
+      "step": 7512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.373836619633358e-06,
+      "loss": 0.2062,
+      "step": 7513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.36368874137527e-06,
+      "loss": 0.1794,
+      "step": 7514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.353545791502466e-06,
+      "loss": 0.1759,
+      "step": 7515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.3434077712451e-06,
+      "loss": 0.177,
+      "step": 7516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.333274681832738e-06,
+      "loss": 0.2016,
+      "step": 7517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.323146524494281e-06,
+      "loss": 0.1473,
+      "step": 7518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.313023300458118e-06,
+      "loss": 0.2309,
+      "step": 7519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.302905010951962e-06,
+      "loss": 0.1407,
+      "step": 7520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.292791657202988e-06,
+      "loss": 0.1992,
+      "step": 7521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.282683240437722e-06,
+      "loss": 0.1708,
+      "step": 7522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.272579761882149e-06,
+      "loss": 0.151,
+      "step": 7523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.262481222761587e-06,
+      "loss": 0.1853,
+      "step": 7524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.252387624300823e-06,
+      "loss": 0.1327,
+      "step": 7525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.242298967723977e-06,
+      "loss": 0.2041,
+      "step": 7526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.232215254254627e-06,
+      "loss": 0.2648,
+      "step": 7527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.22213648511574e-06,
+      "loss": 0.1436,
+      "step": 7528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.21206266152964e-06,
+      "loss": 0.1641,
+      "step": 7529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.201993784718104e-06,
+      "loss": 0.2286,
+      "step": 7530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.191929855902304e-06,
+      "loss": 0.2598,
+      "step": 7531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.181870876302761e-06,
+      "loss": 0.1466,
+      "step": 7532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.171816847139448e-06,
+      "loss": 0.2172,
+      "step": 7533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.161767769631741e-06,
+      "loss": 0.2109,
+      "step": 7534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.151723644998356e-06,
+      "loss": 0.1891,
+      "step": 7535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.141684474457474e-06,
+      "loss": 0.2192,
+      "step": 7536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.131650259226665e-06,
+      "loss": 0.2439,
+      "step": 7537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.12162100052284e-06,
+      "loss": 0.2234,
+      "step": 7538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.11159669956237e-06,
+      "loss": 0.1502,
+      "step": 7539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.101577357561015e-06,
+      "loss": 0.2462,
+      "step": 7540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.091562975733909e-06,
+      "loss": 0.1781,
+      "step": 7541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.081553555295602e-06,
+      "loss": 0.2553,
+      "step": 7542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.071549097460058e-06,
+      "loss": 0.1763,
+      "step": 7543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.061549603440595e-06,
+      "loss": 0.1368,
+      "step": 7544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.05155507444998e-06,
+      "loss": 0.2146,
+      "step": 7545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.041565511700328e-06,
+      "loss": 0.1583,
+      "step": 7546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.031580916403188e-06,
+      "loss": 0.1719,
+      "step": 7547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.021601289769505e-06,
+      "loss": 0.1825,
+      "step": 7548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.011626633009595e-06,
+      "loss": 0.2157,
+      "step": 7549
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.001656947333187e-06,
+      "loss": 0.237,
+      "step": 7550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.991692233949433e-06,
+      "loss": 0.1903,
+      "step": 7551
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.981732494066825e-06,
+      "loss": 0.1619,
+      "step": 7552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.9717777288933e-06,
+      "loss": 0.2091,
+      "step": 7553
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.961827939636196e-06,
+      "loss": 0.1837,
+      "step": 7554
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.95188312750218e-06,
+      "loss": 0.3214,
+      "step": 7555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.941943293697386e-06,
+      "loss": 0.1479,
+      "step": 7556
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.93200843942733e-06,
+      "loss": 0.2154,
+      "step": 7557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.922078565896892e-06,
+      "loss": 0.2092,
+      "step": 7558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.912153674310385e-06,
+      "loss": 0.1657,
+      "step": 7559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.90223376587151e-06,
+      "loss": 0.1758,
+      "step": 7560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.892318841783337e-06,
+      "loss": 0.2042,
+      "step": 7561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.882408903248362e-06,
+      "loss": 0.211,
+      "step": 7562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.872503951468481e-06,
+      "loss": 0.1903,
+      "step": 7563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.86260398764494e-06,
+      "loss": 0.1771,
+      "step": 7564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.852709012978439e-06,
+      "loss": 0.2203,
+      "step": 7565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.842819028669014e-06,
+      "loss": 0.2147,
+      "step": 7566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.832934035916141e-06,
+      "loss": 0.2336,
+      "step": 7567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.823054035918687e-06,
+      "loss": 0.1654,
+      "step": 7568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.813179029874874e-06,
+      "loss": 0.1783,
+      "step": 7569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.803309018982382e-06,
+      "loss": 0.3104,
+      "step": 7570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.793444004438212e-06,
+      "loss": 0.1817,
+      "step": 7571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.783583987438826e-06,
+      "loss": 0.1504,
+      "step": 7572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.773728969180023e-06,
+      "loss": 0.2043,
+      "step": 7573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.763878950857051e-06,
+      "loss": 0.182,
+      "step": 7574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.754033933664495e-06,
+      "loss": 0.2092,
+      "step": 7575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.744193918796378e-06,
+      "loss": 0.195,
+      "step": 7576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.734358907446116e-06,
+      "loss": 0.124,
+      "step": 7577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.724528900806467e-06,
+      "loss": 0.1991,
+      "step": 7578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.714703900069637e-06,
+      "loss": 0.2168,
+      "step": 7579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.704883906427213e-06,
+      "loss": 0.2353,
+      "step": 7580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.695068921070154e-06,
+      "loss": 0.2539,
+      "step": 7581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.685258945188824e-06,
+      "loss": 0.189,
+      "step": 7582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.675453979972998e-06,
+      "loss": 0.2252,
+      "step": 7583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.665654026611796e-06,
+      "loss": 0.1608,
+      "step": 7584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.655859086293795e-06,
+      "loss": 0.2096,
+      "step": 7585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.646069160206894e-06,
+      "loss": 0.2422,
+      "step": 7586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.636284249538446e-06,
+      "loss": 0.2028,
+      "step": 7587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.626504355475146e-06,
+      "loss": 0.1885,
+      "step": 7588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.616729479203123e-06,
+      "loss": 0.1609,
+      "step": 7589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.606959621907851e-06,
+      "loss": 0.1374,
+      "step": 7590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.597194784774243e-06,
+      "loss": 0.1927,
+      "step": 7591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.587434968986591e-06,
+      "loss": 0.2286,
+      "step": 7592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.577680175728536e-06,
+      "loss": 0.2736,
+      "step": 7593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.567930406183178e-06,
+      "loss": 0.1679,
+      "step": 7594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.558185661532941e-06,
+      "loss": 0.205,
+      "step": 7595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.548445942959682e-06,
+      "loss": 0.1454,
+      "step": 7596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.53871125164466e-06,
+      "loss": 0.2104,
+      "step": 7597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.528981588768464e-06,
+      "loss": 0.1937,
+      "step": 7598
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.519256955511128e-06,
+      "loss": 0.2021,
+      "step": 7599
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.509537353052072e-06,
+      "loss": 0.2616,
+      "step": 7600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.499822782570065e-06,
+      "loss": 0.1579,
+      "step": 7601
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.490113245243314e-06,
+      "loss": 0.1912,
+      "step": 7602
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.480408742249413e-06,
+      "loss": 0.314,
+      "step": 7603
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.470709274765277e-06,
+      "loss": 0.218,
+      "step": 7604
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.461014843967285e-06,
+      "loss": 0.1982,
+      "step": 7605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.4513254510312e-06,
+      "loss": 0.161,
+      "step": 7606
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.441641097132119e-06,
+      "loss": 0.1138,
+      "step": 7607
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.431961783444586e-06,
+      "loss": 0.1552,
+      "step": 7608
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.422287511142523e-06,
+      "loss": 0.2074,
+      "step": 7609
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.412618281399198e-06,
+      "loss": 0.1851,
+      "step": 7610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.40295409538731e-06,
+      "loss": 0.2152,
+      "step": 7611
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.393294954278957e-06,
+      "loss": 0.2233,
+      "step": 7612
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.383640859245573e-06,
+      "loss": 0.2802,
+      "step": 7613
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.373991811458026e-06,
+      "loss": 0.1636,
+      "step": 7614
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.364347812086549e-06,
+      "loss": 0.2778,
+      "step": 7615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.354708862300764e-06,
+      "loss": 0.2157,
+      "step": 7616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.345074963269706e-06,
+      "loss": 0.2411,
+      "step": 7617
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.335446116161755e-06,
+      "loss": 0.1788,
+      "step": 7618
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.325822322144727e-06,
+      "loss": 0.1868,
+      "step": 7619
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.316203582385778e-06,
+      "loss": 0.212,
+      "step": 7620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.306589898051464e-06,
+      "loss": 0.2126,
+      "step": 7621
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.296981270307746e-06,
+      "loss": 0.2179,
+      "step": 7622
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.287377700319982e-06,
+      "loss": 0.1517,
+      "step": 7623
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.277779189252866e-06,
+      "loss": 0.2089,
+      "step": 7624
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.268185738270517e-06,
+      "loss": 0.1744,
+      "step": 7625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.25859734853645e-06,
+      "loss": 0.1883,
+      "step": 7626
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.249014021213518e-06,
+      "loss": 0.1774,
+      "step": 7627
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.239435757464009e-06,
+      "loss": 0.2324,
+      "step": 7628
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.229862558449591e-06,
+      "loss": 0.1851,
+      "step": 7629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.22029442533127e-06,
+      "loss": 0.2629,
+      "step": 7630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.210731359269496e-06,
+      "loss": 0.2568,
+      "step": 7631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.201173361424086e-06,
+      "loss": 0.1493,
+      "step": 7632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.19162043295421e-06,
+      "loss": 0.2184,
+      "step": 7633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.182072575018485e-06,
+      "loss": 0.1459,
+      "step": 7634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.172529788774847e-06,
+      "loss": 0.2351,
+      "step": 7635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.162992075380677e-06,
+      "loss": 0.2071,
+      "step": 7636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.153459435992683e-06,
+      "loss": 0.2131,
+      "step": 7637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.143931871767014e-06,
+      "loss": 0.2071,
+      "step": 7638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.13440938385915e-06,
+      "loss": 0.13,
+      "step": 7639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.124891973424003e-06,
+      "loss": 0.1734,
+      "step": 7640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.115379641615827e-06,
+      "loss": 0.2372,
+      "step": 7641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.105872389588298e-06,
+      "loss": 0.269,
+      "step": 7642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.096370218494458e-06,
+      "loss": 0.1961,
+      "step": 7643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.086873129486721e-06,
+      "loss": 0.2311,
+      "step": 7644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.07738112371691e-06,
+      "loss": 0.1804,
+      "step": 7645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.067894202336218e-06,
+      "loss": 0.167,
+      "step": 7646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.05841236649521e-06,
+      "loss": 0.1721,
+      "step": 7647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.04893561734385e-06,
+      "loss": 0.2102,
+      "step": 7648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.0394639560315e-06,
+      "loss": 0.1499,
+      "step": 7649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.029997383706861e-06,
+      "loss": 0.1432,
+      "step": 7650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.02053590151805e-06,
+      "loss": 0.1787,
+      "step": 7651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.011079510612585e-06,
+      "loss": 0.1944,
+      "step": 7652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.00162821213729e-06,
+      "loss": 0.2732,
+      "step": 7653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.992182007238452e-06,
+      "loss": 0.2099,
+      "step": 7654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.982740897061713e-06,
+      "loss": 0.2253,
+      "step": 7655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.97330488275208e-06,
+      "loss": 0.2101,
+      "step": 7656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.96387396545396e-06,
+      "loss": 0.1632,
+      "step": 7657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.954448146311155e-06,
+      "loss": 0.2001,
+      "step": 7658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.9450274264668e-06,
+      "loss": 0.1927,
+      "step": 7659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.935611807063476e-06,
+      "loss": 0.1291,
+      "step": 7660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.92620128924308e-06,
+      "loss": 0.1842,
+      "step": 7661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.916795874146943e-06,
+      "loss": 0.1944,
+      "step": 7662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.907395562915765e-06,
+      "loss": 0.1849,
+      "step": 7663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.898000356689594e-06,
+      "loss": 0.2274,
+      "step": 7664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.888610256607893e-06,
+      "loss": 0.1762,
+      "step": 7665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.879225263809515e-06,
+      "loss": 0.1502,
+      "step": 7666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.86984537943265e-06,
+      "loss": 0.1179,
+      "step": 7667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.8604706046149e-06,
+      "loss": 0.1799,
+      "step": 7668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.851100940493272e-06,
+      "loss": 0.1946,
+      "step": 7669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.841736388204068e-06,
+      "loss": 0.1903,
+      "step": 7670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.832376948883048e-06,
+      "loss": 0.1851,
+      "step": 7671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.823022623665344e-06,
+      "loss": 0.1531,
+      "step": 7672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.813673413685429e-06,
+      "loss": 0.2184,
+      "step": 7673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.804329320077181e-06,
+      "loss": 0.1905,
+      "step": 7674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.79499034397388e-06,
+      "loss": 0.2357,
+      "step": 7675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.785656486508126e-06,
+      "loss": 0.1891,
+      "step": 7676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.776327748811946e-06,
+      "loss": 0.1347,
+      "step": 7677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.767004132016742e-06,
+      "loss": 0.1554,
+      "step": 7678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.75768563725327e-06,
+      "loss": 0.2012,
+      "step": 7679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.7483722656517e-06,
+      "loss": 0.1657,
+      "step": 7680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.739064018341535e-06,
+      "loss": 0.2094,
+      "step": 7681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.729760896451693e-06,
+      "loss": 0.1939,
+      "step": 7682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.720462901110476e-06,
+      "loss": 0.14,
+      "step": 7683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.711170033445519e-06,
+      "loss": 0.1984,
+      "step": 7684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.701882294583896e-06,
+      "loss": 0.1743,
+      "step": 7685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.692599685651997e-06,
+      "loss": 0.2612,
+      "step": 7686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.683322207775645e-06,
+      "loss": 0.1328,
+      "step": 7687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.67404986207999e-06,
+      "loss": 0.3051,
+      "step": 7688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.664782649689611e-06,
+      "loss": 0.1519,
+      "step": 7689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.655520571728414e-06,
+      "loss": 0.2175,
+      "step": 7690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.646263629319717e-06,
+      "loss": 0.2069,
+      "step": 7691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.637011823586221e-06,
+      "loss": 0.1988,
+      "step": 7692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.627765155649963e-06,
+      "loss": 0.235,
+      "step": 7693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.6185236266323956e-06,
+      "loss": 0.1671,
+      "step": 7694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.609287237654344e-06,
+      "loss": 0.176,
+      "step": 7695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.600055989835975e-06,
+      "loss": 0.1728,
+      "step": 7696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.590829884296869e-06,
+      "loss": 0.158,
+      "step": 7697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.581608922155986e-06,
+      "loss": 0.2454,
+      "step": 7698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.572393104531622e-06,
+      "loss": 0.1887,
+      "step": 7699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5631824325415044e-06,
+      "loss": 0.2251,
+      "step": 7700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.553976907302685e-06,
+      "loss": 0.1896,
+      "step": 7701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.544776529931602e-06,
+      "loss": 0.158,
+      "step": 7702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.535581301544104e-06,
+      "loss": 0.1779,
+      "step": 7703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.526391223255386e-06,
+      "loss": 0.1805,
+      "step": 7704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.517206296180013e-06,
+      "loss": 0.1576,
+      "step": 7705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5080265214319435e-06,
+      "loss": 0.2399,
+      "step": 7706
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4988519001245095e-06,
+      "loss": 0.159,
+      "step": 7707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.489682433370404e-06,
+      "loss": 0.1634,
+      "step": 7708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.48051812228171e-06,
+      "loss": 0.2086,
+      "step": 7709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.47135896796986e-06,
+      "loss": 0.2164,
+      "step": 7710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4622049715457e-06,
+      "loss": 0.1829,
+      "step": 7711
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.453056134119429e-06,
+      "loss": 0.1113,
+      "step": 7712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.443912456800606e-06,
+      "loss": 0.2255,
+      "step": 7713
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.434773940698181e-06,
+      "loss": 0.1817,
+      "step": 7714
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.425640586920501e-06,
+      "loss": 0.2513,
+      "step": 7715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.416512396575226e-06,
+      "loss": 0.177,
+      "step": 7716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.407389370769452e-06,
+      "loss": 0.1792,
+      "step": 7717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3982715106096335e-06,
+      "loss": 0.1478,
+      "step": 7718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.389158817201542e-06,
+      "loss": 0.1747,
+      "step": 7719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.380051291650397e-06,
+      "loss": 0.2198,
+      "step": 7720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.370948935060773e-06,
+      "loss": 0.2017,
+      "step": 7721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.361851748536586e-06,
+      "loss": 0.219,
+      "step": 7722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.352759733181147e-06,
+      "loss": 0.1639,
+      "step": 7723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.343672890097158e-06,
+      "loss": 0.1607,
+      "step": 7724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.334591220386649e-06,
+      "loss": 0.204,
+      "step": 7725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.325514725151061e-06,
+      "loss": 0.1898,
+      "step": 7726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.316443405491208e-06,
+      "loss": 0.1861,
+      "step": 7727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.307377262507237e-06,
+      "loss": 0.2039,
+      "step": 7728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.298316297298713e-06,
+      "loss": 0.1808,
+      "step": 7729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.28926051096453e-06,
+      "loss": 0.1866,
+      "step": 7730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.280209904602997e-06,
+      "loss": 0.1948,
+      "step": 7731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.271164479311776e-06,
+      "loss": 0.2088,
+      "step": 7732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.262124236187878e-06,
+      "loss": 0.1867,
+      "step": 7733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.253089176327737e-06,
+      "loss": 0.2209,
+      "step": 7734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.244059300827105e-06,
+      "loss": 0.2205,
+      "step": 7735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.235034610781127e-06,
+      "loss": 0.2044,
