diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,20059 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999251328891218,
+  "global_step": 3339,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9801980198019803e-07,
+      "loss": 2.6954,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9603960396039606e-07,
+      "loss": 3.1209,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.940594059405941e-07,
+      "loss": 2.9452,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.920792079207921e-07,
+      "loss": 3.3805,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.900990099009902e-07,
+      "loss": 2.2747,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1881188118811881e-06,
+      "loss": 2.7651,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3861386138613863e-06,
+      "loss": 2.7982,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5841584158415842e-06,
+      "loss": 2.6226,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7821782178217822e-06,
+      "loss": 2.64,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9801980198019803e-06,
+      "loss": 2.8978,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1782178217821785e-06,
+      "loss": 1.9478,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3762376237623762e-06,
+      "loss": 1.8533,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5742574257425744e-06,
+      "loss": 1.885,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7722772277227726e-06,
+      "loss": 2.015,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9702970297029703e-06,
+      "loss": 1.4504,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1683168316831685e-06,
+      "loss": 1.2375,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3663366336633666e-06,
+      "loss": 1.1839,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5643564356435644e-06,
+      "loss": 1.2132,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7623762376237625e-06,
+      "loss": 1.0111,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.960396039603961e-06,
+      "loss": 1.0197,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.158415841584159e-06,
+      "loss": 0.8044,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.356435643564357e-06,
+      "loss": 0.8853,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.554455445544555e-06,
+      "loss": 0.7622,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7524752475247525e-06,
+      "loss": 0.905,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.950495049504951e-06,
+      "loss": 0.983,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.148514851485149e-06,
+      "loss": 0.9614,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.346534653465347e-06,
+      "loss": 0.9154,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.544554455445545e-06,
+      "loss": 0.7946,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7425742574257425e-06,
+      "loss": 0.8096,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.940594059405941e-06,
+      "loss": 0.6829,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.138613861386139e-06,
+      "loss": 0.7524,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.336633663366337e-06,
+      "loss": 0.7673,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.534653465346535e-06,
+      "loss": 0.7025,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.732673267326733e-06,
+      "loss": 0.7116,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.930693069306931e-06,
+      "loss": 0.7366,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.128712871287129e-06,
+      "loss": 0.7717,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.326732673267327e-06,
+      "loss": 0.734,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.524752475247525e-06,
+      "loss": 0.7666,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.722772277227724e-06,
+      "loss": 0.85,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.920792079207921e-06,
+      "loss": 0.7113,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.11881188118812e-06,
+      "loss": 0.801,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.316831683168318e-06,
+      "loss": 0.7893,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.514851485148515e-06,
+      "loss": 0.6858,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.712871287128714e-06,
+      "loss": 0.6544,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.910891089108911e-06,
+      "loss": 0.7228,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.10891089108911e-06,
+      "loss": 0.7456,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.306930693069308e-06,
+      "loss": 0.6126,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.504950495049505e-06,
+      "loss": 0.7048,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.702970297029704e-06,
+      "loss": 0.6302,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.900990099009901e-06,
+      "loss": 0.7098,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.00990099009901e-05,
+      "loss": 0.6182,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0297029702970298e-05,
+      "loss": 0.7459,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0495049504950497e-05,
+      "loss": 0.6916,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0693069306930694e-05,
+      "loss": 0.6999,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0891089108910893e-05,
+      "loss": 0.6926,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.108910891089109e-05,
+      "loss": 0.5983,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1287128712871288e-05,
+      "loss": 0.7035,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1485148514851485e-05,
+      "loss": 0.5669,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1683168316831684e-05,
+      "loss": 0.6225,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1881188118811881e-05,
+      "loss": 0.6002,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.207920792079208e-05,
+      "loss": 0.5697,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2277227722772278e-05,
+      "loss": 0.6518,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2475247524752477e-05,
+      "loss": 0.738,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2673267326732674e-05,
+      "loss": 0.657,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2871287128712873e-05,
+      "loss": 0.6278,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.306930693069307e-05,
+      "loss": 0.7762,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.326732673267327e-05,
+      "loss": 0.7414,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3465346534653467e-05,
+      "loss": 0.6907,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3663366336633666e-05,
+      "loss": 0.448,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3861386138613861e-05,
+      "loss": 0.6363,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.405940594059406e-05,
+      "loss": 0.7554,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4257425742574257e-05,
+      "loss": 0.5781,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4455445544554456e-05,
+      "loss": 0.5574,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4653465346534654e-05,
+      "loss": 0.6171,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4851485148514853e-05,
+      "loss": 0.6589,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.504950495049505e-05,
+      "loss": 0.6051,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5247524752475249e-05,
+      "loss": 0.6637,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5445544554455448e-05,
+      "loss": 0.6486,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5643564356435644e-05,
+      "loss": 0.7367,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5841584158415843e-05,
+      "loss": 0.6415,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6039603960396042e-05,
+      "loss": 0.7304,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.623762376237624e-05,
+      "loss": 0.6668,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6435643564356436e-05,
+      "loss": 0.5821,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6633663366336635e-05,
+      "loss": 0.6202,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.683168316831683e-05,
+      "loss": 0.4944,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.702970297029703e-05,
+      "loss": 0.715,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.722772277227723e-05,
+      "loss": 0.7211,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7425742574257428e-05,
+      "loss": 0.572,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7623762376237624e-05,
+      "loss": 0.53,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7821782178217823e-05,
+      "loss": 0.5419,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8019801980198022e-05,
+      "loss": 0.6437,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.821782178217822e-05,
+      "loss": 0.7834,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.841584158415842e-05,
+      "loss": 0.6778,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8613861386138615e-05,
+      "loss": 0.5748,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.881188118811881e-05,
+      "loss": 0.5118,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.900990099009901e-05,
+      "loss": 0.7188,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.920792079207921e-05,
+      "loss": 0.5768,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9405940594059408e-05,
+      "loss": 0.5485,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9603960396039604e-05,
+      "loss": 0.726,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9801980198019803e-05,
+      "loss": 0.6541,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.6772,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999995293305322e-05,
+      "loss": 0.7654,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998117322571e-05,
+      "loss": 0.5524,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999957639774465e-05,
+      "loss": 0.6711,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999924692973738e-05,
+      "loss": 0.5769,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999882332854538e-05,
+      "loss": 0.6976,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999983055945674e-05,
+      "loss": 0.6251,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999769372829087e-05,
+      "loss": 0.5697,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999969877302917e-05,
+      "loss": 0.6742,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999618760123454e-05,
+      "loss": 0.6831,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999952933418725e-05,
+      "loss": 0.5632,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999430495304744e-05,
+      "loss": 0.7367,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999932224356898e-05,
+      "loss": 0.772,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999920457908185e-05,
+      "loss": 0.6203,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999907750195412e-05,
+      "loss": 0.6064,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998941012305418e-05,
+      "loss": 0.5311,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999879511026422e-05,
+      "loss": 0.6531,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998639795967874e-05,
+      "loss": 0.6927,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999847506956258e-05,
+      "loss": 0.6519,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998300931203402e-05,
+      "loss": 0.5964,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998117381054265e-05,
+      "loss": 0.5661,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999792441928795e-05,
+      "loss": 0.6025,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99977220460861e-05,
+      "loss": 0.5784,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997510261639218e-05,
+      "loss": 0.6536,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997289066146665e-05,
+      "loss": 0.6256,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999705845981666e-05,
+      "loss": 0.6695,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999681844286628e-05,
+      "loss": 0.7123,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996569015521468e-05,
+      "loss": 0.5594,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996310178017012e-05,
+      "loss": 0.567,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999604193059657e-05,
+      "loss": 0.6816,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995764273512652e-05,
+      "loss": 0.766,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995477207026632e-05,
+      "loss": 0.6083,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999518073140873e-05,
+      "loss": 0.6198,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999487484693804e-05,
+      "loss": 0.6413,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994559553902486e-05,
+      "loss": 0.5829,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994234852598884e-05,
+      "loss": 0.5704,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993900743332877e-05,
+      "loss": 0.6975,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999355722641898e-05,
+      "loss": 0.597,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993204302180557e-05,
+      "loss": 0.6635,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992841970949827e-05,
+      "loss": 0.5701,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992470233067875e-05,
+      "loss": 0.6312,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999208908888462e-05,
+      "loss": 0.7587,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991698538758855e-05,
+      "loss": 0.5587,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999129858305822e-05,
+      "loss": 0.5972,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990889222159207e-05,
+      "loss": 0.5872,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990470456447165e-05,
+      "loss": 0.5725,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990042286316294e-05,
+      "loss": 0.5374,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989604712169646e-05,
+      "loss": 0.5733,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998915773441913e-05,
+      "loss": 0.5412,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99887013534855e-05,
+      "loss": 0.6664,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988235569798364e-05,
+      "loss": 0.7423,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987760383796188e-05,
+      "loss": 0.7103,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998727579592628e-05,
+      "loss": 0.5473,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986781806644798e-05,
+      "loss": 0.6413,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998627841641676e-05,
+      "loss": 0.5082,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985765625716017e-05,
+      "loss": 0.6988,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998524343502529e-05,
+      "loss": 0.5884,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984711844836132e-05,
+      "loss": 0.7375,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998417085564895e-05,
+      "loss": 0.6932,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983620467973002e-05,
+      "loss": 0.615,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998306068232638e-05,
+      "loss": 0.6848,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998249149923604e-05,
+      "loss": 0.6069,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981912919237775e-05,
+      "loss": 0.5462,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998132494287622e-05,
+      "loss": 0.6868,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998072757070487e-05,
+      "loss": 0.5329,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980120803286044e-05,
+      "loss": 0.6595,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979504641190922e-05,
+      "loss": 0.6259,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978879084999518e-05,
+      "loss": 0.5556,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978244135300694e-05,
+      "loss": 0.5827,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977599792692156e-05,
+      "loss": 0.7282,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976946057780438e-05,
+      "loss": 0.6349,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976282931180942e-05,
+      "loss": 0.6636,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975610413517886e-05,
+      "loss": 0.5582,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974928505424333e-05,
+      "loss": 0.7258,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99742372075422e-05,
+      "loss": 0.6817,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973536520522224e-05,
+      "loss": 0.6334,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972826445023993e-05,
+      "loss": 0.6285,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972106981715927e-05,
+      "loss": 0.6882,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971378131275287e-05,
+      "loss": 0.7238,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970639894388162e-05,
+      "loss": 0.4588,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969892271749492e-05,
+      "loss": 0.7667,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969135264063037e-05,
+      "loss": 0.6412,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968368872041405e-05,
+      "loss": 0.6219,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967593096406023e-05,
+      "loss": 0.5398,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966807937887157e-05,
+      "loss": 0.7794,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966013397223918e-05,
+      "loss": 0.5782,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996520947516423e-05,
+      "loss": 0.5966,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964396172464857e-05,
+      "loss": 0.5677,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963573489891398e-05,
+      "loss": 0.6409,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962741428218265e-05,
+      "loss": 0.6746,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996189998822872e-05,
+      "loss": 0.6021,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996104917071484e-05,
+      "loss": 0.6159,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960188976477535e-05,
+      "loss": 0.5632,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959319406326535e-05,
+      "loss": 0.5771,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99584404610804e-05,
+      "loss": 0.5689,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995755214156652e-05,
+      "loss": 0.651,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99566544486211e-05,
+      "loss": 0.659,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955747383089178e-05,
+      "loss": 0.5663,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954830945824605e-05,
+      "loss": 0.6433,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953905137690067e-05,
+      "loss": 0.5648,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952969959557055e-05,
+      "loss": 0.622,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952025412305887e-05,