+      "step": 7736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.226015107284328e-06,
+      "loss": 0.2096,
+      "step": 7737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.217000791430606e-06,
+      "loss": 0.2352,
+      "step": 7738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.207991664313202e-06,
+      "loss": 0.1609,
+      "step": 7739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.198987727024753e-06,
+      "loss": 0.1512,
+      "step": 7740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.189988980657275e-06,
+      "loss": 0.2505,
+      "step": 7741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.180995426302106e-06,
+      "loss": 0.2054,
+      "step": 7742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.172007065050013e-06,
+      "loss": 0.1787,
+      "step": 7743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.163023897991111e-06,
+      "loss": 0.1587,
+      "step": 7744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.154045926214859e-06,
+      "loss": 0.1746,
+      "step": 7745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1450731508101165e-06,
+      "loss": 0.205,
+      "step": 7746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.136105572865126e-06,
+      "loss": 0.1789,
+      "step": 7747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.127143193467445e-06,
+      "loss": 0.2611,
+      "step": 7748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.118186013704065e-06,
+      "loss": 0.1982,
+      "step": 7749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.109234034661289e-06,
+      "loss": 0.1724,
+      "step": 7750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.100287257424837e-06,
+      "loss": 0.1126,
+      "step": 7751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.091345683079753e-06,
+      "loss": 0.1603,
+      "step": 7752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.082409312710503e-06,
+      "loss": 0.2581,
+      "step": 7753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.073478147400869e-06,
+      "loss": 0.2025,
+      "step": 7754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.064552188234047e-06,
+      "loss": 0.2381,
+      "step": 7755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.055631436292553e-06,
+      "loss": 0.2168,
+      "step": 7756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.046715892658318e-06,
+      "loss": 0.1709,
+      "step": 7757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.03780555841263e-06,
+      "loss": 0.1696,
+      "step": 7758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.028900434636115e-06,
+      "loss": 0.2634,
+      "step": 7759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.020000522408798e-06,
+      "loss": 0.1427,
+      "step": 7760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.011105822810077e-06,
+      "loss": 0.2051,
+      "step": 7761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.002216336918682e-06,
+      "loss": 0.1477,
+      "step": 7762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.993332065812747e-06,
+      "loss": 0.2373,
+      "step": 7763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.984453010569758e-06,
+      "loss": 0.2211,
+      "step": 7764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.975579172266566e-06,
+      "loss": 0.2213,
+      "step": 7765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.966710551979388e-06,
+      "loss": 0.2097,
+      "step": 7766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9578471507838475e-06,
+      "loss": 0.2712,
+      "step": 7767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9489889697548436e-06,
+      "loss": 0.2567,
+      "step": 7768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.940136009966736e-06,
+      "loss": 0.2514,
+      "step": 7769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9312882724932136e-06,
+      "loss": 0.1649,
+      "step": 7770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.922445758407315e-06,
+      "loss": 0.2091,
+      "step": 7771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.913608468781479e-06,
+      "loss": 0.1896,
+      "step": 7772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.904776404687497e-06,
+      "loss": 0.218,
+      "step": 7773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.895949567196514e-06,
+      "loss": 0.2167,
+      "step": 7774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8871279573790635e-06,
+      "loss": 0.2712,
+      "step": 7775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8783115763050135e-06,
+      "loss": 0.1593,
+      "step": 7776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.869500425043634e-06,
+      "loss": 0.2349,
+      "step": 7777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.860694504663551e-06,
+      "loss": 0.2221,
+      "step": 7778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.85189381623273e-06,
+      "loss": 0.1513,
+      "step": 7779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.843098360818523e-06,
+      "loss": 0.212,
+      "step": 7780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8343081394876715e-06,
+      "loss": 0.2462,
+      "step": 7781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.825523153306223e-06,
+      "loss": 0.2293,
+      "step": 7782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.816743403339654e-06,
+      "loss": 0.2457,
+      "step": 7783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.807968890652755e-06,
+      "loss": 0.224,
+      "step": 7784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7991996163097036e-06,
+      "loss": 0.1283,
+      "step": 7785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.790435581374038e-06,
+      "loss": 0.1871,
+      "step": 7786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.781676786908681e-06,
+      "loss": 0.139,
+      "step": 7787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.772923233975881e-06,
+      "loss": 0.2388,
+      "step": 7788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.764174923637279e-06,
+      "loss": 0.1884,
+      "step": 7789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.755431856953881e-06,
+      "loss": 0.1947,
+      "step": 7790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.74669403498604e-06,
+      "loss": 0.1784,
+      "step": 7791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7379614587934796e-06,
+      "loss": 0.183,
+      "step": 7792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7292341294353075e-06,
+      "loss": 0.2083,
+      "step": 7793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7205120479699565e-06,
+      "loss": 0.2167,
+      "step": 7794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.711795215455257e-06,
+      "loss": 0.1643,
+      "step": 7795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7030836329483755e-06,
+      "loss": 0.1459,
+      "step": 7796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.694377301505866e-06,
+      "loss": 0.1386,
+      "step": 7797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.68567622218364e-06,
+      "loss": 0.1902,
+      "step": 7798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.676980396036953e-06,
+      "loss": 0.2,
+      "step": 7799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.668289824120455e-06,
+      "loss": 0.1741,
+      "step": 7800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.659604507488121e-06,
+      "loss": 0.1919,
+      "step": 7801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.65092444719333e-06,
+      "loss": 0.2191,
+      "step": 7802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.642249644288778e-06,
+      "loss": 0.1478,
+      "step": 7803
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.633580099826581e-06,
+      "loss": 0.1582,
+      "step": 7804
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.624915814858146e-06,
+      "loss": 0.206,
+      "step": 7805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.616256790434305e-06,
+      "loss": 0.2151,
+      "step": 7806
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.607603027605231e-06,
+      "loss": 0.1473,
+      "step": 7807
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.598954527420436e-06,
+      "loss": 0.2201,
+      "step": 7808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5903112909288245e-06,
+      "loss": 0.1828,
+      "step": 7809
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.581673319178661e-06,
+      "loss": 0.1443,
+      "step": 7810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.573040613217534e-06,
+      "loss": 0.1917,
+      "step": 7811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.564413174092443e-06,
+      "loss": 0.2734,
+      "step": 7812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.555791002849732e-06,
+      "loss": 0.1827,
+      "step": 7813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.547174100535075e-06,
+      "loss": 0.2537,
+      "step": 7814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.53856246819356e-06,
+      "loss": 0.2113,
+      "step": 7815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5299561068696e-06,
+      "loss": 0.1519,
+      "step": 7816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5213550176069615e-06,
+      "loss": 0.1761,
+      "step": 7817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.512759201448798e-06,
+      "loss": 0.1442,
+      "step": 7818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.504168659437626e-06,
+      "loss": 0.1908,
+      "step": 7819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.495583392615284e-06,
+      "loss": 0.1662,
+      "step": 7820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.487003402023012e-06,
+      "loss": 0.2016,
+      "step": 7821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.478428688701405e-06,
+      "loss": 0.1758,
+      "step": 7822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.46985925369038e-06,
+      "loss": 0.209,
+      "step": 7823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4612950980292685e-06,
+      "loss": 0.1674,
+      "step": 7824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.452736222756706e-06,
+      "loss": 0.1441,
+      "step": 7825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.444182628910733e-06,
+      "loss": 0.1933,
+      "step": 7826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.435634317528738e-06,
+      "loss": 0.1694,
+      "step": 7827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4270912896474465e-06,
+      "loss": 0.1626,
+      "step": 7828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.418553546302963e-06,
+      "loss": 0.2029,
+      "step": 7829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.410021088530771e-06,
+      "loss": 0.1767,
+      "step": 7830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4014939173656526e-06,
+      "loss": 0.1694,
+      "step": 7831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.392972033841809e-06,
+      "loss": 0.2248,
+      "step": 7832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.384455438992798e-06,
+      "loss": 0.1734,
+      "step": 7833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.375944133851464e-06,
+      "loss": 0.2088,
+      "step": 7834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.367438119450086e-06,
+      "loss": 0.2053,
+      "step": 7835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.358937396820291e-06,
+      "loss": 0.2176,
+      "step": 7836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.35044196699302e-06,
+      "loss": 0.17,
+      "step": 7837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.341951830998622e-06,
+      "loss": 0.1741,
+      "step": 7838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3334669898667875e-06,
+      "loss": 0.1847,
+      "step": 7839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.32498744462654e-06,
+      "loss": 0.2349,
+      "step": 7840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.316513196306295e-06,
+      "loss": 0.1705,
+      "step": 7841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.30804424593382e-06,
+      "loss": 0.1661,
+      "step": 7842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.299580594536214e-06,
+      "loss": 0.2094,
+      "step": 7843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.291122243139969e-06,
+      "loss": 0.2534,
+      "step": 7844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.282669192770896e-06,
+      "loss": 0.1833,
+      "step": 7845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.274221444454198e-06,
+      "loss": 0.2226,
+      "step": 7846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.265778999214428e-06,
+      "loss": 0.1723,
+      "step": 7847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.257341858075466e-06,
+      "loss": 0.1946,
+      "step": 7848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.248910022060595e-06,
+      "loss": 0.171,
+      "step": 7849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.240483492192423e-06,
+      "loss": 0.203,
+      "step": 7850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.232062269492906e-06,
+      "loss": 0.2014,
+      "step": 7851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.223646354983387e-06,
+      "loss": 0.1757,
+      "step": 7852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.215235749684561e-06,
+      "loss": 0.178,
+      "step": 7853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.206830454616447e-06,
+      "loss": 0.2376,
+      "step": 7854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.198430470798455e-06,
+      "loss": 0.1539,
+      "step": 7855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.1900357992493445e-06,
+      "loss": 0.238,
+      "step": 7856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.181646440987204e-06,
+      "loss": 0.1893,
+      "step": 7857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.173262397029511e-06,
+      "loss": 0.2301,
+      "step": 7858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.16488366839309e-06,
+      "loss": 0.237,
+      "step": 7859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1565102560941005e-06,
+      "loss": 0.2468,
+      "step": 7860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1481421611480796e-06,
+      "loss": 0.1535,
+      "step": 7861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.139779384569927e-06,
+      "loss": 0.1469,
+      "step": 7862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.131421927373859e-06,
+      "loss": 0.1859,
+      "step": 7863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.123069790573494e-06,
+      "loss": 0.2452,
+      "step": 7864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.114722975181763e-06,
+      "loss": 0.2262,
+      "step": 7865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1063814822109765e-06,
+      "loss": 0.1411,
+      "step": 7866
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.09804531267279e-06,
+      "loss": 0.2109,
+      "step": 7867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.089714467578233e-06,
+      "loss": 0.2254,
+      "step": 7868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.08138894793765e-06,
+      "loss": 0.1629,
+      "step": 7869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.073068754760786e-06,
+      "loss": 0.168,
+      "step": 7870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.06475388905669e-06,
+      "loss": 0.1814,
+      "step": 7871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.056444351833812e-06,
+      "loss": 0.1962,
+      "step": 7872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.048140144099939e-06,
+      "loss": 0.1385,
+      "step": 7873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0398412668621895e-06,
+      "loss": 0.1587,
+      "step": 7874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.031547721127062e-06,
+      "loss": 0.1683,
+      "step": 7875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.023259507900419e-06,
+      "loss": 0.1505,
+      "step": 7876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.014976628187424e-06,
+      "loss": 0.2102,
+      "step": 7877
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.006699082992645e-06,
+      "loss": 0.1572,
+      "step": 7878
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.998426873320001e-06,
+      "loss": 0.1614,
+      "step": 7879
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.990160000172717e-06,
+      "loss": 0.2139,
+      "step": 7880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.981898464553415e-06,
+      "loss": 0.1393,
+      "step": 7881
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.973642267464086e-06,
+      "loss": 0.1573,
+      "step": 7882
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.965391409905991e-06,
+      "loss": 0.2578,
+      "step": 7883
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9571458928798285e-06,
+      "loss": 0.1556,
+      "step": 7884
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.948905717385622e-06,
+      "loss": 0.1912,
+      "step": 7885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9406708844227164e-06,
+      "loss": 0.1936,
+      "step": 7886
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.932441394989857e-06,
+      "loss": 0.1651,
+      "step": 7887
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.924217250085123e-06,
+      "loss": 0.2223,
+      "step": 7888
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.915998450705923e-06,
+      "loss": 0.2291,
+      "step": 7889
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.907784997849053e-06,
+      "loss": 0.147,
+      "step": 7890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.899576892510628e-06,
+      "loss": 0.1278,
+      "step": 7891
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.891374135686134e-06,
+      "loss": 0.1634,
+      "step": 7892
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.883176728370415e-06,
+      "loss": 0.2214,
+      "step": 7893
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.874984671557637e-06,
+      "loss": 0.1546,
+      "step": 7894
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.86679796624135e-06,
+      "loss": 0.2059,
+      "step": 7895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8586166134144415e-06,
+      "loss": 0.1724,
+      "step": 7896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.850440614069136e-06,
+      "loss": 0.2106,
+      "step": 7897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.842269969197039e-06,
+      "loss": 0.1646,
+      "step": 7898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.834104679789077e-06,
+      "loss": 0.2025,
+      "step": 7899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.825944746835527e-06,
+      "loss": 0.2205,
+      "step": 7900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.817790171326043e-06,
+      "loss": 0.2195,
+      "step": 7901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.809640954249629e-06,
+      "loss": 0.1945,
+      "step": 7902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.801497096594593e-06,
+      "loss": 0.1989,
+      "step": 7903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.793358599348642e-06,
+      "loss": 0.256,
+      "step": 7904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.785225463498828e-06,
+      "loss": 0.2518,
+      "step": 7905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.777097690031513e-06,
+      "loss": 0.2839,
+      "step": 7906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7689752799324486e-06,
+      "loss": 0.1612,
+      "step": 7907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.760858234186739e-06,
+      "loss": 0.2335,
+      "step": 7908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.752746553778798e-06,
+      "loss": 0.2117,
+      "step": 7909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.744640239692428e-06,
+      "loss": 0.2115,
+      "step": 7910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7365392929107585e-06,
+      "loss": 0.1888,
+      "step": 7911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.728443714416271e-06,
+      "loss": 0.1747,
+      "step": 7912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.720353505190818e-06,
+      "loss": 0.1942,
+      "step": 7913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.712268666215559e-06,
+      "loss": 0.2413,
+      "step": 7914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.704189198471044e-06,
+      "loss": 0.1791,
+      "step": 7915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.696115102937144e-06,
+      "loss": 0.1943,
+      "step": 7916
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.688046380593098e-06,
+      "loss": 0.1849,
+      "step": 7917
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.679983032417463e-06,
+      "loss": 0.1946,
+      "step": 7918
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.671925059388195e-06,
+      "loss": 0.2222,
+      "step": 7919
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.663872462482534e-06,
+      "loss": 0.1225,
+      "step": 7920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.655825242677121e-06,
+      "loss": 0.1812,
+      "step": 7921
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.647783400947931e-06,
+      "loss": 0.2334,
+      "step": 7922
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.639746938270263e-06,
+      "loss": 0.2344,
+      "step": 7923
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.631715855618797e-06,
+      "loss": 0.1794,
+      "step": 7924
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.623690153967542e-06,
+      "loss": 0.1793,
+      "step": 7925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6156698342898495e-06,
+      "loss": 0.1242,
+      "step": 7926
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.607654897558434e-06,
+      "loss": 0.1912,
+      "step": 7927
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5996453447453575e-06,
+      "loss": 0.2073,
+      "step": 7928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.591641176822004e-06,
+      "loss": 0.1674,
+      "step": 7929
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.583642394759137e-06,
+      "loss": 0.1584,
+      "step": 7930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.575648999526844e-06,
+      "loss": 0.2669,
+      "step": 7931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5676609920945585e-06,
+      "loss": 0.2454,
+      "step": 7932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.559678373431082e-06,
+      "loss": 0.1646,
+      "step": 7933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.551701144504545e-06,
+      "loss": 0.1868,
+      "step": 7934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.543729306282425e-06,
+      "loss": 0.2255,
+      "step": 7935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.535762859731547e-06,
+      "loss": 0.2236,
+      "step": 7936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.527801805818105e-06,
+      "loss": 0.1646,
+      "step": 7937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.51984614550759e-06,
+      "loss": 0.1931,
+      "step": 7938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.511895879764883e-06,
+      "loss": 0.1857,
+      "step": 7939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.503951009554192e-06,
+      "loss": 0.2249,
+      "step": 7940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.496011535839063e-06,
+      "loss": 0.1796,
+      "step": 7941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.488077459582425e-06,
+      "loss": 0.1584,
+      "step": 7942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.480148781746497e-06,
+      "loss": 0.2045,
+      "step": 7943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4722255032928826e-06,
+      "loss": 0.1837,
+      "step": 7944
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.46430762518253e-06,
+      "loss": 0.1953,
+      "step": 7945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.456395148375698e-06,
+      "loss": 0.1699,
+      "step": 7946
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.448488073832048e-06,
+      "loss": 0.2066,
+      "step": 7947
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4405864025105225e-06,
+      "loss": 0.1705,
+      "step": 7948
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.432690135369444e-06,
+      "loss": 0.2094,
+      "step": 7949
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.424799273366482e-06,
+      "loss": 0.1582,
+      "step": 7950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.416913817458652e-06,
+      "loss": 0.1595,
+      "step": 7951
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.4090337686022784e-06,
+      "loss": 0.1735,
+      "step": 7952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.401159127753075e-06,
+      "loss": 0.2457,
+      "step": 7953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.393289895866083e-06,
+      "loss": 0.201,
+      "step": 7954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.385426073895672e-06,
+      "loss": 0.2534,
+      "step": 7955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.37756766279558e-06,
+      "loss": 0.2205,
+      "step": 7956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.369714663518876e-06,
+      "loss": 0.1393,
+      "step": 7957
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.361867077017968e-06,
+      "loss": 0.2925,
+      "step": 7958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.354024904244631e-06,
+      "loss": 0.1343,
+      "step": 7959
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.346188146149939e-06,
+      "loss": 0.278,