+      "loss": 0.5405,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951071496825715e-05,
+      "loss": 0.5506,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995010821401448e-05,
+      "loss": 0.6642,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994913556477897e-05,
+      "loss": 0.5762,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948153550034778e-05,
+      "loss": 0.7733,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947162170706304e-05,
+      "loss": 0.6083,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946161427726775e-05,
+      "loss": 0.5951,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945151322038233e-05,
+      "loss": 0.5985,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994413185459153e-05,
+      "loss": 0.6717,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943103026346327e-05,
+      "loss": 0.5626,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942064838271098e-05,
+      "loss": 0.5061,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941017291343136e-05,
+      "loss": 0.6156,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993996038654853e-05,
+      "loss": 0.5954,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938894124882194e-05,
+      "loss": 0.6028,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993781850734784e-05,
+      "loss": 0.4896,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936733534957985e-05,
+      "loss": 0.5428,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935639208733956e-05,
+      "loss": 0.5641,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934535529705883e-05,
+      "loss": 0.5927,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933422498912707e-05,
+      "loss": 0.5421,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993230011740217e-05,
+      "loss": 0.5383,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931168386230807e-05,
+      "loss": 0.6685,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993002730646396e-05,
+      "loss": 0.6318,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928876879175773e-05,
+      "loss": 0.5521,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927717105449194e-05,
+      "loss": 0.5541,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926547986375957e-05,
+      "loss": 0.7174,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99253695230566e-05,
+      "loss": 0.5646,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924181716600457e-05,
+      "loss": 0.6453,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922984568125657e-05,
+      "loss": 0.5733,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992177807875912e-05,
+      "loss": 0.5083,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920562249636567e-05,
+      "loss": 0.7351,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99193370819025e-05,
+      "loss": 0.6467,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991810257671022e-05,
+      "loss": 0.557,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991685873522181e-05,
+      "loss": 0.6238,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915605558608156e-05,
+      "loss": 0.6396,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991434304804891e-05,
+      "loss": 0.6323,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913071204732534e-05,
+      "loss": 0.7573,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911790029856255e-05,
+      "loss": 0.6379,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910499524626098e-05,
+      "loss": 0.6332,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990919969025686e-05,
+      "loss": 0.6141,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907890527972135e-05,
+      "loss": 0.515,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906572039004275e-05,
+      "loss": 0.6894,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990524422459444e-05,
+      "loss": 0.7074,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903907085992543e-05,
+      "loss": 0.5172,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990256062445729e-05,
+      "loss": 0.5659,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990120484125615e-05,
+      "loss": 0.6884,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899839737665383e-05,
+      "loss": 0.6405,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989846531497001e-05,
+      "loss": 0.7448,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897081574463834e-05,
+      "loss": 0.5384,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895688517449416e-05,
+      "loss": 0.5957,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98942861452381e-05,
+      "loss": 0.7099,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892874459149994e-05,
+      "loss": 0.6242,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891453460513967e-05,
+      "loss": 0.5719,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890023150667667e-05,
+      "loss": 0.5103,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98885835309575e-05,
+      "loss": 0.6297,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887134602738624e-05,
+      "loss": 0.5221,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988567636737499e-05,
+      "loss": 0.4799,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988420882623928e-05,
+      "loss": 0.5597,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882731980712954e-05,
+      "loss": 0.5519,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881245832186217e-05,
+      "loss": 0.6928,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879750382058045e-05,
+      "loss": 0.519,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878245631736158e-05,
+      "loss": 0.6014,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987673158263704e-05,
+      "loss": 0.5398,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875208236185926e-05,
+      "loss": 0.4896,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873675593816795e-05,
+      "loss": 0.482,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987213365697239e-05,
+      "loss": 0.5971,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987058242710419e-05,
+      "loss": 0.6148,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986902190567243e-05,
+      "loss": 0.6135,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867452094146094e-05,
+      "loss": 0.5839,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98658729940029e-05,
+      "loss": 0.5307,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986428460672932e-05,
+      "loss": 0.7045,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862686933820567e-05,
+      "loss": 0.5501,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861079976780585e-05,
+      "loss": 0.6668,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859463737122075e-05,
+      "loss": 0.6742,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985783821636646e-05,
+      "loss": 0.5782,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856203416043908e-05,
+      "loss": 0.5667,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985455933769332e-05,
+      "loss": 0.6995,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985290598286233e-05,
+      "loss": 0.5264,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851243353107302e-05,
+      "loss": 0.68,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849571449993342e-05,
+      "loss": 0.6634,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847890275094274e-05,
+      "loss": 0.6145,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984619982999265e-05,
+      "loss": 0.564,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844500116279758e-05,
+      "loss": 0.5531,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98427911355556e-05,
+      "loss": 0.6641,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984107288942891e-05,
+      "loss": 0.5726,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839345379517135e-05,
+      "loss": 0.6229,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837608607446452e-05,
+      "loss": 0.5764,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983586257485175e-05,
+      "loss": 0.6053,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834107283376634e-05,
+      "loss": 0.6057,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832342734673435e-05,
+      "loss": 0.6674,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830568930403187e-05,
+      "loss": 0.59,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982878587223564e-05,
+      "loss": 0.6937,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826993561849262e-05,
+      "loss": 0.7042,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982519200093122e-05,
+      "loss": 0.4996,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823381191177395e-05,
+      "loss": 0.5425,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982156113429237e-05,
+      "loss": 0.4302,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819731831989436e-05,
+      "loss": 0.533,
+      "step": 297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981789328599059e-05,
+      "loss": 0.6042,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816045498026528e-05,
+      "loss": 0.4766,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814188469836638e-05,
+      "loss": 0.6128,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812322203169018e-05,
+      "loss": 0.5039,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810446699780456e-05,
+      "loss": 0.7042,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980856196143643e-05,
+      "loss": 0.5281,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806667989911133e-05,