+      "step": 7960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.338356803684352e-06,
+      "loss": 0.202,
+      "step": 7961
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.330530877797668e-06,
+      "loss": 0.1918,
+      "step": 7962
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.322710369438999e-06,
+      "loss": 0.2136,
+      "step": 7963
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.314895279556831e-06,
+      "loss": 0.1804,
+      "step": 7964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.307085609098983e-06,
+      "loss": 0.2343,
+      "step": 7965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.299281359012592e-06,
+      "loss": 0.152,
+      "step": 7966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.291482530244179e-06,
+      "loss": 0.1666,
+      "step": 7967
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.283689123739599e-06,
+      "loss": 0.2389,
+      "step": 7968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.275901140444011e-06,
+      "loss": 0.2005,
+      "step": 7969
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.268118581301962e-06,
+      "loss": 0.1423,
+      "step": 7970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.260341447257327e-06,
+      "loss": 0.2429,
+      "step": 7971
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.252569739253299e-06,
+      "loss": 0.2539,
+      "step": 7972
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.244803458232455e-06,
+      "loss": 0.1588,
+      "step": 7973
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2370426051366885e-06,
+      "loss": 0.207,
+      "step": 7974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.229287180907227e-06,
+      "loss": 0.1419,
+      "step": 7975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.221537186484654e-06,
+      "loss": 0.2266,
+      "step": 7976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.213792622808905e-06,
+      "loss": 0.1428,
+      "step": 7977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.20605349081923e-06,
+      "loss": 0.1896,
+      "step": 7978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.198319791454237e-06,
+      "loss": 0.2167,
+      "step": 7979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.190591525651878e-06,
+      "loss": 0.164,
+      "step": 7980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.182868694349419e-06,
+      "loss": 0.2257,
+      "step": 7981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.175151298483505e-06,
+      "loss": 0.2785,
+      "step": 7982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1674393389901096e-06,
+      "loss": 0.1748,
+      "step": 7983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.159732816804524e-06,
+      "loss": 0.2008,
+      "step": 7984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1520317328614176e-06,
+      "loss": 0.1245,
+      "step": 7985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.144336088094759e-06,
+      "loss": 0.1858,
+      "step": 7986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.136645883437885e-06,
+      "loss": 0.1968,
+      "step": 7987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.128961119823483e-06,
+      "loss": 0.1262,
+      "step": 7988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.121281798183547e-06,
+      "loss": 0.1267,
+      "step": 7989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.113607919449426e-06,
+      "loss": 0.18,
+      "step": 7990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.105939484551825e-06,
+      "loss": 0.1632,
+      "step": 7991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0982764944207564e-06,
+      "loss": 0.1718,
+      "step": 7992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.090618949985604e-06,
+      "loss": 0.1673,
+      "step": 7993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.082966852175075e-06,
+      "loss": 0.0974,
+      "step": 7994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.075320201917211e-06,
+      "loss": 0.1027,
+      "step": 7995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.067679000139419e-06,
+      "loss": 0.158,
+      "step": 7996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.060043247768409e-06,
+      "loss": 0.2265,
+      "step": 7997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.05241294573024e-06,
+      "loss": 0.2087,
+      "step": 7998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.044788094950331e-06,
+      "loss": 0.1958,
+      "step": 7999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.037168696353434e-06,
+      "loss": 0.1445,
+      "step": 8000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.029554750863613e-06,
+      "loss": 0.2043,
+      "step": 8001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.021946259404298e-06,
+      "loss": 0.2014,
+      "step": 8002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.01434322289826e-06,
+      "loss": 0.2444,
+      "step": 8003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0067456422675794e-06,
+      "loss": 0.1642,
+      "step": 8004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9991535184337136e-06,
+      "loss": 0.2616,
+      "step": 8005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.99156685231742e-06,
+      "loss": 0.1806,
+      "step": 8006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9839856448388126e-06,
+      "loss": 0.212,
+      "step": 8007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.976409896917361e-06,
+      "loss": 0.1527,
+      "step": 8008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.968839609471837e-06,
+      "loss": 0.1782,
+      "step": 8009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.961274783420366e-06,
+      "loss": 0.2848,
+      "step": 8010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.953715419680432e-06,
+      "loss": 0.2146,
+      "step": 8011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.946161519168818e-06,
+      "loss": 0.2297,
+      "step": 8012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.938613082801674e-06,
+      "loss": 0.162,
+      "step": 8013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.931070111494479e-06,
+      "loss": 0.1657,
+      "step": 8014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.923532606162029e-06,
+      "loss": 0.1678,
+      "step": 8015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.916000567718487e-06,
+      "loss": 0.1512,
+      "step": 8016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.908473997077351e-06,
+      "loss": 0.179,
+      "step": 8017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.900952895151428e-06,
+      "loss": 0.2066,
+      "step": 8018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8934372628528846e-06,
+      "loss": 0.1913,
+      "step": 8019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.885927101093235e-06,
+      "loss": 0.205,
+      "step": 8020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.878422410783296e-06,
+      "loss": 0.2303,
+      "step": 8021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.870923192833238e-06,
+      "loss": 0.2347,
+      "step": 8022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.86342944815259e-06,
+      "loss": 0.2301,
+      "step": 8023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.85594117765017e-06,
+      "loss": 0.2243,
+      "step": 8024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.848458382234178e-06,
+      "loss": 0.1714,
+      "step": 8025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.840981062812111e-06,
+      "loss": 0.1106,
+      "step": 8026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8335092202908325e-06,
+      "loss": 0.2124,
+      "step": 8027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.82604285557654e-06,
+      "loss": 0.1675,
+      "step": 8028
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.818581969574742e-06,
+      "loss": 0.2272,
+      "step": 8029
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.811126563190288e-06,
+      "loss": 0.2727,
+      "step": 8030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.803676637327392e-06,
+      "loss": 0.2326,
+      "step": 8031
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.796232192889577e-06,
+      "loss": 0.1869,
+      "step": 8032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.788793230779703e-06,
+      "loss": 0.1674,
+      "step": 8033
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.781359751899983e-06,
+      "loss": 0.1278,
+      "step": 8034
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.773931757151928e-06,
+      "loss": 0.193,
+      "step": 8035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.766509247436429e-06,
+      "loss": 0.2555,
+      "step": 8036
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7590922236536885e-06,
+      "loss": 0.1689,
+      "step": 8037
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.751680686703231e-06,
+      "loss": 0.1391,
+      "step": 8038
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.744274637483936e-06,
+      "loss": 0.1849,
+      "step": 8039
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.736874076894032e-06,
+      "loss": 0.2317,
+      "step": 8040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.729479005831028e-06,
+      "loss": 0.2157,
+      "step": 8041
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.722089425191823e-06,
+      "loss": 0.2075,
+      "step": 8042
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.714705335872626e-06,
+      "loss": 0.1643,
+      "step": 8043
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.707326738768975e-06,
+      "loss": 0.1638,
+      "step": 8044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.699953634775755e-06,
+      "loss": 0.139,
+      "step": 8045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.692586024787182e-06,
+      "loss": 0.1841,
+      "step": 8046
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6852239096967785e-06,
+      "loss": 0.1799,
+      "step": 8047
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.677867290397448e-06,
+      "loss": 0.2892,
+      "step": 8048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.670516167781403e-06,
+      "loss": 0.1533,
+      "step": 8049
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.663170542740169e-06,
+      "loss": 0.1531,
+      "step": 8050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.655830416164647e-06,
+      "loss": 0.1759,
+      "step": 8051
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.648495788945051e-06,
+      "loss": 0.1503,
+      "step": 8052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.641166661970914e-06,
+      "loss": 0.1307,
+      "step": 8053
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.633843036131124e-06,
+      "loss": 0.2311,
+      "step": 8054
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.626524912313884e-06,
+      "loss": 0.1555,
+      "step": 8055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.619212291406744e-06,
+      "loss": 0.1694,
+      "step": 8056
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.611905174296588e-06,
+      "loss": 0.1864,
+      "step": 8057
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.604603561869614e-06,
+      "loss": 0.1728,
+      "step": 8058
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.597307455011363e-06,
+      "loss": 0.1934,
+      "step": 8059
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.590016854606727e-06,
+      "loss": 0.2512,
+      "step": 8060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.582731761539888e-06,
+      "loss": 0.2171,
+      "step": 8061
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.575452176694412e-06,
+      "loss": 0.1858,
+      "step": 8062
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568178100953146e-06,
+      "loss": 0.2328,
+      "step": 8063
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.560909535198299e-06,
+      "loss": 0.1761,
+      "step": 8064
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.553646480311402e-06,
+      "loss": 0.1585,
+      "step": 8065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5463889371733345e-06,
+      "loss": 0.2192,
+      "step": 8066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.539136906664276e-06,
+      "loss": 0.1772,
+      "step": 8067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5318903896637676e-06,
+      "loss": 0.1668,
+      "step": 8068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5246493870506665e-06,
+      "loss": 0.193,
+      "step": 8069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.51741389970316e-06,
+      "loss": 0.2117,
+      "step": 8070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.510183928498779e-06,
+      "loss": 0.2588,
+      "step": 8071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.502959474314366e-06,
+      "loss": 0.182,
+      "step": 8072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.495740538026105e-06,
+      "loss": 0.1925,
+      "step": 8073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.488527120509528e-06,
+      "loss": 0.1938,
+      "step": 8074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.481319222639463e-06,
+      "loss": 0.1434,
+      "step": 8075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4741168452900885e-06,
+      "loss": 0.2141,
+      "step": 8076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.466919989334923e-06,
+      "loss": 0.1237,
+      "step": 8077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4597286556467886e-06,
+      "loss": 0.2072,
+      "step": 8078
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.452542845097862e-06,
+      "loss": 0.252,
+      "step": 8079
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.445362558559646e-06,
+      "loss": 0.2628,
+      "step": 8080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4381877969029485e-06,
+      "loss": 0.2025,
+      "step": 8081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.43101856099794e-06,
+      "loss": 0.2727,
+      "step": 8082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.42385485171411e-06,
+      "loss": 0.1536,
+      "step": 8083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4166966699202675e-06,
+      "loss": 0.2282,
+      "step": 8084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4095440164845624e-06,
+      "loss": 0.1981,
+      "step": 8085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.402396892274485e-06,
+      "loss": 0.2079,
+      "step": 8086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.395255298156814e-06,
+      "loss": 0.1586,
+      "step": 8087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.388119234997701e-06,
+      "loss": 0.2167,
+      "step": 8088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.380988703662614e-06,
+      "loss": 0.1705,
+      "step": 8089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.373863705016335e-06,
+      "loss": 0.1633,
+      "step": 8090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.366744239922998e-06,
+      "loss": 0.1477,
+      "step": 8091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.359630309246038e-06,
+      "loss": 0.2377,
+      "step": 8092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3525219138482396e-06,
+      "loss": 0.1679,
+      "step": 8093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.345419054591732e-06,
+      "loss": 0.1606,
+      "step": 8094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.33832173233793e-06,
+      "loss": 0.1965,
+      "step": 8095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.331229947947596e-06,
+      "loss": 0.2162,
+      "step": 8096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3241437022808365e-06,
+      "loss": 0.1588,
+      "step": 8097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.31706299619708e-06,
+      "loss": 0.2135,
+      "step": 8098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.309987830555057e-06,
+      "loss": 0.187,
+      "step": 8099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.302918206212864e-06,
+      "loss": 0.2514,
+      "step": 8100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.295854124027887e-06,
+      "loss": 0.1503,
+      "step": 8101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.288795584856881e-06,
+      "loss": 0.1816,
+      "step": 8102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.281742589555904e-06,
+      "loss": 0.1388,
+      "step": 8103
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.274695138980328e-06,
+      "loss": 0.1773,
+      "step": 8104
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.267653233984892e-06,
+      "loss": 0.2091,
+      "step": 8105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.260616875423634e-06,
+      "loss": 0.1299,
+      "step": 8106
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.253586064149917e-06,
+      "loss": 0.2086,
+      "step": 8107
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.246560801016447e-06,
+      "loss": 0.1603,
+      "step": 8108
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.239541086875265e-06,
+      "loss": 0.1692,
+      "step": 8109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.232526922577696e-06,
+      "loss": 0.2211,
+      "step": 8110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.225518308974441e-06,
+      "loss": 0.165,
+      "step": 8111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2185152469154994e-06,
+      "loss": 0.1578,
+      "step": 8112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.211517737250198e-06,
+      "loss": 0.225,
+      "step": 8113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.204525780827201e-06,
+      "loss": 0.1711,
+      "step": 8114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1975393784945135e-06,
+      "loss": 0.1743,
+      "step": 8115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.19055853109942e-06,
+      "loss": 0.1803,
+      "step": 8116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.183583239488575e-06,
+      "loss": 0.1947,
+      "step": 8117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.176613504507954e-06,
+      "loss": 0.2154,
+      "step": 8118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.169649327002828e-06,
+      "loss": 0.208,
+      "step": 8119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.162690707817829e-06,
+      "loss": 0.2137,
+      "step": 8120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1557376477968926e-06,
+      "loss": 0.1575,
+      "step": 8121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.148790147783288e-06,
+      "loss": 0.1949,
+      "step": 8122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.14184820861962e-06,
+      "loss": 0.2023,
+      "step": 8123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134911831147797e-06,
+      "loss": 0.2,
+      "step": 8124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1279810162090755e-06,
+      "loss": 0.2163,
+      "step": 8125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.121055764644027e-06,
+      "loss": 0.1691,
+      "step": 8126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.114136077292541e-06,
+      "loss": 0.1464,
+      "step": 8127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.107221954993845e-06,
+      "loss": 0.1599,
+      "step": 8128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1003133985864865e-06,
+      "loss": 0.1405,
+      "step": 8129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.093410408908327e-06,
+      "loss": 0.1956,
+      "step": 8130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.086512986796564e-06,
+      "loss": 0.1909,
+      "step": 8131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.079621133087741e-06,
+      "loss": 0.1509,
+      "step": 8132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.072734848617682e-06,
+      "loss": 0.1619,
+      "step": 8133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.065854134221564e-06,
+      "loss": 0.2012,
+      "step": 8134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.058978990733891e-06,
+      "loss": 0.1593,
+      "step": 8135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.05210941898847e-06,
+      "loss": 0.15,
+      "step": 8136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.045245419818449e-06,
+      "loss": 0.1685,
+      "step": 8137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.038386994056303e-06,
+      "loss": 0.2099,
+      "step": 8138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.031534142533816e-06,
+      "loss": 0.2213,
+      "step": 8139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0246868660821116e-06,
+      "loss": 0.2006,
+      "step": 8140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.017845165531614e-06,
+      "loss": 0.2088,
+      "step": 8141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.011009041712099e-06,
+      "loss": 0.2066,
+      "step": 8142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.004178495452665e-06,
+      "loss": 0.2142,
+      "step": 8143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.997353527581704e-06,
+      "loss": 0.1801,
+      "step": 8144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.990534138926955e-06,
+      "loss": 0.21,
+      "step": 8145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.983720330315471e-06,
+      "loss": 0.2848,
+      "step": 8146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.976912102573655e-06,
+      "loss": 0.2394,
+      "step": 8147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.970109456527182e-06,
+      "loss": 0.2308,
+      "step": 8148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.963312393001106e-06,
+      "loss": 0.2197,
+      "step": 8149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.956520912819756e-06,
+      "loss": 0.1326,
+      "step": 8150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.94973501680681e-06,
+      "loss": 0.2001,
+      "step": 8151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.942954705785274e-06,
+      "loss": 0.2033,
+      "step": 8152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.936179980577453e-06,
+      "loss": 0.2211,
+      "step": 8153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9294108420049934e-06,
+      "loss": 0.1556,
+      "step": 8154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.92264729088887e-06,
+      "loss": 0.1919,
+      "step": 8155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.915889328049343e-06,
+      "loss": 0.1812,
+      "step": 8156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.909136954306042e-06,
+      "loss": 0.1781,
+      "step": 8157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.902390170477898e-06,
+      "loss": 0.1568,
+      "step": 8158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.895648977383143e-06,
+      "loss": 0.1544,
+      "step": 8159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.888913375839376e-06,
+      "loss": 0.1321,
+      "step": 8160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.882183366663477e-06,
+      "loss": 0.145,
+      "step": 8161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.875458950671662e-06,
+      "loss": 0.1794,
+      "step": 8162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.868740128679471e-06,
+      "loss": 0.2061,
+      "step": 8163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.86202690150178e-06,
+      "loss": 0.2693,
+      "step": 8164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.855319269952756e-06,
+      "loss": 0.1701,
+      "step": 8165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8486172348459135e-06,
+      "loss": 0.173,
+      "step": 8166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.841920796994064e-06,
+      "loss": 0.1618,
+      "step": 8167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.835229957209363e-06,
+      "loss": 0.2201,
+      "step": 8168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.828544716303284e-06,
+      "loss": 0.2331,
+      "step": 8169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8218650750866e-06,
+      "loss": 0.2,
+      "step": 8170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.815191034369426e-06,
+      "loss": 0.0907,
+      "step": 8171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8085225949612003e-06,
+      "loss": 0.2004,
+      "step": 8172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8018597576706616e-06,
+      "loss": 0.1197,
+      "step": 8173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7952025233058852e-06,
+      "loss": 0.1724,
+      "step": 8174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7885508926742698e-06,
+      "loss": 0.1506,
+      "step": 8175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.781904866582514e-06,
+      "loss": 0.157,
+      "step": 8176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.775264445836668e-06,
+      "loss": 0.2368,
+      "step": 8177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.768629631242071e-06,
+      "loss": 0.2157,
+      "step": 8178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.76200042360339e-06,
+      "loss": 0.1697,
+      "step": 8179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.755376823724627e-06,
+      "loss": 0.2369,
+      "step": 8180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7487588324090993e-06,
+      "loss": 0.1608,
+      "step": 8181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.742146450459427e-06,
+      "loss": 0.1929,
+      "step": 8182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.735539678677563e-06,
+      "loss": 0.1468,
+      "step": 8183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.728938517864794e-06,
+      "loss": 0.192,