+      "loss": 0.617,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980476478698742e-05,
+      "loss": 0.7268,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802852354456856e-05,
+      "loss": 0.588,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980093069411969e-05,
+      "loss": 0.6001,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798999807784848e-05,
+      "loss": 0.7233,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797059697269958e-05,
+      "loss": 0.5678,
+      "step": 309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795110364401313e-05,
+      "loss": 0.6132,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793151811013903e-05,
+      "loss": 0.6301,
+      "step": 311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791184038951387e-05,
+      "loss": 0.5701,
+      "step": 312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789207050066104e-05,
+      "loss": 0.582,
+      "step": 313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787220846219075e-05,
+      "loss": 0.4692,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785225429279987e-05,
+      "loss": 0.5731,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978322080112721e-05,
+      "loss": 0.5944,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781206963647767e-05,
+      "loss": 0.5432,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977918391873737e-05,
+      "loss": 0.651,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777151668300393e-05,
+      "loss": 0.5822,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775110214249863e-05,
+      "loss": 0.511,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773059558507484e-05,
+      "loss": 0.5919,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770999703003622e-05,
+      "loss": 0.633,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768930649677298e-05,
+      "loss": 0.6803,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976685240047619e-05,
+      "loss": 0.4619,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764764957356633e-05,
+      "loss": 0.6416,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976266832228362e-05,
+      "loss": 0.6461,
+      "step": 326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97605624972308e-05,
+      "loss": 0.6106,
+      "step": 327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758447484180467e-05,
+      "loss": 0.6411,
+      "step": 328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756323285123557e-05,
+      "loss": 0.5834,
+      "step": 329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975418990205967e-05,
+      "loss": 0.5695,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975204733699704e-05,
+      "loss": 0.7151,
+      "step": 331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749895591952545e-05,
+      "loss": 0.5321,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747734668951707e-05,
+      "loss": 0.6896,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974556457002869e-05,
+      "loss": 0.555,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743385297226284e-05,
+      "loss": 0.6247,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741196852595937e-05,
+      "loss": 0.5501,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738999238197705e-05,
+      "loss": 0.59,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736792456100298e-05,
+      "loss": 0.4725,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734576508381037e-05,
+      "loss": 0.4614,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732351397125885e-05,
+      "loss": 0.538,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730117124429424e-05,
+      "loss": 0.4781,
+      "step": 341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727873692394863e-05,
+      "loss": 0.5858,
+      "step": 342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972562110313403e-05,
+      "loss": 0.5906,
+      "step": 343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723359358767374e-05,
+      "loss": 0.5313,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972108846142397e-05,
+      "loss": 0.6469,
+      "step": 345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718808413241493e-05,
+      "loss": 0.5752,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971651921636625e-05,
+      "loss": 0.6171,
+      "step": 347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971422087295314e-05,
+      "loss": 0.4966,
+      "step": 348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711913385165697e-05,
+      "loss": 0.5946,
+      "step": 349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970959675517604e-05,
+      "loss": 0.6607,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9707270985164903e-05,
+      "loss": 0.6308,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970493607732163e-05,
+      "loss": 0.6797,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970259203384415e-05,
+      "loss": 0.5635,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700238856939013e-05,
+      "loss": 0.5409,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697876548821352e-05,
+      "loss": 0.6131,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695505111714896e-05,
+      "loss": 0.4933,
+      "step": 356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693124547851978e-05,
+      "loss": 0.5981,
+      "step": 357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969073485947351e-05,
+      "loss": 0.5688,
+      "step": 358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688336048829e-05,
+      "loss": 0.5993,
+      "step": 359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685928118176547e-05,
+      "loss": 0.5721,
+      "step": 360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683511069782822e-05,
+      "loss": 0.5633,
+      "step": 361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681084905923092e-05,
+      "loss": 0.6658,
+      "step": 362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678649628881193e-05,
+      "loss": 0.5445,
+      "step": 363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676205240949557e-05,
+      "loss": 0.6024,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673751744429174e-05,
+      "loss": 0.6359,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671289141629613e-05,
+      "loss": 0.6644,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668817434869026e-05,
+      "loss": 0.6302,
+      "step": 367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966633662647412e-05,
+      "loss": 0.639,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663846718780184e-05,
+      "loss": 0.7623,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661347714131055e-05,
+      "loss": 0.5962,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658839614879152e-05,
+      "loss": 0.5981,
+      "step": 371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656322423385445e-05,
+      "loss": 0.5895,
+      "step": 372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965379614201946e-05,
+      "loss": 0.5485,
+      "step": 373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651260773159287e-05,
+      "loss": 0.6007,
+      "step": 374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648716319191567e-05,
+      "loss": 0.5798,
+      "step": 375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646162782511494e-05,
+      "loss": 0.7008,
+      "step": 376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964360016552281e-05,
+      "loss": 0.6235,
+      "step": 377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964102847063781e-05,
+      "loss": 0.5634,
+      "step": 378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638447700277327e-05,
+      "loss": 0.4726,
+      "step": 379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635857856870742e-05,
+      "loss": 0.579,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633258942855974e-05,
+      "loss": 0.5028,
+      "step": 381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963065096067948e-05,
+      "loss": 0.5203,
+      "step": 382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628033912796263e-05,
+      "loss": 0.5761,
+      "step": 383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625407801669846e-05,
+      "loss": 0.721,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622772629772287e-05,
+      "loss": 0.5572,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620128399584187e-05,
+      "loss": 0.5377,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961747511359465e-05,
+      "loss": 0.6419,
+      "step": 387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614812774301325e-05,
+      "loss": 0.5385,
+      "step": 388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612141384210374e-05,
+      "loss": 0.6411,
+      "step": 389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609460945836477e-05,
+      "loss": 0.6461,
+      "step": 390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960677146170284e-05,
+      "loss": 0.42,
+      "step": 391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604072934341178e-05,
+      "loss": 0.6431,
+      "step": 392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601365366291716e-05,
+      "loss": 0.524,
+      "step": 393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95986487601032e-05,
+      "loss": 0.7097,
+      "step": 394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595923118332874e-05,
+      "loss": 0.5287,
+      "step": 395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593188443546487e-05,
+      "loss": 0.5921,
+      "step": 396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95904447383183e-05,
+      "loss": 0.6481,