+      "step": 8184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.722342968821696e-06,
+      "loss": 0.2253,
+      "step": 8185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7157530323481847e-06,
+      "loss": 0.2178,
+      "step": 8186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.709168709243488e-06,
+      "loss": 0.1363,
+      "step": 8187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.70259000030615e-06,
+      "loss": 0.2167,
+      "step": 8188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6960169063340545e-06,
+      "loss": 0.2522,
+      "step": 8189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6894494281243698e-06,
+      "loss": 0.2028,
+      "step": 8190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6828875664736082e-06,
+      "loss": 0.1624,
+      "step": 8191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6763313221776065e-06,
+      "loss": 0.2615,
+      "step": 8192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6697806960314997e-06,
+      "loss": 0.2177,
+      "step": 8193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6632356888297414e-06,
+      "loss": 0.1752,
+      "step": 8194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.656696301366125e-06,
+      "loss": 0.1999,
+      "step": 8195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6501625344337377e-06,
+      "loss": 0.1914,
+      "step": 8196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.643634388825001e-06,
+      "loss": 0.1597,
+      "step": 8197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6371118653316604e-06,
+      "loss": 0.1762,
+      "step": 8198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6305949647447545e-06,
+      "loss": 0.3127,
+      "step": 8199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6240836878546626e-06,
+      "loss": 0.1884,
+      "step": 8200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6175780354510856e-06,
+      "loss": 0.2067,
+      "step": 8201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.61107800832301e-06,
+      "loss": 0.154,
+      "step": 8202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6045836072587712e-06,
+      "loss": 0.2116,
+      "step": 8203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.598094833046023e-06,
+      "loss": 0.147,
+      "step": 8204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.591611686471702e-06,
+      "loss": 0.1878,
+      "step": 8205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5851341683221127e-06,
+      "loss": 0.1794,
+      "step": 8206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5786622793828273e-06,
+      "loss": 0.1619,
+      "step": 8207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5721960204387726e-06,
+      "loss": 0.2211,
+      "step": 8208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5657353922741834e-06,
+      "loss": 0.2397,
+      "step": 8209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.559280395672593e-06,
+      "loss": 0.1734,
+      "step": 8210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.55283103141687e-06,
+      "loss": 0.1415,
+      "step": 8211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.546387300289189e-06,
+      "loss": 0.1926,
+      "step": 8212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539949203071069e-06,
+      "loss": 0.217,
+      "step": 8213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5335167405433025e-06,
+      "loss": 0.1771,
+      "step": 8214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.527089913486037e-06,
+      "loss": 0.25,
+      "step": 8215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.520668722678705e-06,
+      "loss": 0.2197,
+      "step": 8216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.514253168900078e-06,
+      "loss": 0.1922,
+      "step": 8217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5078432529282447e-06,
+      "loss": 0.2335,
+      "step": 8218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.501438975540583e-06,
+      "loss": 0.2273,
+      "step": 8219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4950403375138217e-06,
+      "loss": 0.1833,
+      "step": 8220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4886473396239904e-06,
+      "loss": 0.1516,
+      "step": 8221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4822599826464187e-06,
+      "loss": 0.1995,
+      "step": 8222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4758782673557755e-06,
+      "loss": 0.1378,
+      "step": 8223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.469502194526053e-06,
+      "loss": 0.1889,
+      "step": 8224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.463131764930516e-06,
+      "loss": 0.1502,
+      "step": 8225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4567669793417967e-06,
+      "loss": 0.1724,
+      "step": 8226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.450407838531805e-06,
+      "loss": 0.1926,
+      "step": 8227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.444054343271774e-06,
+      "loss": 0.1609,
+      "step": 8228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.437706494332266e-06,
+      "loss": 0.2884,
+      "step": 8229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.431364292483152e-06,
+      "loss": 0.2263,
+      "step": 8230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4250277384936134e-06,
+      "loss": 0.2091,
+      "step": 8231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4186968331321454e-06,
+      "loss": 0.1804,
+      "step": 8232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4123715771665786e-06,
+      "loss": 0.1609,
+      "step": 8233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4060519713640216e-06,
+      "loss": 0.151,
+      "step": 8234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3997380164909387e-06,
+      "loss": 0.1489,
+      "step": 8235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3934297133130677e-06,
+      "loss": 0.1435,
+      "step": 8236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3871270625954856e-06,
+      "loss": 0.1978,
+      "step": 8237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.380830065102597e-06,
+      "loss": 0.2134,
+      "step": 8238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.374538721598086e-06,
+      "loss": 0.1755,
+      "step": 8239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.368253032844976e-06,
+      "loss": 0.2337,
+      "step": 8240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3619729996056003e-06,
+      "loss": 0.1654,
+      "step": 8241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3556986226415955e-06,
+      "loss": 0.204,
+      "step": 8242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3494299027139297e-06,
+      "loss": 0.1804,
+      "step": 8243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3431668405828675e-06,
+      "loss": 0.1826,
+      "step": 8244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3369094370079956e-06,
+      "loss": 0.2195,
+      "step": 8245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3306576927482126e-06,
+      "loss": 0.1967,
+      "step": 8246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.324411608561745e-06,
+      "loss": 0.1436,
+      "step": 8247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.318171185206098e-06,
+      "loss": 0.1532,
+      "step": 8248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.311936423438128e-06,
+      "loss": 0.1866,
+      "step": 8249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.305707324013996e-06,
+      "loss": 0.1889,
+      "step": 8250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2994838876891477e-06,
+      "loss": 0.2096,
+      "step": 8251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2932661152183743e-06,
+      "loss": 0.1762,
+      "step": 8252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2870540073557775e-06,
+      "loss": 0.1521,
+      "step": 8253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.28084756485475e-06,
+      "loss": 0.2315,
+      "step": 8254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.274646788468022e-06,
+      "loss": 0.2172,
+      "step": 8255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2684516789476206e-06,
+      "loss": 0.2304,
+      "step": 8256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.262262237044883e-06,
+      "loss": 0.2169,
+      "step": 8257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.256078463510487e-06,
+      "loss": 0.2233,
+      "step": 8258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2499003590943876e-06,
+      "loss": 0.2117,
+      "step": 8259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.243727924545864e-06,
+      "loss": 0.1962,
+      "step": 8260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2375611606135228e-06,
+      "loss": 0.1197,
+      "step": 8261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2314000680452594e-06,
+      "loss": 0.158,
+      "step": 8262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2252446475883045e-06,
+      "loss": 0.1849,
+      "step": 8263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2190948999891833e-06,
+      "loss": 0.2355,
+      "step": 8264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.212950825993738e-06,
+      "loss": 0.212,
+      "step": 8265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2068124263471276e-06,
+      "loss": 0.2939,
+      "step": 8266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.200679701793824e-06,
+      "loss": 0.1527,
+      "step": 8267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.194552653077593e-06,
+      "loss": 0.1502,
+      "step": 8268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1884312809415286e-06,
+      "loss": 0.1879,
+      "step": 8269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1823155861280485e-06,
+      "loss": 0.1467,
+      "step": 8270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1762055693788426e-06,
+      "loss": 0.2218,
+      "step": 8271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.170101231434952e-06,
+      "loss": 0.154,
+      "step": 8272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1640025730367117e-06,
+      "loss": 0.1745,
+      "step": 8273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.157909594923758e-06,
+      "loss": 0.1695,
+      "step": 8274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1518222978350664e-06,
+      "loss": 0.1979,
+      "step": 8275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1457406825088963e-06,
+      "loss": 0.2727,
+      "step": 8276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1396647496828247e-06,
+      "loss": 0.1856,
+      "step": 8277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1335945000937393e-06,
+      "loss": 0.2091,
+      "step": 8278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1275299344778575e-06,
+      "loss": 0.1815,
+      "step": 8279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.121471053570679e-06,
+      "loss": 0.1185,
+      "step": 8280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.115417858107039e-06,
+      "loss": 0.1897,
+      "step": 8281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1093703488210556e-06,
+      "loss": 0.1905,
+      "step": 8282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1033285264461754e-06,
+      "loss": 0.1586,
+      "step": 8283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0972923917151732e-06,
+      "loss": 0.1892,
+      "step": 8284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0912619453600854e-06,
+      "loss": 0.1529,
+      "step": 8285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0852371881123e-06,
+      "loss": 0.184,
+      "step": 8286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0792181207025094e-06,
+      "loss": 0.1646,
+      "step": 8287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0732047438606914e-06,
+      "loss": 0.1857,
+      "step": 8288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.067197058316157e-06,
+      "loss": 0.1924,
+      "step": 8289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0611950647975284e-06,
+      "loss": 0.1786,
+      "step": 8290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.055198764032713e-06,
+      "loss": 0.1416,
+      "step": 8291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.049208156748967e-06,
+      "loss": 0.1106,
+      "step": 8292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0432232436728157e-06,
+      "loss": 0.1904,
+      "step": 8293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0372440255301116e-06,
+      "loss": 0.142,
+      "step": 8294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.031270503046019e-06,
+      "loss": 0.2084,
+      "step": 8295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.025302676945013e-06,
+      "loss": 0.1415,
+      "step": 8296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0193405479508663e-06,
+      "loss": 0.1685,
+      "step": 8297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0133841167866715e-06,
+      "loss": 0.1663,
+      "step": 8298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.007433384174835e-06,
+      "loss": 0.1691,
+      "step": 8299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0014883508370505e-06,
+      "loss": 0.1576,
+      "step": 8300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.995549017494342e-06,
+      "loss": 0.2096,
+      "step": 8301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9896153848670285e-06,
+      "loss": 0.1416,
+      "step": 8302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9836874536747384e-06,
+      "loss": 0.1981,
+      "step": 8303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9777652246364306e-06,
+      "loss": 0.1577,
+      "step": 8304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9718486984703366e-06,
+      "loss": 0.1583,
+      "step": 8305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9659378758940205e-06,
+      "loss": 0.1405,
+      "step": 8306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.960032757624359e-06,
+      "loss": 0.1928,
+      "step": 8307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9541333443775243e-06,
+      "loss": 0.2295,
+      "step": 8308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.948239636868977e-06,
+      "loss": 0.2017,
+      "step": 8309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.942351635813534e-06,
+      "loss": 0.1876,
+      "step": 8310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.93646934192528e-06,
+      "loss": 0.1882,
+      "step": 8311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9305927559176217e-06,
+      "loss": 0.1757,
+      "step": 8312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9247218785032893e-06,
+      "loss": 0.1332,
+      "step": 8313
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.91885671039428e-06,
+      "loss": 0.1986,
+      "step": 8314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.912997252301941e-06,
+      "loss": 0.1374,
+      "step": 8315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.90714350493691e-06,
+      "loss": 0.1282,
+      "step": 8316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9012954690091188e-06,
+      "loss": 0.2253,
+      "step": 8317
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8954531452278275e-06,
+      "loss": 0.1922,
+      "step": 8318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.889616534301598e-06,
+      "loss": 0.143,
+      "step": 8319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8837856369382853e-06,
+      "loss": 0.1952,
+      "step": 8320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.87796045384508e-06,
+      "loss": 0.1502,
+      "step": 8321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8721409857284444e-06,
+      "loss": 0.1736,
+      "step": 8322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8663272332941694e-06,
+      "loss": 0.2204,
+      "step": 8323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.860519197247358e-06,
+      "loss": 0.201,
+      "step": 8324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8547168782924127e-06,
+      "loss": 0.2502,
+      "step": 8325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8489202771330216e-06,
+      "loss": 0.1405,
+      "step": 8326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.843129394472205e-06,
+      "loss": 0.1853,
+      "step": 8327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.837344231012301e-06,
+      "loss": 0.3114,
+      "step": 8328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8315647874549155e-06,
+      "loss": 0.0955,
+      "step": 8329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.825791064500993e-06,
+      "loss": 0.1567,
+      "step": 8330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.820023062850763e-06,
+      "loss": 0.1562,
+      "step": 8331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.814260783203776e-06,
+      "loss": 0.1276,
+      "step": 8332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.808504226258896e-06,
+      "loss": 0.0945,
+      "step": 8333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8027533927142525e-06,
+      "loss": 0.1539,
+      "step": 8334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.797008283267327e-06,
+      "loss": 0.1003,
+      "step": 8335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7912688986148903e-06,
+      "loss": 0.1951,
+      "step": 8336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.785535239453002e-06,
+      "loss": 0.2335,
+      "step": 8337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.779807306477056e-06,
+      "loss": 0.1957,
+      "step": 8338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.774085100381735e-06,
+      "loss": 0.2054,
+      "step": 8339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7683686218610285e-06,
+      "loss": 0.1625,
+      "step": 8340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.762657871608232e-06,
+      "loss": 0.2045,
+      "step": 8341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7569528503159525e-06,
+      "loss": 0.157,
+      "step": 8342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.751253558676087e-06,
+      "loss": 0.1398,
+      "step": 8343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7455599973798486e-06,
+      "loss": 0.1945,
+      "step": 8344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7398721671177687e-06,
+      "loss": 0.1768,
+      "step": 8345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7341900685796505e-06,
+      "loss": 0.1361,
+      "step": 8346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.728513702454627e-06,
+      "loss": 0.1736,
+      "step": 8347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7228430694311413e-06,
+      "loss": 0.2279,
+      "step": 8348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7171781701969155e-06,
+      "loss": 0.1852,
+      "step": 8349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7115190054390007e-06,
+      "loss": 0.1658,
+      "step": 8350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7058655758437303e-06,
+      "loss": 0.1703,
+      "step": 8351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7002178820967615e-06,
+      "loss": 0.1709,
+      "step": 8352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6945759248830525e-06,
+      "loss": 0.2355,
+      "step": 8353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6889397048868505e-06,
+      "loss": 0.1962,
+      "step": 8354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6833092227917247e-06,
+      "loss": 0.1277,
+      "step": 8355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.677684479280551e-06,
+      "loss": 0.2183,
+      "step": 8356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6720654750354903e-06,
+      "loss": 0.1416,
+      "step": 8357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6664522107380074e-06,
+      "loss": 0.1509,
+      "step": 8358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.660844687068903e-06,
+      "loss": 0.1829,
+      "step": 8359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6552429047082383e-06,
+      "loss": 0.1573,
+      "step": 8360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6496468643354087e-06,
+      "loss": 0.1933,
+      "step": 8361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6440565666291104e-06,
+      "loss": 0.1945,
+      "step": 8362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.638472012267329e-06,
+      "loss": 0.2002,
+      "step": 8363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6328932019273554e-06,
+      "loss": 0.1643,
+      "step": 8364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6273201362858046e-06,
+      "loss": 0.1938,
+      "step": 8365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.621752816018569e-06,
+      "loss": 0.1622,
+      "step": 8366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6161912418008526e-06,
+      "loss": 0.1795,
+      "step": 8367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6106354143071777e-06,
+      "loss": 0.1671,
+      "step": 8368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6050853342113434e-06,
+      "loss": 0.1928,
+      "step": 8369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5995410021864787e-06,
+      "loss": 0.1987,
+      "step": 8370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5940024189049892e-06,
+      "loss": 0.1775,
+      "step": 8371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5884695850385997e-06,
+      "loss": 0.1784,
+      "step": 8372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5829425012583397e-06,
+      "loss": 0.1899,
+      "step": 8373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.577421168234534e-06,
+      "loss": 0.1518,
+      "step": 8374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5719055866368024e-06,
+      "loss": 0.2477,
+      "step": 8375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5663957571340936e-06,
+      "loss": 0.1314,
+      "step": 8376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5608916803946227e-06,
+      "loss": 0.243,
+      "step": 8377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5553933570859333e-06,
+      "loss": 0.1976,
+      "step": 8378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.549900787874876e-06,
+      "loss": 0.2198,
+      "step": 8379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5444139734275673e-06,
+      "loss": 0.1873,
+      "step": 8380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.53893291440947e-06,
+      "loss": 0.1772,
+      "step": 8381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5334576114853237e-06,
+      "loss": 0.1587,
+      "step": 8382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.52798806531917e-06,
+      "loss": 0.2172,
+      "step": 8383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5225242765743616e-06,
+      "loss": 0.1646,
+      "step": 8384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.517066245913552e-06,
+      "loss": 0.2298,
+      "step": 8385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.511613973998683e-06,
+      "loss": 0.2297,
+      "step": 8386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.50616746149101e-06,
+      "loss": 0.1915,
+      "step": 8387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5007267090511043e-06,
+      "loss": 0.2028,
+      "step": 8388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.495291717338799e-06,
+      "loss": 0.1153,
+      "step": 8389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.489862487013267e-06,
+      "loss": 0.2012,
+      "step": 8390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.484439018732965e-06,
+      "loss": 0.199,
+      "step": 8391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.479021313155644e-06,
+      "loss": 0.1869,
+      "step": 8392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.473609370938368e-06,
+      "loss": 0.2103,
+      "step": 8393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4682031927375117e-06,
+      "loss": 0.1678,
+      "step": 8394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.462802779208723e-06,
+      "loss": 0.1401,
+      "step": 8395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4574081310069774e-06,
+      "loss": 0.202,
+      "step": 8396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.45201924878653e-06,
+      "loss": 0.1569,
+      "step": 8397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.446636133200947e-06,
+      "loss": 0.1056,
+      "step": 8398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4412587849031e-06,
+      "loss": 0.1336,
+      "step": 8399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4358872045451563e-06,
+      "loss": 0.2112,
+      "step": 8400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.430521392778573e-06,
+      "loss": 0.237,
+      "step": 8401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4251613502541347e-06,
+      "loss": 0.2214,
+      "step": 8402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4198070776218885e-06,
+      "loss": 0.1686,
+      "step": 8403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4144585755312143e-06,
+      "loss": 0.188,
+      "step": 8404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4091158446307827e-06,
+      "loss": 0.1651,
+      "step": 8405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4037788855685583e-06,
+      "loss": 0.1884,
+      "step": 8406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.398447698991807e-06,
+      "loss": 0.1455,
+      "step": 8407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3931222855471056e-06,
+      "loss": 0.2124,
+      "step": 8408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.387802645880305e-06,
+      "loss": 0.2198,
+      "step": 8409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3824887806365825e-06,
+      "loss": 0.2308,
+      "step": 8410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3771806904604123e-06,
+      "loss": 0.1945,
+      "step": 8411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.371878375995551e-06,
+      "loss": 0.167,
+      "step": 8412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.366581837885068e-06,
+      "loss": 0.1827,
+      "step": 8413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3612910767713327e-06,
+      "loss": 0.1497,
+      "step": 8414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.356006093296009e-06,
+      "loss": 0.1729,
+      "step": 8415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.350726888100058e-06,
+      "loss": 0.2295,
+      "step": 8416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3454534618237432e-06,
+      "loss": 0.1627,
+      "step": 8417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3401858151066325e-06,
+      "loss": 0.122,
+      "step": 8418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3349239485875918e-06,
+      "loss": 0.1338,
+      "step": 8419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3296678629047663e-06,
+      "loss": 0.201,
+      "step": 8420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.324417558695624e-06,
+      "loss": 0.1523,
+      "step": 8421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.319173036596933e-06,
+      "loss": 0.2424,
+      "step": 8422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.313934297244741e-06,
+      "loss": 0.2088,
+      "step": 8423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3087013412744e-06,
+      "loss": 0.1171,
+      "step": 8424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3034741693205743e-06,
+      "loss": 0.1511,
+      "step": 8425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.298252782017207e-06,
+      "loss": 0.2522,
+      "step": 8426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2930371799975594e-06,
+      "loss": 0.1466,
+      "step": 8427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2878273638941794e-06,
+      "loss": 0.1705,
+      "step": 8428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.282623334338907e-06,
+      "loss": 0.1882,
+      "step": 8429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2774250919628982e-06,
+      "loss": 0.1466,
+      "step": 8430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2722326373965983e-06,
+      "loss": 0.1828,
+      "step": 8431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2670459712697377e-06,
+      "loss": 0.2024,
+      "step": 8432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2618650942113683e-06,
+      "loss": 0.2301,
+      "step": 8433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.256690006849832e-06,
+      "loss": 0.1496,
+      "step": 8434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.251520709812749e-06,
+      "loss": 0.1403,
+      "step": 8435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.246357203727073e-06,
+      "loss": 0.1369,
+      "step": 8436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2411994892190147e-06,
+      "loss": 0.1843,
+      "step": 8437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2360475669141178e-06,
+      "loss": 0.1915,
+      "step": 8438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2309014374372106e-06,
+      "loss": 0.1325,
+      "step": 8439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.225761101412416e-06,
+      "loss": 0.2132,
+      "step": 8440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2206265594631405e-06,
+      "loss": 0.2306,
+      "step": 8441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2154978122121194e-06,
+      "loss": 0.1697,
+      "step": 8442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.210374860281367e-06,
+      "loss": 0.2428,
+      "step": 8443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.205257704292185e-06,
+      "loss": 0.1691,
+      "step": 8444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.200146344865206e-06,
+      "loss": 0.1292,
+      "step": 8445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.195040782620311e-06,
+      "loss": 0.1968,
+      "step": 8446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1899410181767223e-06,
+      "loss": 0.2067,
+      "step": 8447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1848470521529384e-06,
+      "loss": 0.2167,
+      "step": 8448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1797588851667496e-06,
+      "loss": 0.19,
+      "step": 8449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1746765178352558e-06,
+      "loss": 0.1948,
+      "step": 8450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1695999507748534e-06,
+      "loss": 0.2088,
+      "step": 8451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1645291846012216e-06,
+      "loss": 0.1974,
+      "step": 8452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.159464219929347e-06,
+      "loss": 0.2524,
+      "step": 8453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1544050573735154e-06,
+      "loss": 0.2079,
+      "step": 8454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1493516975472914e-06,
+      "loss": 0.2045,
+      "step": 8455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1443041410635635e-06,
+      "loss": 0.2217,
+      "step": 8456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1392623885344964e-06,
+      "loss": 0.1763,
+      "step": 8457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1342264405715406e-06,
+      "loss": 0.1644,
+      "step": 8458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.129196297785474e-06,
+      "loss": 0.2052,
+      "step": 8459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.124171960786353e-06,
+      "loss": 0.1901,
+      "step": 8460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1191534301835235e-06,
+      "loss": 0.1406,
+      "step": 8461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1141407065856376e-06,
+      "loss": 0.1615,
+      "step": 8462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1091337906006482e-06,
+      "loss": 0.1387,
+      "step": 8463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1041326828357798e-06,
+      "loss": 0.1758,
+      "step": 8464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.099137383897587e-06,
+      "loss": 0.1956,
+      "step": 8465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0941478943918792e-06,
+      "loss": 0.195,
+      "step": 8466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0891642149238e-06,
+      "loss": 0.1687,
+      "step": 8467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.084186346097772e-06,
+      "loss": 0.1964,
+      "step": 8468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.079214288517506e-06,
+      "loss": 0.1757,
+      "step": 8469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.074248042786009e-06,
+      "loss": 0.223,
+      "step": 8470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.06928760950561e-06,
+      "loss": 0.1508,
+      "step": 8471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.064332989277895e-06,
+      "loss": 0.2454,
+      "step": 8472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.059384182703761e-06,
+      "loss": 0.1577,
+      "step": 8473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0544411903834106e-06,
+      "loss": 0.2229,
+      "step": 8474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.04950401291632e-06,
+      "loss": 0.2214,
+      "step": 8475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0445726509012773e-06,
+      "loss": 0.1647,
+      "step": 8476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.039647104936371e-06,
+      "loss": 0.2031,
+      "step": 8477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0347273756189554e-06,
+      "loss": 0.2034,
+      "step": 8478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0298134635456987e-06,
+      "loss": 0.1794,
+      "step": 8479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0249053693125795e-06,
+      "loss": 0.1707,
+      "step": 8480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.020003093514833e-06,
+      "loss": 0.2271,
+      "step": 8481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.015106636747016e-06,
+      "loss": 0.172,
+      "step": 8482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.010215999602977e-06,
+      "loss": 0.2015,
+      "step": 8483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.005331182675846e-06,
+      "loss": 0.1597,
+      "step": 8484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.000452186558066e-06,
+      "loss": 0.1975,
+      "step": 8485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.995579011841353e-06,
+      "loss": 0.1733,
+      "step": 8486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9907116591167287e-06,
+      "loss": 0.201,
+      "step": 8487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9858501289745146e-06,
+      "loss": 0.1524,
+      "step": 8488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.980994422004312e-06,
+      "loss": 0.1363,
+      "step": 8489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9761445387950217e-06,
+      "loss": 0.2113,
+      "step": 8490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9713004799348455e-06,
+      "loss": 0.1634,
+      "step": 8491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.966462246011258e-06,
+      "loss": 0.1629,
+      "step": 8492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.961629837611062e-06,
+      "loss": 0.1989,
+      "step": 8493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.956803255320322e-06,
+      "loss": 0.1488,
+      "step": 8494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9519824997244087e-06,
+      "loss": 0.1389,
+      "step": 8495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.947167571407982e-06,
+      "loss": 0.1503,
+      "step": 8496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.942358470955008e-06,
+      "loss": 0.1971,
+      "step": 8497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9375551989487262e-06,
+      "loss": 0.2169,
+      "step": 8498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9327577559716816e-06,
+      "loss": 0.1789,
+      "step": 8499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.927966142605714e-06,
+      "loss": 0.1328,
+      "step": 8500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9231803594319475e-06,
+      "loss": 0.1843,
+      "step": 8501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9184004070308015e-06,
+      "loss": 0.1773,
+      "step": 8502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.913626285982001e-06,
+      "loss": 0.2102,
+      "step": 8503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.908857996864533e-06,
+      "loss": 0.1616,
+      "step": 8504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9040955402567228e-06,
+      "loss": 0.2555,
+      "step": 8505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.899338916736143e-06,
+      "loss": 0.1752,
+      "step": 8506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8945881268796818e-06,
+      "loss": 0.1821,
+      "step": 8507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8898431712635222e-06,
+      "loss": 0.2178,
+      "step": 8508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8851040504631322e-06,
+      "loss": 0.1127,
+      "step": 8509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8803707650532686e-06,
+      "loss": 0.1598,
+      "step": 8510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.875643315608e-06,
+      "loss": 0.1293,
+      "step": 8511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.870921702700651e-06,
+      "loss": 0.1953,
+      "step": 8512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.86620592690388e-06,
+      "loss": 0.1252,
+      "step": 8513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8614959887896077e-06,
+      "loss": 0.1816,
+      "step": 8514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8567918889290602e-06,
+      "loss": 0.1936,
+      "step": 8515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.852093627892748e-06,
+      "loss": 0.1602,
+      "step": 8516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.847401206250493e-06,
+      "loss": 0.1779,
+      "step": 8517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8427146245713733e-06,
+      "loss": 0.1506,
+      "step": 8518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8380338834237841e-06,
+      "loss": 0.1798,
+      "step": 8519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8333589833754216e-06,
+      "loss": 0.1433,
+      "step": 8520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8286899249932487e-06,
+      "loss": 0.2316,
+      "step": 8521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8240267088435236e-06,
+      "loss": 0.1576,
+      "step": 8522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8193693354918163e-06,
+      "loss": 0.2085,
+      "step": 8523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8147178055029579e-06,
+      "loss": 0.1714,
+      "step": 8524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.810072119441103e-06,
+      "loss": 0.2023,
+      "step": 8525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8054322778696786e-06,
+      "loss": 0.1881,
+      "step": 8526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8007982813514013e-06,
+      "loss": 0.2128,
+      "step": 8527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7961701304482826e-06,
+      "loss": 0.1772,
+      "step": 8528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.79154782572164e-06,
+      "loss": 0.1935,
+      "step": 8529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7869313677320475e-06,
+      "loss": 0.1954,
+      "step": 8530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7823207570394074e-06,
+      "loss": 0.2062,
+      "step": 8531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7777159942028887e-06,
+      "loss": 0.164,
+      "step": 8532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.773117079780956e-06,
+      "loss": 0.1577,
+      "step": 8533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7685240143313798e-06,
+      "loss": 0.1993,
+      "step": 8534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7639367984111922e-06,
+      "loss": 0.2283,
+      "step": 8535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7593554325767426e-06,
+      "loss": 0.2395,
+      "step": 8536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7547799173836587e-06,
+      "loss": 0.2725,
+      "step": 8537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7502102533868637e-06,
+      "loss": 0.2116,
+      "step": 8538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7456464411405527e-06,
+      "loss": 0.178,
+      "step": 8539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7410884811982497e-06,
+      "loss": 0.2125,
+      "step": 8540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7365363741127238e-06,
+      "loss": 0.2589,
+      "step": 8541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7319901204360666e-06,
+      "loss": 0.2073,
+      "step": 8542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7274497207196594e-06,
+      "loss": 0.1839,
+      "step": 8543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7229151755141392e-06,
+      "loss": 0.1381,
+      "step": 8544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.718386485369472e-06,
+      "loss": 0.1758,
+      "step": 8545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7138636508349071e-06,
+      "loss": 0.1629,
+      "step": 8546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7093466724589557e-06,
+      "loss": 0.1916,
+      "step": 8547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7048355507894465e-06,
+      "loss": 0.1716,
+      "step": 8548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7003302863735026e-06,
+      "loss": 0.1947,
+      "step": 8549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.695830879757504e-06,
+      "loss": 0.1864,
+      "step": 8550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6913373314871528e-06,
+      "loss": 0.1829,
+      "step": 8551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6868496421074243e-06,
+      "loss": 0.0992,
+      "step": 8552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.682367812162583e-06,
+      "loss": 0.2006,
+      "step": 8553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6778918421961941e-06,
+      "loss": 0.2233,
+      "step": 8554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.673421732751107e-06,
+      "loss": 0.2653,
+      "step": 8555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6689574843694433e-06,
+      "loss": 0.1771,
+      "step": 8556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6644990975926367e-06,
+      "loss": 0.1159,
+      "step": 8557
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.66004657296141e-06,
+      "loss": 0.2433,
+      "step": 8558
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6555999110157538e-06,
+      "loss": 0.2441,
+      "step": 8559
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.65115911229497e-06,
+      "loss": 0.18,
+      "step": 8560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.646724177337633e-06,
+      "loss": 0.1536,
+      "step": 8561
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6422951066816184e-06,
+      "loss": 0.2101,
+      "step": 8562
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6378719008640854e-06,
+      "loss": 0.1435,
+      "step": 8563
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6334545604214768e-06,
+      "loss": 0.2244,
+      "step": 8564
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6290430858895367e-06,
+      "loss": 0.119,
+      "step": 8565
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6246374778032868e-06,
+      "loss": 0.187,
+      "step": 8566
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6202377366970334e-06,
+      "loss": 0.1298,
+      "step": 8567
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6158438631043882e-06,
+      "loss": 0.1494,
+      "step": 8568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6114558575582417e-06,
+      "loss": 0.1599,
+      "step": 8569
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6070737205907682e-06,
+      "loss": 0.1552,
+      "step": 8570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6026974527334371e-06,
+      "loss": 0.1781,
+      "step": 8571
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.598327054517007e-06,
+      "loss": 0.1879,
+      "step": 8572
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5939625264715042e-06,
+      "loss": 0.1106,
+      "step": 8573
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.589603869126277e-06,
+      "loss": 0.1922,
+      "step": 8574
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5852510830099476e-06,
+      "loss": 0.1789,
+      "step": 8575
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.58090416865041e-06,
+      "loss": 0.2052,
+      "step": 8576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5765631265748649e-06,
+      "loss": 0.1481,
+      "step": 8577
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.572227957309802e-06,
+      "loss": 0.1546,
+      "step": 8578
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5678986613809787e-06,
+      "loss": 0.0945,
+      "step": 8579
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5635752393134695e-06,
+      "loss": 0.2257,
+      "step": 8580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.559257691631605e-06,
+      "loss": 0.1397,
+      "step": 8581
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.554946018859027e-06,
+      "loss": 0.1705,
+      "step": 8582
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5506402215186566e-06,
+      "loss": 0.2227,
+      "step": 8583
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.546340300132698e-06,
+      "loss": 0.1281,
+      "step": 8584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.542046255222651e-06,
+      "loss": 0.2316,
+      "step": 8585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5377580873093045e-06,
+      "loss": 0.1336,
+      "step": 8586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.53347579691272e-06,
+      "loss": 0.1145,
+      "step": 8587
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5291993845522545e-06,
+      "loss": 0.1538,
+      "step": 8588
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5249288507465599e-06,
+      "loss": 0.1206,
+      "step": 8589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5206641960135604e-06,
+      "loss": 0.1375,
+      "step": 8590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5164054208704703e-06,
+      "loss": 0.2301,
+      "step": 8591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5121525258338154e-06,
+      "loss": 0.2855,
+      "step": 8592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.507905511419372e-06,
+      "loss": 0.1374,
+      "step": 8593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5036643781422167e-06,
+      "loss": 0.2662,
+      "step": 8594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4994291265167326e-06,
+      "loss": 0.2225,
+      "step": 8595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4951997570565534e-06,
+      "loss": 0.1986,
+      "step": 8596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.49097627027463e-06,
+      "loss": 0.2749,
+      "step": 8597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4867586666831857e-06,
+      "loss": 0.1997,
+      "step": 8598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4825469467937336e-06,
+      "loss": 0.2248,
+      "step": 8599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4783411111170708e-06,
+      "loss": 0.125,
+      "step": 8600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4741411601632781e-06,
+      "loss": 0.2395,
+      "step": 8601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.469947094441737e-06,
+      "loss": 0.1202,
+      "step": 8602
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4657589144611017e-06,
+      "loss": 0.1531,
+      "step": 8603
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4615766207293157e-06,
+      "loss": 0.2537,
+      "step": 8604
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4574002137536014e-06,
+      "loss": 0.209,
+      "step": 8605
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.453229694040481e-06,
+      "loss": 0.1858,
+      "step": 8606
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4490650620957557e-06,
+      "loss": 0.2372,
+      "step": 8607
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.444906318424516e-06,
+      "loss": 0.1935,
+      "step": 8608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4407534635311414e-06,
+      "loss": 0.2083,
+      "step": 8609
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.436606497919274e-06,
+      "loss": 0.1904,
+      "step": 8610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4324654220918721e-06,
+      "loss": 0.19,
+      "step": 8611
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4283302365511675e-06,
+      "loss": 0.1539,
+      "step": 8612
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4242009417986646e-06,
+      "loss": 0.1671,
+      "step": 8613
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4200775383351794e-06,
+      "loss": 0.194,
+      "step": 8614
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4159600266607953e-06,
+      "loss": 0.2048,
+      "step": 8615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.411848407274885e-06,
+      "loss": 0.1907,
+      "step": 8616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4077426806761052e-06,
+      "loss": 0.1733,
+      "step": 8617
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4036428473624019e-06,
+      "loss": 0.2092,
+      "step": 8618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3995489078310054e-06,
+      "loss": 0.1793,
+      "step": 8619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3954608625784294e-06,
+      "loss": 0.2054,
+      "step": 8620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3913787121004719e-06,
+      "loss": 0.2114,
+      "step": 8621
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3873024568922144e-06,
+      "loss": 0.1649,
+      "step": 8622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3832320974480284e-06,
+      "loss": 0.174,
+      "step": 8623
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.37916763426158e-06,
+      "loss": 0.2164,
+      "step": 8624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3751090678257861e-06,
+      "loss": 0.1832,
+      "step": 8625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3710563986328972e-06,
+      "loss": 0.2096,
+      "step": 8626
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3670096271743982e-06,
+      "loss": 0.1701,
+      "step": 8627
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3629687539410906e-06,
+      "loss": 0.1194,
+      "step": 8628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3589337794230661e-06,
+      "loss": 0.1551,
+      "step": 8629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3549047041096718e-06,
+      "loss": 0.1548,
+      "step": 8630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.350881528489556e-06,
+      "loss": 0.1396,
+      "step": 8631
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.346864253050667e-06,
+      "loss": 0.1294,