+      "step": 397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958769200523107e-05,
+      "loss": 0.5957,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584930246876045e-05,
+      "loss": 0.6141,
+      "step": 399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958215946585298e-05,
+      "loss": 0.6318,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579379664770122e-05,
+      "loss": 0.6687,
+      "step": 401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576590846244205e-05,
+      "loss": 0.5881,
+      "step": 402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573793012900447e-05,
+      "loss": 0.5793,
+      "step": 403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570986167372563e-05,
+      "loss": 0.7022,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956817031230274e-05,
+      "loss": 0.5837,
+      "step": 405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565345450341662e-05,
+      "loss": 0.6932,
+      "step": 406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956251158414847e-05,
+      "loss": 0.618,
+      "step": 407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95596687163908e-05,
+      "loss": 0.4939,
+      "step": 408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556816849744756e-05,
+      "loss": 0.6153,
+      "step": 409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955395598689491e-05,
+      "loss": 0.6287,
+      "step": 410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551086130534294e-05,
+      "loss": 0.6066,
+      "step": 411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548207283364426e-05,
+      "loss": 0.7029,
+      "step": 412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545319448095272e-05,
+      "loss": 0.6269,
+      "step": 413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542422627445264e-05,
+      "loss": 0.6712,
+      "step": 414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539516824141298e-05,
+      "loss": 0.6228,
+      "step": 415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536602040918712e-05,
+      "loss": 0.5642,
+      "step": 416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953367828052131e-05,
+      "loss": 0.639,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530745545701338e-05,
+      "loss": 0.4925,
+      "step": 418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527803839219493e-05,
+      "loss": 0.569,
+      "step": 419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524853163844922e-05,
+      "loss": 0.583,
+      "step": 420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952189352235521e-05,
+      "loss": 0.5729,
+      "step": 421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951892491753638e-05,
+      "loss": 0.535,
+      "step": 422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95159473521829e-05,
+      "loss": 0.548,
+      "step": 423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951296082909766e-05,
+      "loss": 0.5671,
+      "step": 424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509965351091998e-05,
+      "loss": 0.651,
+      "step": 425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506960920985668e-05,
+      "loss": 0.5577,
+      "step": 426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503947541606866e-05,
+      "loss": 0.4993,
+      "step": 427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500925215792194e-05,
+      "loss": 0.6122,
+      "step": 428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949789394638669e-05,
+      "loss": 0.5614,
+      "step": 429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494853736243803e-05,
+      "loss": 0.7051,
+      "step": 430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491804588225408e-05,
+      "loss": 0.4516,
+      "step": 431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488746505201774e-05,
+      "loss": 0.5827,
+      "step": 432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485679490051604e-05,
+      "loss": 0.4856,
+      "step": 433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482603545662e-05,
+      "loss": 0.4823,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479518674928458e-05,
+      "loss": 0.6917,
+      "step": 435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94764248807549e-05,
+      "loss": 0.6136,
+      "step": 436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473322166053622e-05,
+      "loss": 0.5264,
+      "step": 437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470210533745336e-05,
+      "loss": 0.6577,
+      "step": 438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946708998675915e-05,
+      "loss": 0.591,
+      "step": 439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463960528032542e-05,
+      "loss": 0.5863,
+      "step": 440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460822160511404e-05,
+      "loss": 0.5821,
+      "step": 441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457674887149997e-05,
+      "loss": 0.6257,
+      "step": 442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945451871091098e-05,
+      "loss": 0.6554,
+      "step": 443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451353634765376e-05,
+      "loss": 0.4248,
+      "step": 444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448179661692598e-05,
+      "loss": 0.5471,
+      "step": 445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444996794680433e-05,
+      "loss": 0.6663,
+      "step": 446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441805036725036e-05,
+      "loss": 0.5354,
+      "step": 447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438604390830927e-05,
+      "loss": 0.6591,
+      "step": 448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435394860011007e-05,
+      "loss": 0.6939,
+      "step": 449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943217644728653e-05,
+      "loss": 0.5254,
+      "step": 450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9428949155687112e-05,
+      "loss": 0.5728,
+      "step": 451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425712988250726e-05,
+      "loss": 0.5812,
+      "step": 452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422467948023705e-05,
+      "loss": 0.5943,
+      "step": 453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419214038060738e-05,
+      "loss": 0.5578,
+      "step": 454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415951261424845e-05,
+      "loss": 0.5646,
+      "step": 455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941267962118741e-05,
+      "loss": 0.6179,
+      "step": 456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409399120428158e-05,
+      "loss": 0.4206,
+      "step": 457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940610976223515e-05,
+      "loss": 0.552,
+      "step": 458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402811549704786e-05,
+      "loss": 0.5266,
+      "step": 459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399504485941806e-05,
+      "loss": 0.6541,
+      "step": 460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396188574059273e-05,
+      "loss": 0.6881,
+      "step": 461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939286381717858e-05,
+      "loss": 0.5448,
+      "step": 462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938953021842946e-05,
+      "loss": 0.6436,
+      "step": 463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386187780949956e-05,
+      "loss": 0.6005,
+      "step": 464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382836507886436e-05,
+      "loss": 0.5762,
+      "step": 465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379476402393577e-05,
+      "loss": 0.6223,
+      "step": 466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376107467634383e-05,
+      "loss": 0.4984,
+      "step": 467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372729706780162e-05,
+      "loss": 0.5284,
+      "step": 468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936934312301053e-05,
+      "loss": 0.5632,
+      "step": 469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365947719513417e-05,
+      "loss": 0.6314,
+      "step": 470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362543499485037e-05,
+      "loss": 0.644,
+      "step": 471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359130466129926e-05,
+      "loss": 0.5142,
+      "step": 472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93557086226609e-05,
+      "loss": 0.589,
+      "step": 473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352277972299072e-05,
+      "loss": 0.563,
+      "step": 474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934883851827385e-05,
+      "loss": 0.7014,
+      "step": 475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345390263822925e-05,
+      "loss": 0.5705,
+      "step": 476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934193321219227e-05,
+      "loss": 0.565,
+      "step": 477
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933846736663615e-05,
+      "loss": 0.6085,
+      "step": 478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334992730417092e-05,
+      "loss": 0.6343,
+      "step": 479
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331509306805912e-05,
+      "loss": 0.4687,
+      "step": 480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328017099081688e-05,
+      "loss": 0.5377,
+      "step": 481
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324516110531774e-05,
+      "loss": 0.4698,
+      "step": 482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932100634445179e-05,
+      "loss": 0.5083,
+      "step": 483
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317487804145606e-05,
+      "loss": 0.4828,
+      "step": 484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931396049292537e-05,
+      "loss": 0.5947,
+      "step": 485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310424414111473e-05,
+      "loss": 0.6254,
+      "step": 486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306879571032566e-05,
+      "loss": 0.4171,
+      "step": 487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303325967025546e-05,
+      "loss": 0.5334,
+      "step": 488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929976360543556e-05,
+      "loss": 0.6015,
+      "step": 489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296192489615992e-05,
+      "loss": 0.554,
+      "step": 490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292612622928482e-05,
+      "loss": 0.6288,
+      "step": 491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289024008742895e-05,
+      "loss": 0.6467,
+      "step": 492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285426650437326e-05,
+      "loss": 0.5877,
+      "step": 493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281820551398117e-05,
+      "loss": 0.5462,
+      "step": 494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278205715019828e-05,
+      "loss": 0.5483,
+      "step": 495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927458214470524e-05,
+      "loss": 0.5675,
+      "step": 496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927094984386537e-05,
+      "loss": 0.5814,
+      "step": 497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926730881591943e-05,
+      "loss": 0.5079,
+      "step": 498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263659064294877e-05,
+      "loss": 0.6087,
+      "step": 499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260000592427354e-05,
+      "loss": 0.589,
+      "step": 500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925633340376073e-05,
+      "loss": 0.5958,
+      "step": 501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252657501747066e-05,
+      "loss": 0.5637,
+      "step": 502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248972889846632e-05,
+      "loss": 0.6145,
+      "step": 503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245279571527905e-05,
+      "loss": 0.661,
+      "step": 504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241577550267537e-05,
+      "loss": 0.6614,
+      "step": 505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237866829550395e-05,
+      "loss": 0.7292,
+      "step": 506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923414741286952e-05,
+      "loss": 0.6501,
+      "step": 507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923041930372615e-05,
+      "loss": 0.5259,
+      "step": 508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226682505629688e-05,
+      "loss": 0.66,
+      "step": 509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222937022097737e-05,
+      "loss": 0.5337,
+      "step": 510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219182856656063e-05,
+      "loss": 0.5447,
+      "step": 511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215420012838608e-05,
+      "loss": 0.4617,
+      "step": 512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921164849418748e-05,
+      "loss": 0.5306,
+      "step": 513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207868304252963e-05,
+      "loss": 0.5256,
+      "step": 514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204079446593494e-05,
+      "loss": 0.5579,
+      "step": 515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200281924775672e-05,
+      "loss": 0.5938,
+      "step": 516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919647574237425e-05,
+      "loss": 0.4979,
+      "step": 517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9192660902972146e-05,
+      "loss": 0.4901,
+      "step": 518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188837410160403e-05,
+      "loss": 0.5557,
+      "step": 519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185005267538227e-05,
+      "loss": 0.6571,
+      "step": 520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918116447871297e-05,
+      "loss": 0.6361,
+      "step": 521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917731504730011e-05,
+      "loss": 0.6691,
+      "step": 522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173456976923273e-05,
+      "loss": 0.5331,
+      "step": 523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169590271214203e-05,
+      "loss": 0.5732,
+      "step": 524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916571493381278e-05,
+      "loss": 0.5812,
+      "step": 525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161830968367017e-05,
+      "loss": 0.4706,
+      "step": 526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915793837853304e-05,
+      "loss": 0.5034,
+      "step": 527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154037167975095e-05,
+      "loss": 0.5332,
+      "step": 528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150127340365538e-05,
+      "loss": 0.5235,
+      "step": 529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914620889938485e-05,
+      "loss": 0.5592,
+      "step": 530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914228184872161e-05,
+      "loss": 0.4977,
+      "step": 531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91383461920725e-05,
+      "loss": 0.5723,
+      "step": 532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913440193314231e-05,
+      "loss": 0.4567,
+      "step": 533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913044907564392e-05,
+      "loss": 0.709,
+      "step": 534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126487623298317e-05,
+      "loss": 0.6287,
+      "step": 535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122517579834566e-05,
+      "loss": 0.5212,
+      "step": 536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911853894898982e-05,
+      "loss": 0.673,
+      "step": 537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911455173450932e-05,
+      "loss": 0.5847,
+      "step": 538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110555940146395e-05,
+      "loss": 0.5873,
+      "step": 539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910655156966243e-05,
+      "loss": 0.5889,
+      "step": 540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102538626826905e-05,
+      "loss": 0.5961,
+      "step": 541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098517115417353e-05,
+      "loss": 0.6719,
+      "step": 542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094487039219376e-05,
+      "loss": 0.6105,
+      "step": 543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090448402026652e-05,
+      "loss": 0.5078,
+      "step": 544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086401207640898e-05,
+      "loss": 0.5859,
+      "step": 545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082345459871902e-05,
+      "loss": 0.5802,
+      "step": 546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078281162537492e-05,
+      "loss": 0.5236,
+      "step": 547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074208319463558e-05,
+      "loss": 0.5815,
+      "step": 548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070126934484014e-05,
+      "loss": 0.5173,
+      "step": 549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066037011440835e-05,
+      "loss": 0.6751,
+      "step": 550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.906193855418402e-05,
+      "loss": 0.656,
+      "step": 551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905783156657161e-05,
+      "loss": 0.5327,
+      "step": 552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053716052469677e-05,
+      "loss": 0.646,
+      "step": 553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90495920157523e-05,
+      "loss": 0.6325,
+      "step": 554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045459460301607e-05,
+      "loss": 0.5169,
+      "step": 555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904131839000773e-05,
+      "loss": 0.775,
+      "step": 556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037168808768824e-05,
+      "loss": 0.5615,
+      "step": 557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033010720491042e-05,
+      "loss": 0.5576,
+      "step": 558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028844129088563e-05,
+      "loss": 0.6372,
+      "step": 559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902466903848356e-05,
+      "loss": 0.6018,
+      "step": 560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90204854526062e-05,
+      "loss": 0.5806,
+      "step": 561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901629337539467e-05,
+      "loss": 0.4229,
+      "step": 562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012092810795125e-05,
+      "loss": 0.5306,
+      "step": 563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007883762761722e-05,
+      "loss": 0.646,
+      "step": 564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90036662352566e-05,
+      "loss": 0.591,
+      "step": 565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999440232249887e-05,
+      "loss": 0.5734,
+      "step": 566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995205757719684e-05,
+      "loss": 0.5493,
+      "step": 567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899096281565206e-05,
+      "loss": 0.7467,
+      "step": 568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898671141004107e-05,