+      "step": 8632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3428528782801985e-06,
+      "loss": 0.1674,
+      "step": 8633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.338847404664667e-06,
+      "loss": 0.1654,
+      "step": 8634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3348478326898562e-06,
+      "loss": 0.1989,
+      "step": 8635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3308541628408388e-06,
+      "loss": 0.218,
+      "step": 8636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.32686639560195e-06,
+      "loss": 0.1825,
+      "step": 8637
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3228845314568472e-06,
+      "loss": 0.1764,
+      "step": 8638
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3189085708884386e-06,
+      "loss": 0.1914,
+      "step": 8639
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3149385143789382e-06,
+      "loss": 0.1669,
+      "step": 8640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3109743624098392e-06,
+      "loss": 0.109,
+      "step": 8641
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3070161154619009e-06,
+      "loss": 0.1657,
+      "step": 8642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3030637740151953e-06,
+      "loss": 0.1719,
+      "step": 8643
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2991173385490607e-06,
+      "loss": 0.1676,
+      "step": 8644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2951768095421146e-06,
+      "loss": 0.1963,
+      "step": 8645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2912421874722747e-06,
+      "loss": 0.1402,
+      "step": 8646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2873134728167258e-06,
+      "loss": 0.2015,
+      "step": 8647
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2833906660519423e-06,
+      "loss": 0.1284,
+      "step": 8648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2794737676536994e-06,
+      "loss": 0.2021,
+      "step": 8649
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2755627780970225e-06,
+      "loss": 0.1435,
+      "step": 8650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2716576978562434e-06,
+      "loss": 0.1702,
+      "step": 8651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2677585274049774e-06,
+      "loss": 0.2183,
+      "step": 8652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2638652672161077e-06,
+      "loss": 0.1404,
+      "step": 8653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2599779177618177e-06,
+      "loss": 0.2293,
+      "step": 8654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2560964795135632e-06,
+      "loss": 0.2164,
+      "step": 8655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2522209529420792e-06,
+      "loss": 0.1792,
+      "step": 8656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2483513385174006e-06,
+      "loss": 0.121,
+      "step": 8657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2444876367088354e-06,
+      "loss": 0.2102,
+      "step": 8658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.24062984798497e-06,
+      "loss": 0.1929,
+      "step": 8659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2367779728136796e-06,
+      "loss": 0.1874,
+      "step": 8660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2329320116621247e-06,
+      "loss": 0.17,
+      "step": 8661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.229091964996737e-06,
+      "loss": 0.1663,
+      "step": 8662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2252578332832442e-06,
+      "loss": 0.1668,
+      "step": 8663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.221429616986658e-06,
+      "loss": 0.1515,
+      "step": 8664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.217607316571251e-06,
+      "loss": 0.1864,
+      "step": 8665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2137909325006025e-06,
+      "loss": 0.1361,
+      "step": 8666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.209980465237559e-06,
+      "loss": 0.156,
+      "step": 8667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2061759152442565e-06,
+      "loss": 0.0951,
+      "step": 8668
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.20237728298212e-06,
+      "loss": 0.1628,
+      "step": 8669
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.198584568911848e-06,
+      "loss": 0.2276,
+      "step": 8670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1947977734934056e-06,
+      "loss": 0.1948,
+      "step": 8671
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1910168971860702e-06,
+      "loss": 0.2668,
+      "step": 8672
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1872419404483914e-06,
+      "loss": 0.1509,
+      "step": 8673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1834729037381865e-06,
+      "loss": 0.2214,
+      "step": 8674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1797097875125785e-06,
+      "loss": 0.1797,
+      "step": 8675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1759525922279412e-06,
+      "loss": 0.1745,
+      "step": 8676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1722013183399606e-06,
+      "loss": 0.2251,
+      "step": 8677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1684559663035942e-06,
+      "loss": 0.1824,
+      "step": 8678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1647165365730738e-06,
+      "loss": 0.1648,
+      "step": 8679
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1609830296019143e-06,
+      "loss": 0.2398,
+      "step": 8680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1572554458429362e-06,
+      "loss": 0.1774,
+      "step": 8681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1535337857482008e-06,
+      "loss": 0.2075,
+      "step": 8682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1498180497690747e-06,
+      "loss": 0.1714,
+      "step": 8683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1461082383562194e-06,
+      "loss": 0.2494,
+      "step": 8684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1424043519595529e-06,
+      "loss": 0.1594,
+      "step": 8685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.138706391028277e-06,
+      "loss": 0.2157,
+      "step": 8686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1350143560108938e-06,
+      "loss": 0.1589,
+      "step": 8687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1313282473551623e-06,
+      "loss": 0.192,
+      "step": 8688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1276480655081412e-06,
+      "loss": 0.1703,
+      "step": 8689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.123973810916168e-06,
+      "loss": 0.2141,
+      "step": 8690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1203054840248472e-06,
+      "loss": 0.1987,
+      "step": 8691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1166430852790789e-06,
+      "loss": 0.1914,
+      "step": 8692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1129866151230517e-06,
+      "loss": 0.2274,
+      "step": 8693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1093360740002057e-06,
+      "loss": 0.1238,
+      "step": 8694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1056914623532977e-06,
+      "loss": 0.2468,
+      "step": 8695
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1020527806243298e-06,
+      "loss": 0.1457,
+      "step": 8696
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.09842002925461e-06,
+      "loss": 0.1797,
+      "step": 8697
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0947932086847245e-06,
+      "loss": 0.1704,
+      "step": 8698
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.091172319354522e-06,
+      "loss": 0.156,
+      "step": 8699
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0875573617031566e-06,
+      "loss": 0.1894,
+      "step": 8700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.083948336169055e-06,
+      "loss": 0.22,
+      "step": 8701
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0803452431899175e-06,
+      "loss": 0.2041,
+      "step": 8702
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0767480832027166e-06,
+      "loss": 0.1656,
+      "step": 8703
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0731568566437312e-06,
+      "loss": 0.1799,
+      "step": 8704
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.069571563948496e-06,
+      "loss": 0.1816,
+      "step": 8705
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0659922055518467e-06,
+      "loss": 0.1276,
+      "step": 8706
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0624187818878862e-06,
+      "loss": 0.24,
+      "step": 8707
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0588512933899896e-06,
+      "loss": 0.1482,
+      "step": 8708
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.055289740490839e-06,
+      "loss": 0.1648,
+      "step": 8709
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0517341236223776e-06,
+      "loss": 0.2166,
+      "step": 8710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0481844432158161e-06,
+      "loss": 0.2003,
+      "step": 8711
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0446406997016822e-06,
+      "loss": 0.2488,
+      "step": 8712
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.041102893509749e-06,
+      "loss": 0.3207,
+      "step": 8713
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0375710250690896e-06,
+      "loss": 0.2337,
+      "step": 8714
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.034045094808045e-06,
+      "loss": 0.1181,
+      "step": 8715
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0305251031542452e-06,
+      "loss": 0.1427,
+      "step": 8716
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0270110505345931e-06,
+      "loss": 0.1376,
+      "step": 8717
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0235029373752758e-06,
+      "loss": 0.1992,
+      "step": 8718
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0200007641017583e-06,
+      "loss": 0.1794,
+      "step": 8719
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0165045311387844e-06,
+      "loss": 0.14,
+      "step": 8720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0130142389103813e-06,
+      "loss": 0.1639,
+      "step": 8721
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.009529887839844e-06,
+      "loss": 0.1716,
+      "step": 8722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0060514783497676e-06,
+      "loss": 0.2013,
+      "step": 8723
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0025790108620092e-06,
+      "loss": 0.1863,
+      "step": 8724
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.991124857977098e-07,
+      "loss": 0.2162,
+      "step": 8725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.956519035772883e-07,
+      "loss": 0.1514,
+      "step": 8726
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.921972646204536e-07,
+      "loss": 0.2114,
+      "step": 8727
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.887485693461763e-07,
+      "loss": 0.1574,
+      "step": 8728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.853058181727215e-07,
+      "loss": 0.1743,
+      "step": 8729
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.818690115176278e-07,
+      "loss": 0.1781,
+      "step": 8730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.784381497977113e-07,
+      "loss": 0.171,
+      "step": 8731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.750132334290618e-07,
+      "loss": 0.2329,
+      "step": 8732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.715942628270636e-07,
+      "loss": 0.1413,
+      "step": 8733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.681812384063626e-07,
+      "loss": 0.1891,
+      "step": 8734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.647741605808947e-07,
+      "loss": 0.1975,
+      "step": 8735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.61373029763879e-07,
+      "loss": 0.1457,
+      "step": 8736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.579778463677858e-07,
+      "loss": 0.2048,
+      "step": 8737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.545886108044021e-07,
+      "loss": 0.1621,
+      "step": 8738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.512053234847773e-07,
+      "loss": 0.1827,
+      "step": 8739
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.478279848192217e-07,
+      "loss": 0.1589,
+      "step": 8740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.444565952173523e-07,
+      "loss": 0.1605,
+      "step": 8741
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.410911550880475e-07,
+      "loss": 0.195,
+      "step": 8742
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.377316648394696e-07,
+      "loss": 0.1873,
+      "step": 8743
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.34378124879065e-07,
+      "loss": 0.1335,
+      "step": 8744
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.310305356135418e-07,
+      "loss": 0.1925,
+      "step": 8745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.276888974489029e-07,
+      "loss": 0.1883,
+      "step": 8746
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.24353210790424e-07,
+      "loss": 0.178,
+      "step": 8747
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.210234760426484e-07,
+      "loss": 0.1352,
+      "step": 8748
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.176996936094195e-07,
+      "loss": 0.1754,
+      "step": 8749
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.143818638938484e-07,
+      "loss": 0.1307,
+      "step": 8750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.110699872983131e-07,
+      "loss": 0.1579,
+      "step": 8751
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.077640642244811e-07,
+      "loss": 0.1872,
+      "step": 8752
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.044640950732986e-07,
+      "loss": 0.1844,
+      "step": 8753
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.011700802449841e-07,
+      "loss": 0.2002,
+      "step": 8754
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.978820201390348e-07,
+      "loss": 0.1931,
+      "step": 8755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.945999151542428e-07,
+      "loss": 0.2714,
+      "step": 8756
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.913237656886398e-07,
+      "loss": 0.1166,
+      "step": 8757
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.880535721395744e-07,
+      "loss": 0.227,
+      "step": 8758
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.847893349036517e-07,
+      "loss": 0.2077,
+      "step": 8759
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.815310543767607e-07,
+      "loss": 0.2747,
+      "step": 8760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.782787309540685e-07,
+      "loss": 0.1593,
+      "step": 8761
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.750323650300152e-07,
+      "loss": 0.1949,
+      "step": 8762
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.717919569983191e-07,
+      "loss": 0.2667,
+      "step": 8763
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.685575072519825e-07,
+      "loss": 0.1967,
+      "step": 8764
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.65329016183275e-07,
+      "loss": 0.1781,
+      "step": 8765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.6210648418375e-07,
+      "loss": 0.2114,
+      "step": 8766
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.588899116442506e-07,
+      "loss": 0.1493,
+      "step": 8767
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.556792989548701e-07,
+      "loss": 0.2917,
+      "step": 8768
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.52474646504986e-07,
+      "loss": 0.1784,
+      "step": 8769
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.492759546832818e-07,
+      "loss": 0.2269,
+      "step": 8770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.460832238776751e-07,
+      "loss": 0.188,
+      "step": 8771
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.428964544753893e-07,
+      "loss": 0.1499,
+      "step": 8772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.397156468629208e-07,
+      "loss": 0.1902,
+      "step": 8773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.365408014260279e-07,
+      "loss": 0.1684,
+      "step": 8774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.333719185497691e-07,
+      "loss": 0.2216,
+      "step": 8775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.302089986184646e-07,
+      "loss": 0.1272,
+      "step": 8776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.270520420157135e-07,
+      "loss": 0.1933,
+      "step": 8777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.239010491243871e-07,
+      "loss": 0.1439,
+      "step": 8778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.207560203266463e-07,
+      "loss": 0.238,
+      "step": 8779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.176169560039137e-07,
+      "loss": 0.214,
+      "step": 8780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.144838565369073e-07,
+      "loss": 0.2155,
+      "step": 8781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.113567223056007e-07,
+      "loss": 0.2245,
+      "step": 8782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.082355536892516e-07,
+      "loss": 0.1346,
+      "step": 8783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.051203510664129e-07,
+      "loss": 0.1692,
+      "step": 8784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.020111148148823e-07,
+      "loss": 0.2516,
+      "step": 8785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.989078453117472e-07,
+      "loss": 0.1861,
+      "step": 8786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.958105429333784e-07,
+      "loss": 0.2001,
+      "step": 8787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.927192080554258e-07,
+      "loss": 0.2514,
+      "step": 8788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.896338410527948e-07,
+      "loss": 0.2044,
+      "step": 8789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.865544422996862e-07,
+      "loss": 0.1882,
+      "step": 8790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.834810121695679e-07,
+      "loss": 0.1638,
+      "step": 8791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.804135510351862e-07,
+      "loss": 0.1969,
+      "step": 8792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.773520592685657e-07,
+      "loss": 0.1637,
+      "step": 8793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.742965372410093e-07,
+      "loss": 0.1523,
+      "step": 8794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.71246985323082e-07,
+      "loss": 0.2221,
+      "step": 8795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.682034038846431e-07,
+      "loss": 0.1689,
+      "step": 8796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.651657932948087e-07,
+      "loss": 0.1881,
+      "step": 8797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.62134153921995e-07,
+      "loss": 0.228,
+      "step": 8798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.591084861338749e-07,
+      "loss": 0.1378,
+      "step": 8799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.56088790297399e-07,
+      "loss": 0.2015,
+      "step": 8800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.530750667787967e-07,
+      "loss": 0.1409,
+      "step": 8801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.500673159435756e-07,
+      "loss": 0.1647,
+      "step": 8802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.470655381565217e-07,
+      "loss": 0.1536,
+      "step": 8803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.44069733781677e-07,
+      "loss": 0.2301,
+      "step": 8804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.4107990318239e-07,
+      "loss": 0.2667,
+      "step": 8805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.380960467212594e-07,
+      "loss": 0.1061,
+      "step": 8806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.351181647601735e-07,
+      "loss": 0.1491,
+      "step": 8807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.321462576602877e-07,
+      "loss": 0.2074,
+      "step": 8808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.291803257820306e-07,
+      "loss": 0.2141,
+      "step": 8809
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.262203694851255e-07,
+      "loss": 0.2035,
+      "step": 8810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.232663891285407e-07,
+      "loss": 0.1768,
+      "step": 8811
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.203183850705452e-07,
+      "loss": 0.1486,
+      "step": 8812
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.173763576686753e-07,
+      "loss": 0.1172,
+      "step": 8813
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.144403072797345e-07,
+      "loss": 0.2099,
+      "step": 8814
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.115102342598101e-07,
+      "loss": 0.1593,
+      "step": 8815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.085861389642678e-07,
+      "loss": 0.1467,
+      "step": 8816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.056680217477407e-07,
+      "loss": 0.1681,
+      "step": 8817
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.02755882964129e-07,
+      "loss": 0.2243,
+      "step": 8818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.998497229666335e-07,
+      "loss": 0.2637,
+      "step": 8819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.969495421076999e-07,
+      "loss": 0.1287,
+      "step": 8820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.940553407390693e-07,
+      "loss": 0.2187,
+      "step": 8821
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.911671192117608e-07,
+      "loss": 0.2503,
+      "step": 8822
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.882848778760387e-07,
+      "loss": 0.1762,
+      "step": 8823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.854086170814788e-07,
+      "loss": 0.1314,
+      "step": 8824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.825383371769134e-07,
+      "loss": 0.1731,
+      "step": 8825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.796740385104416e-07,
+      "loss": 0.1906,
+      "step": 8826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.768157214294524e-07,
+      "loss": 0.1868,
+      "step": 8827
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.739633862806072e-07,
+      "loss": 0.1303,
+      "step": 8828
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.711170334098293e-07,
+      "loss": 0.281,
+      "step": 8829
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.682766631623316e-07,
+      "loss": 0.1672,
+      "step": 8830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.65442275882594e-07,
+      "loss": 0.1655,
+      "step": 8831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.626138719143692e-07,
+      "loss": 0.2339,
+      "step": 8832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.597914516006942e-07,
+      "loss": 0.1839,
+      "step": 8833
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.569750152838672e-07,
+      "loss": 0.165,
+      "step": 8834
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.54164563305465e-07,
+      "loss": 0.2224,
+      "step": 8835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.513600960063482e-07,
+      "loss": 0.1939,
+      "step": 8836
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.485616137266337e-07,
+      "loss": 0.1766,
+      "step": 8837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.457691168057223e-07,
+      "loss": 0.1787,
+      "step": 8838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.429826055822985e-07,
+      "loss": 0.2342,
+      "step": 8839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.402020803943032e-07,
+      "loss": 0.2134,
+      "step": 8840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.374275415789666e-07,
+      "loss": 0.2319,
+      "step": 8841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.346589894727806e-07,
+      "loss": 0.252,
+      "step": 8842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3189642441151e-07,
+      "loss": 0.1247,
+      "step": 8843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.291398467302146e-07,
+      "loss": 0.1568,
+      "step": 8844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.263892567631991e-07,
+      "loss": 0.2464,
+      "step": 8845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.236446548440633e-07,
+      "loss": 0.1644,
+      "step": 8846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2090604130568e-07,
+      "loss": 0.1515,
+      "step": 8847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.181734164801723e-07,
+      "loss": 0.132,
+      "step": 8848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.154467806989639e-07,
+      "loss": 0.2029,
+      "step": 8849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.127261342927404e-07,
+      "loss": 0.1941,
+      "step": 8850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.100114775914711e-07,
+      "loss": 0.1224,
+      "step": 8851
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.073028109243706e-07,
+      "loss": 0.239,
+      "step": 8852
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.046001346199648e-07,
+      "loss": 0.153,
+      "step": 8853
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.019034490060305e-07,
+      "loss": 0.1923,
+      "step": 8854
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.992127544096171e-07,
+      "loss": 0.2529,
+      "step": 8855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.965280511570637e-07,