+      "loss": 0.6451,
+      "step": 569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982451544888722e-05,
+      "loss": 0.5566,
+      "step": 570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978183224204994e-05,
+      "loss": 0.7828,
+      "step": 571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897390645200782e-05,
+      "loss": 0.528,
+      "step": 572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969621232323102e-05,
+      "loss": 0.5307,
+      "step": 573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896532756918467e-05,
+      "loss": 0.6964,
+      "step": 574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961025466634325e-05,
+      "loss": 0.594,
+      "step": 575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956714928721807e-05,
+      "loss": 0.6491,
+      "step": 576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952395959504782e-05,
+      "loss": 0.68,
+      "step": 577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948068563048877e-05,
+      "loss": 0.6499,
+      "step": 578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943732743427625e-05,
+      "loss": 0.6792,
+      "step": 579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939388504722512e-05,
+      "loss": 0.5468,
+      "step": 580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935035851022936e-05,
+      "loss": 0.5742,
+      "step": 581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893067478642622e-05,
+      "loss": 0.4539,
+      "step": 582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89263053150376e-05,
+      "loss": 0.6544,
+      "step": 583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921927440970234e-05,
+      "loss": 0.6194,
+      "step": 584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8917541168345188e-05,
+      "loss": 0.5728,
+      "step": 585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891314650129142e-05,
+      "loss": 0.482,
+      "step": 586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908743443945816e-05,
+      "loss": 0.5986,
+      "step": 587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904332000453134e-05,
+      "loss": 0.5032,
+      "step": 588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889991217496604e-05,
+      "loss": 0.5747,
+      "step": 589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895483971645093e-05,
+      "loss": 0.6525,
+      "step": 590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891047394658726e-05,
+      "loss": 0.5651,
+      "step": 591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886602448183264e-05,
+      "loss": 0.478,
+      "step": 592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888214913640291e-05,
+      "loss": 0.6513,
+      "step": 593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877687463509742e-05,
+      "loss": 0.5262,
+      "step": 594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873217433703702e-05,
+      "loss": 0.6056,
+      "step": 595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868739051192602e-05,
+      "loss": 0.5354,
+      "step": 596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864252320192125e-05,
+      "loss": 0.6152,
+      "step": 597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859757244925795e-05,
+      "loss": 0.5842,
+      "step": 598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885525382962501e-05,
+      "loss": 0.6121,
+      "step": 599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850742078529008e-05,
+      "loss": 0.5832,
+      "step": 600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846221995884876e-05,
+      "loss": 0.4945,
+      "step": 601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884169358594754e-05,
+      "loss": 0.5319,
+      "step": 602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837156852979775e-05,
+      "loss": 0.5722,
+      "step": 603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883261180125218e-05,
+      "loss": 0.6379,
+      "step": 604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828058435043188e-05,
+      "loss": 0.4889,
+      "step": 605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823496758639065e-05,
+      "loss": 0.6678,
+      "step": 606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881892677633389e-05,
+      "loss": 0.6402,
+      "step": 607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814348492429568e-05,
+      "loss": 0.644,
+      "step": 608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809761911235816e-05,
+      "loss": 0.5241,
+      "step": 609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880516703707016e-05,
+      "loss": 0.617,
+      "step": 610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880056387425793e-05,
+      "loss": 0.5249,
+      "step": 611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879595242713227e-05,
+      "loss": 0.5357,
+      "step": 612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879133270003411e-05,
+      "loss": 0.5661,
+      "step": 613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786704697312185e-05,
+      "loss": 0.5723,
+      "step": 614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878206842332301e-05,
+      "loss": 0.5806,
+      "step": 615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777423882430885e-05,
+      "loss": 0.5704,
+      "step": 616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772771079007904e-05,
+      "loss": 0.5373,
+      "step": 617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8768110017433932e-05,
+      "loss": 0.6163,
+      "step": 618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87634407020966e-05,
+      "loss": 0.6566,
+      "step": 619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875876313739133e-05,
+      "loss": 0.606,
+      "step": 620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754077327721283e-05,
+      "loss": 0.5599,
+      "step": 621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749383277497403e-05,
+      "loss": 0.6038,
+      "step": 622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874468099113837e-05,
+      "loss": 0.5305,
+      "step": 623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739970473070643e-05,
+      "loss": 0.5348,
+      "step": 624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735251727728414e-05,
+      "loss": 0.6944,
+      "step": 625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730524759553616e-05,
+      "loss": 0.6038,
+      "step": 626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725789572995925e-05,
+      "loss": 0.5353,
+      "step": 627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721046172512766e-05,
+      "loss": 0.5538,
+      "step": 628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716294562569284e-05,
+      "loss": 0.6365,
+      "step": 629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711534747638352e-05,
+      "loss": 0.7199,
+      "step": 630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706766732200567e-05,
+      "loss": 0.7401,
+      "step": 631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701990520744256e-05,
+      "loss": 0.4909,
+      "step": 632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697206117765445e-05,
+      "loss": 0.4849,
+      "step": 633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692413527767882e-05,
+      "loss": 0.6009,
+      "step": 634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687612755263016e-05,
+      "loss": 0.5968,
+      "step": 635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682803804770003e-05,
+      "loss": 0.4663,
+      "step": 636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677986680815692e-05,
+      "loss": 0.5477,
+      "step": 637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867316138793464e-05,
+      "loss": 0.4763,
+      "step": 638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668327930669075e-05,
+      "loss": 0.6959,
+      "step": 639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663486313568914e-05,
+      "loss": 0.5139,
+      "step": 640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865863654119177e-05,
+      "loss": 0.4871,
+      "step": 641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653778618102914e-05,
+      "loss": 0.5536,
+      "step": 642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86489125488753e-05,
+      "loss": 0.5504,
+      "step": 643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644038338089556e-05,
+      "loss": 0.5935,
+      "step": 644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639155990333954e-05,
+      "loss": 0.6678,
+      "step": 645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863426551020445e-05,
+      "loss": 0.6319,
+      "step": 646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629366902304634e-05,
+      "loss": 0.5054,
+      "step": 647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862446017124576e-05,
+      "loss": 0.6295,
+      "step": 648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619545321646728e-05,
+      "loss": 0.5496,
+      "step": 649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614622358134075e-05,
+      "loss": 0.4941,
+      "step": 650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609691285341976e-05,
+      "loss": 0.5425,
+      "step": 651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604752107912242e-05,
+      "loss": 0.4131,
+      "step": 652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599804830494316e-05,
+      "loss": 0.6096,
+      "step": 653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594849457745262e-05,