+      "loss": 0.1588,
+      "step": 8856
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.938493395739598e-07,
+      "loss": 0.233,
+      "step": 8857
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.911766199851843e-07,
+      "loss": 0.1422,
+      "step": 8858
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.885098927148947e-07,
+      "loss": 0.1571,
+      "step": 8859
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.858491580864989e-07,
+      "loss": 0.2121,
+      "step": 8860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.831944164226999e-07,
+      "loss": 0.1663,
+      "step": 8861
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.805456680454624e-07,
+      "loss": 0.1586,
+      "step": 8862
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.779029132760294e-07,
+      "loss": 0.2024,
+      "step": 8863
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.752661524349167e-07,
+      "loss": 0.2283,
+      "step": 8864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.726353858419076e-07,
+      "loss": 0.1661,
+      "step": 8865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.700106138160688e-07,
+      "loss": 0.1386,
+      "step": 8866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.673918366757181e-07,
+      "loss": 0.205,
+      "step": 8867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.647790547384846e-07,
+      "loss": 0.1687,
+      "step": 8868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.62172268321226e-07,
+      "loss": 0.2228,
+      "step": 8869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.595714777401062e-07,
+      "loss": 0.1801,
+      "step": 8870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.56976683310545e-07,
+      "loss": 0.1476,
+      "step": 8871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.543878853472462e-07,
+      "loss": 0.1907,
+      "step": 8872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.518050841641698e-07,
+      "loss": 0.2118,
+      "step": 8873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.492282800745707e-07,
+      "loss": 0.1706,
+      "step": 8874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.466574733909546e-07,
+      "loss": 0.1871,
+      "step": 8875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.440926644251221e-07,
+      "loss": 0.1096,
+      "step": 8876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.415338534881131e-07,
+      "loss": 0.1416,
+      "step": 8877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.389810408902796e-07,
+      "loss": 0.1804,
+      "step": 8878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.364342269412237e-07,
+      "loss": 0.1635,
+      "step": 8879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.338934119498207e-07,
+      "loss": 0.1875,
+      "step": 8880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.31358596224224e-07,
+      "loss": 0.0933,
+      "step": 8881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.288297800718545e-07,
+      "loss": 0.1161,
+      "step": 8882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.263069637994112e-07,
+      "loss": 0.1749,
+      "step": 8883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.237901477128548e-07,
+      "loss": 0.1427,
+      "step": 8884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.212793321174358e-07,
+      "loss": 0.1499,
+      "step": 8885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.18774517317655e-07,
+      "loss": 0.1369,
+      "step": 8886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.162757036173083e-07,
+      "loss": 0.2154,
+      "step": 8887
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.137828913194476e-07,
+      "loss": 0.1301,
+      "step": 8888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.112960807263978e-07,
+      "loss": 0.1574,
+      "step": 8889
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.088152721397677e-07,
+      "loss": 0.1976,
+      "step": 8890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.063404658604332e-07,
+      "loss": 0.1609,
+      "step": 8891
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.038716621885264e-07,
+      "loss": 0.1211,
+      "step": 8892
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.014088614234747e-07,
+      "loss": 0.1533,
+      "step": 8893
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.989520638639722e-07,
+      "loss": 0.2052,
+      "step": 8894
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.965012698079641e-07,
+      "loss": 0.1506,
+      "step": 8895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.940564795526958e-07,
+      "loss": 0.1716,
+      "step": 8896
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.916176933946693e-07,
+      "loss": 0.136,
+      "step": 8897
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.891849116296532e-07,
+      "loss": 0.2025,
+      "step": 8898
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.867581345527117e-07,
+      "loss": 0.1855,
+      "step": 8899
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.843373624581593e-07,
+      "loss": 0.169,
+      "step": 8900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.819225956395835e-07,
+      "loss": 0.1524,
+      "step": 8901
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.795138343898442e-07,
+      "loss": 0.1492,
+      "step": 8902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.771110790010913e-07,
+      "loss": 0.1518,
+      "step": 8903
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.747143297647194e-07,
+      "loss": 0.269,
+      "step": 8904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7232358697141266e-07,
+      "loss": 0.1703,
+      "step": 8905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.699388509111224e-07,
+      "loss": 0.1264,
+      "step": 8906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6756012187306185e-07,
+      "loss": 0.1533,
+      "step": 8907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.651874001457335e-07,
+      "loss": 0.168,
+      "step": 8908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.62820686016896e-07,
+      "loss": 0.1371,
+      "step": 8909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6045997977358647e-07,
+      "loss": 0.193,
+      "step": 8910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5810528170211477e-07,
+      "loss": 0.2694,
+      "step": 8911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.55756592088058e-07,
+      "loss": 0.1671,
+      "step": 8912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.534139112162605e-07,
+      "loss": 0.1344,
+      "step": 8913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.510772393708562e-07,
+      "loss": 0.183,
+      "step": 8914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4874657683522945e-07,
+      "loss": 0.233,
+      "step": 8915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4642192389203754e-07,
+      "loss": 0.222,
+      "step": 8916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4410328082322703e-07,
+      "loss": 0.1352,
+      "step": 8917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4179064790999535e-07,
+      "loss": 0.1382,
+      "step": 8918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.394840254328236e-07,
+      "loss": 0.1142,
+      "step": 8919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3718341367145476e-07,
+      "loss": 0.1873,
+      "step": 8920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3488881290491556e-07,
+      "loss": 0.1675,
+      "step": 8921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3260022341149455e-07,
+      "loss": 0.2034,
+      "step": 8922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.303176454687419e-07,
+      "loss": 0.2722,
+      "step": 8923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.280410793535028e-07,
+      "loss": 0.2144,
+      "step": 8924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.257705253418787e-07,
+      "loss": 0.1713,
+      "step": 8925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2350598370923254e-07,
+      "loss": 0.163,
+      "step": 8926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2124745473021696e-07,
+      "loss": 0.1752,
+      "step": 8927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.189949386787462e-07,
+      "loss": 0.1779,
+      "step": 8928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.167484358280016e-07,
+      "loss": 0.1346,
+      "step": 8929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1450794645044866e-07,
+      "loss": 0.1614,
+      "step": 8930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.122734708178089e-07,
+      "loss": 0.1815,
+      "step": 8931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1004500920107657e-07,
+      "loss": 0.1803,
+      "step": 8932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0782256187053005e-07,
+      "loss": 0.1962,
+      "step": 8933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.056061290956981e-07,
+      "loss": 0.1483,
+      "step": 8934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0339571114539895e-07,
+      "loss": 0.1997,
+      "step": 8935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.011913082877072e-07,
+      "loss": 0.153,
+      "step": 8936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9899292078997565e-07,
+      "loss": 0.1676,
+      "step": 8937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.968005489188242e-07,
+      "loss": 0.1591,
+      "step": 8938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.946141929401459e-07,
+      "loss": 0.1272,
+      "step": 8939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.924338531191063e-07,
+      "loss": 0.1762,
+      "step": 8940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.902595297201328e-07,
+      "loss": 0.1885,
+      "step": 8941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8809122300692557e-07,
+      "loss": 0.1911,
+      "step": 8942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8592893324246314e-07,
+      "loss": 0.18,
+      "step": 8943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8377266068898577e-07,
+      "loss": 0.1511,
+      "step": 8944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.81622405608012e-07,
+      "loss": 0.1571,
+      "step": 8945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7947816826031655e-07,
+      "loss": 0.1446,
+      "step": 8946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7733994890595816e-07,
+      "loss": 0.2508,
+      "step": 8947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.752077478042626e-07,
+      "loss": 0.2242,
+      "step": 8948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.730815652138231e-07,
+      "loss": 0.2245,
+      "step": 8949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.709614013925e-07,
+      "loss": 0.1481,
+      "step": 8950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6884725659743216e-07,
+      "loss": 0.1829,
+      "step": 8951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.667391310850199e-07,
+      "loss": 0.1512,
+      "step": 8952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6463702511093656e-07,
+      "loss": 0.152,
+      "step": 8953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6254093893013374e-07,
+      "loss": 0.1978,
+      "step": 8954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.604508727968192e-07,
+      "loss": 0.1554,
+      "step": 8955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5836682696447356e-07,
+      "loss": 0.1692,
+      "step": 8956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.562888016858612e-07,
+      "loss": 0.1392,
+      "step": 8957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5421679721299727e-07,
+      "loss": 0.1503,
+      "step": 8958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5215081379718074e-07,
+      "loss": 0.1889,
+      "step": 8959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5009085168897226e-07,
+      "loss": 0.1961,
+      "step": 8960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.480369111381998e-07,
+      "loss": 0.1714,
+      "step": 8961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.459889923939752e-07,
+      "loss": 0.2053,
+      "step": 8962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.439470957046609e-07,
+      "loss": 0.1525,
+      "step": 8963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.419112213179032e-07,
+      "loss": 0.1415,
+      "step": 8964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398813694806158e-07,
+      "loss": 0.2287,
+      "step": 8965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.37857540438985e-07,
+      "loss": 0.2339,
+      "step": 8966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.358397344384479e-07,
+      "loss": 0.1792,
+      "step": 8967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3382795172373084e-07,
+      "loss": 0.1112,
+      "step": 8968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3182219253882765e-07,
+      "loss": 0.1202,
+      "step": 8969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.298224571269881e-07,
+      "loss": 0.0715,
+      "step": 8970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.278287457307516e-07,
+      "loss": 0.1855,
+      "step": 8971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2584105859190804e-07,
+      "loss": 0.1878,
+      "step": 8972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2385939595153126e-07,
+      "loss": 0.1414,
+      "step": 8973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2188375804995117e-07,
+      "loss": 0.1302,
+      "step": 8974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.199141451267762e-07,
+      "loss": 0.1693,
+      "step": 8975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.179505574208874e-07,
+      "loss": 0.1858,
+      "step": 8976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1599299517042215e-07,
+      "loss": 0.1948,
+      "step": 8977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.14041458612796e-07,
+      "loss": 0.194,
+      "step": 8978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.120959479846919e-07,
+      "loss": 0.1377,
+      "step": 8979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1015646352206553e-07,
+      "loss": 0.1904,
+      "step": 8980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.082230054601343e-07,
+      "loss": 0.1888,
+      "step": 8981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0629557403338837e-07,
+      "loss": 0.1819,
+      "step": 8982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0437416947559084e-07,
+      "loss": 0.1856,
+      "step": 8983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0245879201976634e-07,
+      "loss": 0.1774,
+      "step": 8984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0054944189821244e-07,
+      "loss": 0.1844,
+      "step": 8985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9864611934250497e-07,
+      "loss": 0.1446,
+      "step": 8986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9674882458346486e-07,
+      "loss": 0.2053,
+      "step": 8987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.948575578512025e-07,
+      "loss": 0.2069,
+      "step": 8988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.929723193751011e-07,
+      "loss": 0.1308,
+      "step": 8989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9109310938378877e-07,
+      "loss": 0.1275,
+      "step": 8990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.892199281051833e-07,
+      "loss": 0.1987,
+      "step": 8991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.87352775766464e-07,
+      "loss": 0.2098,
+      "step": 8992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8549165259407763e-07,
+      "loss": 0.1326,
+      "step": 8993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.836365588137491e-07,
+      "loss": 0.1969,
+      "step": 8994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8178749465045396e-07,
+      "loss": 0.1351,
+      "step": 8995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7994446032845734e-07,
+      "loss": 0.2383,
+      "step": 8996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.781074560712749e-07,
+      "loss": 0.1194,
+      "step": 8997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.762764821017061e-07,
+      "loss": 0.1658,
+      "step": 8998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.744515386418067e-07,
+      "loss": 0.1371,
+      "step": 8999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.726326259129164e-07,
+      "loss": 0.1447,
+      "step": 9000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7081974413561417e-07,
+      "loss": 0.1957,
+      "step": 9001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.690128935297853e-07,
+      "loss": 0.1734,
+      "step": 9002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6721207431456006e-07,
+      "loss": 0.1699,
+      "step": 9003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6541728670833597e-07,
+      "loss": 0.1437,
+      "step": 9004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6362853092878894e-07,
+      "loss": 0.1324,
+      "step": 9005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6184580719286756e-07,
+      "loss": 0.1677,
+      "step": 9006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.600691157167712e-07,
+      "loss": 0.315,
+      "step": 9007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.582984567159774e-07,
+      "loss": 0.1973,
+      "step": 9008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5653383040524227e-07,
+      "loss": 0.1946,
+      "step": 9009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.547752369985723e-07,
+      "loss": 0.2247,
+      "step": 9010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.530226767092525e-07,
+      "loss": 0.1789,
+      "step": 9011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5127614974983503e-07,
+      "loss": 0.154,
+      "step": 9012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.495356563321338e-07,
+      "loss": 0.1847,
+      "step": 9013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4780119666724113e-07,
+      "loss": 0.1855,
+      "step": 9014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.460727709655164e-07,
+      "loss": 0.1246,
+      "step": 9015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.443503794365753e-07,
+      "loss": 0.2021,
+      "step": 9016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.426340222893175e-07,
+      "loss": 0.1466,
+      "step": 9017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.409236997318931e-07,
+      "loss": 0.2352,
+      "step": 9018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.392194119717417e-07,
+      "loss": 0.1897,
+      "step": 9019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3752115921555905e-07,
+      "loss": 0.1564,
+      "step": 9020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3582894166930268e-07,
+      "loss": 0.1535,
+      "step": 9021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3414275953820842e-07,
+      "loss": 0.1627,
+      "step": 9022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.324626130267793e-07,
+      "loss": 0.1709,
+      "step": 9023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3078850233878013e-07,
+      "loss": 0.31,
+      "step": 9024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2912042767724852e-07,
+      "loss": 0.2069,
+      "step": 9025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.274583892444948e-07,
+      "loss": 0.2235,
+      "step": 9026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.258023872420856e-07,
+      "loss": 0.2493,
+      "step": 9027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.241524218708657e-07,
+      "loss": 0.1588,
+      "step": 9028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2250849333093626e-07,
+      "loss": 0.2322,
+      "step": 9029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2087060182168218e-07,
+      "loss": 0.1772,
+      "step": 9030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.19238747541739e-07,
+      "loss": 0.2195,
+      "step": 9031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1761293068902622e-07,
+      "loss": 0.1804,
+      "step": 9032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1599315146071942e-07,
+      "loss": 0.2121,
+      "step": 9033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1437941005326145e-07,
+      "loss": 0.1842,
+      "step": 9034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1277170666237912e-07,
+      "loss": 0.2053,
+      "step": 9035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.111700414830442e-07,
+      "loss": 0.1361,
+      "step": 9036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.09574414709518e-07,
+      "loss": 0.2312,
+      "step": 9037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0798482653530682e-07,
+      "loss": 0.1234,
+      "step": 9038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.064012771532009e-07,
+      "loss": 0.1876,
+      "step": 9039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0482376675525773e-07,
+      "loss": 0.2036,
+      "step": 9040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0325229553279647e-07,
+      "loss": 0.2826,
+      "step": 9041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.01686863676398e-07,
+      "loss": 0.1393,
+      "step": 9042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.001274713759327e-07,
+      "loss": 0.1779,
+      "step": 9043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9857411882051035e-07,
+      "loss": 0.1115,
+      "step": 9044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.970268061985303e-07,
+      "loss": 0.1661,
+      "step": 9045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.95485533697648e-07,
+      "loss": 0.2014,
+      "step": 9046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9395030150479166e-07,
+      "loss": 0.1689,
+      "step": 9047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9242110980615124e-07,
+      "loss": 0.2326,
+      "step": 9048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9089795878718951e-07,
+      "loss": 0.1756,
+      "step": 9049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.893808486326365e-07,
+      "loss": 0.1977,
+      "step": 9050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8786977952648944e-07,
+      "loss": 0.174,
+      "step": 9051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8636475165200174e-07,
+      "loss": 0.1247,
+      "step": 9052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8486576519171628e-07,
+      "loss": 0.1318,
+      "step": 9053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8337282032742098e-07,
+      "loss": 0.18,
+      "step": 9054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.818859172401821e-07,
+      "loss": 0.1102,
+      "step": 9055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8040505611033874e-07,
+      "loss": 0.163,
+      "step": 9056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7893023711748613e-07,
+      "loss": 0.1859,
+      "step": 9057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7746146044048117e-07,
+      "loss": 0.2366,
+      "step": 9058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7599872625747583e-07,
+      "loss": 0.1485,
+      "step": 9059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7454203474585597e-07,
+      "loss": 0.1644,
+      "step": 9060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7309138608229692e-07,
+      "loss": 0.1677,
+      "step": 9061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.716467804427302e-07,
+      "loss": 0.1567,
+      "step": 9062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7020821800236564e-07,
+      "loss": 0.1798,
+      "step": 9063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.687756989356637e-07,
+      "loss": 0.2106,
+      "step": 9064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6734922341636316e-07,
+      "loss": 0.2188,
+      "step": 9065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.659287916174701e-07,
+      "loss": 0.1922,
+      "step": 9066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6451440371125226e-07,
+      "loss": 0.2257,
+      "step": 9067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6310605986924465e-07,
+      "loss": 0.1445,
+      "step": 9068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6170376026226065e-07,
+      "loss": 0.1868,
+      "step": 9069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6030750506036419e-07,
+      "loss": 0.1852,
+      "step": 9070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5891729443289204e-07,
+      "loss": 0.1689,
+      "step": 9071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5753312854845937e-07,
+      "loss": 0.1837,
+      "step": 9072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.561550075749263e-07,
+      "loss": 0.1393,
+      "step": 9073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5478293167943692e-07,
+      "loss": 0.1853,
+      "step": 9074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.53416901028397e-07,
+      "loss": 0.1695,
+      "step": 9075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.520569157874796e-07,
+      "loss": 0.1196,
+      "step": 9076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5070297612162497e-07,
+      "loss": 0.1678,
+      "step": 9077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4935508219504067e-07,
+      "loss": 0.1515,
+      "step": 9078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4801323417119594e-07,
+      "loss": 0.1896,
+      "step": 9079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4667743221283282e-07,
+      "loss": 0.2045,
+      "step": 9080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4534767648195503e-07,
+      "loss": 0.1647,
+      "step": 9081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.440239671398391e-07,
+      "loss": 0.1482,
+      "step": 9082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.427063043470178e-07,
+      "loss": 0.1922,
+      "step": 9083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4139468826331325e-07,
+      "loss": 0.1327,
+      "step": 9084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4008911904778154e-07,
+      "loss": 0.1945,
+      "step": 9085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.387895968587738e-07,
+      "loss": 0.1624,
+      "step": 9086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.374961218538917e-07,
+      "loss": 0.1292,
+      "step": 9087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3620869419000982e-07,
+      "loss": 0.1892,
+      "step": 9088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3492731402326432e-07,
+      "loss": 0.1558,
+      "step": 9089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3365198150906976e-07,
+      "loss": 0.1269,
+      "step": 9090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3238269680209137e-07,
+      "loss": 0.1917,
+      "step": 9091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3111946005626707e-07,
+      "loss": 0.1678,
+      "step": 9092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2986227142480767e-07,
+      "loss": 0.196,
+      "step": 9093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2861113106018562e-07,
+      "loss": 0.1885,
+      "step": 9094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.273660391141407e-07,
+      "loss": 0.1633,
+      "step": 9095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2612699573766872e-07,
+      "loss": 0.1485,
+      "step": 9096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2489400108105508e-07,
+      "loss": 0.199,
+      "step": 9097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2366705529382462e-07,
+      "loss": 0.2595,
+      "step": 9098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2244615852479157e-07,
+      "loss": 0.1743,
+      "step": 9099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.212313109220209e-07,
+      "loss": 0.128,
+      "step": 9100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2002251263285026e-07,
+      "loss": 0.1386,
+      "step": 9101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.188197638038846e-07,
+      "loss": 0.1814,
+      "step": 9102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1762306458099614e-07,
+      "loss": 0.1944,
+      "step": 9103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1643241510931324e-07,
+      "loss": 0.213,
+      "step": 9104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1524781553324815e-07,
+      "loss": 0.185,
+      "step": 9105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1406926599646373e-07,
+      "loss": 0.1771,
+      "step": 9106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1289676664189563e-07,
+      "loss": 0.1281,
+      "step": 9107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1173031761174679e-07,
+      "loss": 0.2226,
+      "step": 9108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1056991904748182e-07,
+      "loss": 0.2045,
+      "step": 9109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0941557108983258e-07,
+      "loss": 0.1527,
+      "step": 9110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0826727387880376e-07,
+      "loss": 0.1436,
+      "step": 9111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0712502755366172e-07,
+      "loss": 0.1985,
+      "step": 9112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0598883225293454e-07,
+      "loss": 0.1695,
+      "step": 9113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0485868811441757e-07,
+      "loss": 0.1654,
+      "step": 9114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0373459527518448e-07,
+      "loss": 0.1008,
+      "step": 9115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.02616553871554e-07,
+      "loss": 0.233,
+      "step": 9116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0150456403913433e-07,
+      "loss": 0.1952,
+      "step": 9117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.003986259127787e-07,
+      "loss": 0.125,
+      "step": 9118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.929873962661874e-08,
+      "loss": 0.1665,
+      "step": 9119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.820490531405324e-08,
+      "loss": 0.2136,
+      "step": 9120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.711712310773724e-08,
+      "loss": 0.1897,
+      "step": 9121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.603539313960407e-08,
+      "loss": 0.1751,
+      "step": 9122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.495971554083216e-08,
+      "loss": 0.1941,
+      "step": 9123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.389009044189489e-08,
+      "loss": 0.203,
+      "step": 9124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.282651797251074e-08,
+      "loss": 0.0823,
+      "step": 9125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.176899826167651e-08,
+      "loss": 0.1815,
+      "step": 9126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.071753143763406e-08,
+      "loss": 0.2823,
+      "step": 9127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.967211762792027e-08,
+      "loss": 0.182,
+      "step": 9128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.86327569593115e-08,
+      "loss": 0.2525,
+      "step": 9129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.759944955787358e-08,
+      "loss": 0.1463,
+      "step": 9130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.657219554891738e-08,
+      "loss": 0.1396,
+      "step": 9131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.555099505703212e-08,
+      "loss": 0.2616,
+      "step": 9132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.45358482060632e-08,
+      "loss": 0.1221,
+      "step": 9133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.352675511913987e-08,
+      "loss": 0.1678,
+      "step": 9134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.25237159186365e-08,
+      "loss": 0.2269,
+      "step": 9135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.152673072620576e-08,
+      "loss": 0.2091,
+      "step": 9136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.053579966276204e-08,
+      "loss": 0.1965,
+      "step": 9137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.955092284848143e-08,
+      "loss": 0.1762,
+      "step": 9138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.857210040281837e-08,
+      "loss": 0.1997,
+      "step": 9139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.759933244447238e-08,
+      "loss": 0.1862,
+      "step": 9140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.663261909143792e-08,
+      "loss": 0.1104,
+      "step": 9141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.567196046094349e-08,
+      "loss": 0.144,
+      "step": 9142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.471735666950142e-08,
+      "loss": 0.1221,
+      "step": 9143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.37688078328913e-08,
+      "loss": 0.1768,
+      "step": 9144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.282631406615447e-08,
+      "loss": 0.1253,
+      "step": 9145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.188987548358838e-08,
+      "loss": 0.1386,
+      "step": 9146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.095949219876885e-08,
+      "loss": 0.1224,
+      "step": 9147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.003516432453339e-08,
+      "loss": 0.2148,
+      "step": 9148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.911689197298676e-08,
+      "loss": 0.1693,
+      "step": 9149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.820467525549545e-08,
+      "loss": 0.1753,
+      "step": 9150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.729851428269318e-08,
+      "loss": 0.1562,
+      "step": 9151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.639840916448092e-08,
+      "loss": 0.1867,
+      "step": 9152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.550436001002136e-08,
+      "loss": 0.2143,
+      "step": 9153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.461636692774442e-08,
+      "loss": 0.1855,
+      "step": 9154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.373443002535284e-08,
+      "loss": 0.2022,
+      "step": 9155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.285854940979997e-08,
+      "loss": 0.153,
+      "step": 9156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.198872518731191e-08,
+      "loss": 0.1715,
+      "step": 9157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.11249574633932e-08,
+      "loss": 0.1781,
+      "step": 9158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.026724634279335e-08,
+      "loss": 0.1626,
+      "step": 9159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.941559192952917e-08,
+      "loss": 0.1369,
+      "step": 9160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8569994326906906e-08,
+      "loss": 0.1892,
+      "step": 9161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.773045363746676e-08,
+      "loss": 0.1667,
+      "step": 9162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.689696996303284e-08,
+      "loss": 0.1898,
+      "step": 9163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.606954340469095e-08,
+      "loss": 0.1672,
+      "step": 9164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.524817406279415e-08,
+      "loss": 0.2007,
+      "step": 9165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.443286203695164e-08,
+      "loss": 0.191,
+      "step": 9166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.362360742605099e-08,
+      "loss": 0.1504,
+      "step": 9167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2820410328235925e-08,
+      "loss": 0.1952,
+      "step": 9168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2023270840917405e-08,
+      "loss": 0.1972,
+      "step": 9169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.123218906077365e-08,
+      "loss": 0.1465,
+      "step": 9170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0447165083750136e-08,
+      "loss": 0.1147,
+      "step": 9171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.966819900505404e-08,
+      "loss": 0.1305,
+      "step": 9172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.889529091915424e-08,
+      "loss": 0.1171,
+      "step": 9173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.81284409197924e-08,
+      "loss": 0.2175,
+      "step": 9174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.736764909997193e-08,
+      "loss": 0.1847,
+      "step": 9175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.661291555196345e-08,
+      "loss": 0.1385,
+      "step": 9176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5864240367299304e-08,
+      "loss": 0.1061,
+      "step": 9177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5121623636779075e-08,
+      "loss": 0.1696,
+      "step": 9178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.438506545046961e-08,
+      "loss": 0.1991,
+      "step": 9179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3654565897699454e-08,
+      "loss": 0.2063,
+      "step": 9180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2930125067064394e-08,
+      "loss": 0.1284,
+      "step": 9181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.22117430464275e-08,
+      "loss": 0.2058,
+      "step": 9182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.149941992290795e-08,
+      "loss": 0.1661,
+      "step": 9183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0793155782897776e-08,
+      "loss": 0.1302,
+      "step": 9184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0092950712056207e-08,
+      "loss": 0.1406,
+      "step": 9185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.939880479530422e-08,
+      "loss": 0.1297,
+      "step": 9186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.871071811683003e-08,
+      "loss": 0.1777,
+      "step": 9187
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.802869076007798e-08,
+      "loss": 0.1575,
+      "step": 9188
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.73527228077708e-08,
+      "loss": 0.1073,
+      "step": 9189
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6682814341887364e-08,
+      "loss": 0.1835,
+      "step": 9190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.601896544367378e-08,
+      "loss": 0.1927,
+      "step": 9191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.536117619364343e-08,
+      "loss": 0.1264,
+      "step": 9192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.470944667157694e-08,
+      "loss": 0.1734,
+      "step": 9193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.406377695650553e-08,
+      "loss": 0.1928,
+      "step": 9194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.34241671267499e-08,
+      "loss": 0.1533,
+      "step": 9195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2790617259870204e-08,
+      "loss": 0.2083,
+      "step": 9196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.216312743271055e-08,
+      "loss": 0.1593,
+      "step": 9197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.154169772137117e-08,
+      "loss": 0.1792,
+      "step": 9198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0926328201213994e-08,
+      "loss": 0.1951,
+      "step": 9199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0317018946884876e-08,
+      "loss": 0.1561,
+      "step": 9200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9713770032269163e-08,
+      "loss": 0.187,
+      "step": 9201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.911658153053054e-08,
+      "loss": 0.2155,
+      "step": 9202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8525453514099965e-08,
+      "loss": 0.1219,
+      "step": 9203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7940386054664537e-08,
+      "loss": 0.2215,
+      "step": 9204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.736137922318971e-08,
+      "loss": 0.1661,
+      "step": 9205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.678843308989154e-08,
+      "loss": 0.1419,
+      "step": 9206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6221547724253337e-08,
+      "loss": 0.1849,
+      "step": 9207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5660723195036763e-08,
+      "loss": 0.1754,
+      "step": 9208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.510595957025408e-08,
+      "loss": 0.1548,
+      "step": 9209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.45572569171848e-08,
+      "loss": 0.178,
+      "step": 9210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.401461530238125e-08,
+      "loss": 0.1882,
+      "step": 9211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3478034791651893e-08,
+      "loss": 0.2112,
+      "step": 9212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2947515450072454e-08,
+      "loss": 0.1987,
+      "step": 9213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.242305734198591e-08,
+      "loss": 0.1251,
+      "step": 9214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1904660530996935e-08,
+      "loss": 0.197,
+      "step": 9215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.139232507998301e-08,
+      "loss": 0.2559,
+      "step": 9216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.088605105107222e-08,
+      "loss": 0.1595,
+      "step": 9217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0385838505665444e-08,
+      "loss": 0.2844,
+      "step": 9218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9891687504436373e-08,
+      "loss": 0.1433,
+      "step": 9219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9403598107314846e-08,
+      "loss": 0.2017,
+      "step": 9220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8921570373492403e-08,
+      "loss": 0.1508,
+      "step": 9221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8445604361427838e-08,
+      "loss": 0.1203,
+      "step": 9222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7975700128847196e-08,
+      "loss": 0.1377,
+      "step": 9223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7511857732743773e-08,
+      "loss": 0.2033,
+      "step": 9224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7054077229372578e-08,
+      "loss": 0.1792,
+      "step": 9225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.660235867425586e-08,
+      "loss": 0.152,
+      "step": 9226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6156702122172017e-08,
+      "loss": 0.117,
+      "step": 9227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5717107627172267e-08,
+      "loss": 0.2147,
+      "step": 9228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5283575242569513e-08,
+      "loss": 0.1852,
+      "step": 9229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4856105020949474e-08,
+      "loss": 0.1369,
+      "step": 9230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4434697014148457e-08,
+      "loss": 0.156,
+      "step": 9231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4019351273275582e-08,
+      "loss": 0.2293,
+      "step": 9232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.361006784870722e-08,
+      "loss": 0.2686,
+      "step": 9233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3206846790081439e-08,
+      "loss": 0.2097,
+      "step": 9234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2809688146298015e-08,
+      "loss": 0.2173,
+      "step": 9235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2418591965529525e-08,
+      "loss": 0.1565,
+      "step": 9236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2033558295199143e-08,
+      "loss": 0.1668,
+      "step": 9237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1654587182013953e-08,
+      "loss": 0.144,
+      "step": 9238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1281678671931639e-08,
+      "loss": 0.188,
+      "step": 9239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0914832810177134e-08,
+      "loss": 0.1645,
+      "step": 9240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0554049641242625e-08,
+      "loss": 0.1844,
+      "step": 9241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0199329208887554e-08,
+      "loss": 0.1556,
+      "step": 9242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.850671556127511e-09,
+      "loss": 0.17,
+      "step": 9243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.508076725250892e-09,
+      "loss": 0.2458,
+      "step": 9244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.171544757802241e-09,
+      "loss": 0.1836,
+      "step": 9245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.841075694604462e-09,
+      "loss": 0.1807,
+      "step": 9246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.516669575736602e-09,
+      "loss": 0.1802,
+      "step": 9247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.198326440533865e-09,
+      "loss": 0.1525,
+      "step": 9248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.886046327609808e-09,
+      "loss": 0.1541,
+      "step": 9249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.57982927483969e-09,
+      "loss": 0.1849,
+      "step": 9250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.279675319360469e-09,
+      "loss": 0.2607,
+      "step": 9251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.985584497576359e-09,
+      "loss": 0.2331,
+      "step": 9252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.697556845147723e-09,
+      "loss": 0.265,
+      "step": 9253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.415592397018833e-09,
+      "loss": 0.171,
+      "step": 9254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.1396911873734535e-09,
+      "loss": 0.1617,
+      "step": 9255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.869853249684809e-09,
+      "loss": 0.1745,
+      "step": 9256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.6060786166711735e-09,
+      "loss": 0.1388,
+      "step": 9257
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3483673203236215e-09,
+      "loss": 0.1792,
+      "step": 9258
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.096719391900484e-09,
+      "loss": 0.1046,
+      "step": 9259
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.851134861921791e-09,
+      "loss": 0.1113,
+      "step": 9260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.611613760169275e-09,
+      "loss": 0.2304,
+      "step": 9261
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.378156115697474e-09,
+      "loss": 0.1802,
+      "step": 9262
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.150761956811522e-09,
+      "loss": 0.2284,
+      "step": 9263
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.929431311094911e-09,
+      "loss": 0.1324,
+      "step": 9264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.714164205392834e-09,
+      "loss": 0.1949,
+      "step": 9265
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.504960665812185e-09,
+      "loss": 0.1514,
+      "step": 9266
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.30182071772156e-09,
+      "loss": 0.1548,
+      "step": 9267
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.104744385756808e-09,
+      "loss": 0.1353,
+      "step": 9268
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9137316938265825e-09,
+      "loss": 0.2226,
+      "step": 9269
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.728782665095686e-09,
+      "loss": 0.2962,
+      "step": 9270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.549897321985073e-09,
+      "loss": 0.1666,
+      "step": 9271
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3770756861996037e-09,
+      "loss": 0.1299,
+      "step": 9272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2103177787002882e-09,
+      "loss": 0.2402,
+      "step": 9273
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0496236197042885e-09,
+      "loss": 0.2274,
+      "step": 9274
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8949932287015692e-09,
+      "loss": 0.1777,
+      "step": 9275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7464266244548999e-09,
+      "loss": 0.2375,
+      "step": 9276
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6039238249720977e-09,
+      "loss": 0.1127,
+      "step": 9277
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4674848475393355e-09,
+      "loss": 0.1578,
+      "step": 9278
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.337109708704487e-09,
+      "loss": 0.1569,
+      "step": 9279
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2127984242826796e-09,
+      "loss": 0.1529,
+      "step": 9280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0945510093451904e-09,
+      "loss": 0.1499,
+      "step": 9281
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.823674782305504e-10,
+      "loss": 0.2206,
+      "step": 9282
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.762478445556443e-10,
+      "loss": 0.1411,
+      "step": 9283
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.761921211824063e-10,
+      "loss": 0.1567,
+      "step": 9284
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.82200320245574e-10,
+      "loss": 0.1575,
+      "step": 9285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.942724531471378e-10,
+      "loss": 0.2392,
+      "step": 9286
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.124085305507897e-10,
+      "loss": 0.1959,
+      "step": 9287
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.366085623819238e-10,
+      "loss": 0.1276,
+      "step": 9288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6687255783873777e-10,
+      "loss": 0.1729,
+      "step": 9289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0320052537557987e-10,
+      "loss": 0.1421,
+      "step": 9290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4559247271405126e-10,
+      "loss": 0.194,
+      "step": 9291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9404840684300596e-10,
+      "loss": 0.2171,
+      "step": 9292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4856833401299952e-10,
+      "loss": 0.2116,
+      "step": 9293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0915225973628928e-10,
+      "loss": 0.2106,
+      "step": 9294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.580018879793649e-11,
+      "loss": 0.2169,
+      "step": 9295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8512125244704053e-11,
+      "loss": 0.189,
+      "step": 9296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.728807237950548e-11,
+      "loss": 0.2569,
+      "step": 9297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.21280327836093e-11,
+      "loss": 0.1585,
+      "step": 9298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.032008288883503e-12,
+      "loss": 0.1769,
+      "step": 9299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.1976,
+      "step": 9300
+    },
+    {
+      "epoch": 1.0,
+      "step": 9300,
+      "total_flos": 8.206174795797096e+22,
+      "train_loss": 0.3191605252492171,
+      "train_runtime": 117088.0105,
+      "train_samples_per_second": 0.318,
+      "train_steps_per_second": 0.079
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 9300,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 4650,
+  "total_flos": 8.206174795797096e+22,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}