+      "loss": 0.5523,
+      "step": 654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589885994329763e-05,
+      "loss": 0.7017,
+      "step": 655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584914444920123e-05,
+      "loss": 0.5662,
+      "step": 656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579934814196256e-05,
+      "loss": 0.5702,
+      "step": 657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574947106845677e-05,
+      "loss": 0.7015,
+      "step": 658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569951327563513e-05,
+      "loss": 0.5572,
+      "step": 659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564947481052485e-05,
+      "loss": 0.549,
+      "step": 660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559935572022912e-05,
+      "loss": 0.6433,
+      "step": 661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554915605192696e-05,
+      "loss": 0.5292,
+      "step": 662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549887585287323e-05,
+      "loss": 0.6346,
+      "step": 663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854485151703987e-05,
+      "loss": 0.4444,
+      "step": 664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853980740519098e-05,
+      "loss": 0.5869,
+      "step": 665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534755254488875e-05,
+      "loss": 0.6078,
+      "step": 666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852969506968934e-05,
+      "loss": 0.7132,
+      "step": 667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852462685555573e-05,
+      "loss": 0.5018,
+      "step": 668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519550616858938e-05,
+      "loss": 0.6444,
+      "step": 669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514466358377435e-05,
+      "loss": 0.6493,
+      "step": 670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509374084897233e-05,
+      "loss": 0.6755,
+      "step": 671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850427380121188e-05,
+      "loss": 0.5806,
+      "step": 672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849916551212248e-05,
+      "loss": 0.5782,
+      "step": 673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849404922243766e-05,
+      "loss": 0.6805,
+      "step": 674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848892493697358e-05,
+      "loss": 0.8167,
+      "step": 675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483792660553937e-05,
+      "loss": 0.5368,
+      "step": 676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478652398009935e-05,
+      "loss": 0.6508,
+      "step": 677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473504154180304e-05,
+      "loss": 0.5074,
+      "step": 678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468347933911293e-05,
+      "loss": 0.5354,
+      "step": 679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463183742056644e-05,
+      "loss": 0.565,
+      "step": 680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845801158347762e-05,
+      "loss": 0.6088,
+      "step": 681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845283146304297e-05,
+      "loss": 0.6138,
+      "step": 682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447643385628948e-05,
+      "loss": 0.5786,
+      "step": 683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8442447356119285e-05,
+      "loss": 0.5596,
+      "step": 684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8437243379405215e-05,
+      "loss": 0.5596,
+      "step": 685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843203146038544e-05,
+      "loss": 0.4786,
+      "step": 686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426811603966138e-05,
+      "loss": 0.5663,
+      "step": 687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842158381506097e-05,
+      "loss": 0.5936,
+      "step": 688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416348098591054e-05,
+      "loss": 0.6319,
+      "step": 689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411104459484974e-05,
+      "loss": 0.494,
+      "step": 690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405852902678775e-05,
+      "loss": 0.6389,
+      "step": 691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400593433115944e-05,
+      "loss": 0.5861,
+      "step": 692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839532605574743e-05,
+      "loss": 0.6113,
+      "step": 693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390050775531622e-05,
+      "loss": 0.5968,
+      "step": 694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838476759743434e-05,
+      "loss": 0.5379,
+      "step": 695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379476526428856e-05,
+      "loss": 0.529,
+      "step": 696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374177567495854e-05,
+      "loss": 0.5427,
+      "step": 697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368870725623447e-05,
+      "loss": 0.6654,
+      "step": 698
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363556005807174e-05,
+      "loss": 0.654,
+      "step": 699
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358233413049996e-05,
+      "loss": 0.5141,
+      "step": 700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352902952362267e-05,
+      "loss": 0.5262,
+      "step": 701
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834756462876176e-05,
+      "loss": 0.599,
+      "step": 702
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834221844727365e-05,
+      "loss": 0.6966,
+      "step": 703
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83368644129305e-05,
+      "loss": 0.6494,
+      "step": 704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331502530772276e-05,
+      "loss": 0.615,
+      "step": 705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326132805846327e-05,
+      "loss": 0.589,
+      "step": 706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832075524320738e-05,
+      "loss": 0.4572,
+      "step": 707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315369847917544e-05,
+      "loss": 0.5444,
+      "step": 708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309976625046307e-05,
+      "loss": 0.6879,
+      "step": 709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304575579670513e-05,
+      "loss": 0.5118,
+      "step": 710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829916671687438e-05,
+      "loss": 0.5079,
+      "step": 711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293750041749482e-05,
+      "loss": 0.5476,
+      "step": 712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8288325559394745e-05,
+      "loss": 0.5496,
+      "step": 713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828289327491644e-05,
+      "loss": 0.5361,
+      "step": 714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277453193428197e-05,
+      "loss": 0.5177,
+      "step": 715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272005320050973e-05,
+      "loss": 0.6382,
+      "step": 716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8266549659913062e-05,
+      "loss": 0.5763,
+      "step": 717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.826108621815009e-05,
+      "loss": 0.5042,
+      "step": 718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8255614999905006e-05,
+      "loss": 0.7477,
+      "step": 719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250136010328085e-05,
+      "loss": 0.5495,
+      "step": 720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244649254576906e-05,
+      "loss": 0.5103,
+      "step": 721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239154737816377e-05,
+      "loss": 0.6496,
+      "step": 722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233652465218687e-05,
+      "loss": 0.5603,
+      "step": 723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822814244196335e-05,
+      "loss": 0.5907,
+      "step": 724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822262467323716e-05,
+      "loss": 0.5509,
+      "step": 725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217099164234213e-05,
+      "loss": 0.6322,
+      "step": 726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821156592015588e-05,
+      "loss": 0.3714,
+      "step": 727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820602494621082e-05,
+      "loss": 0.6093,
+      "step": 728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820047624761497e-05,
+      "loss": 0.533,
+      "step": 729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194919829591532e-05,
+      "loss": 0.4948,
+      "step": 730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189355697370983e-05,
+      "loss": 0.5605,
+      "step": 731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183783856191058e-05,
+      "loss": 0.5552,
+      "step": 732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178204311296744e-05,
+      "loss": 0.4659,
+      "step": 733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172617067940285e-05,
+      "loss": 0.538,
+      "step": 734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816702213138117e-05,
+      "loss": 0.5112,
+      "step": 735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161419506886136e-05,
+      "loss": 0.6163,
+      "step": 736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155809199729144e-05,
+      "loss": 0.5028,
+      "step": 737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81501912151914e-05,
+      "loss": 0.4856,
+      "step": 738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814456555856133e-05,
+      "loss": 0.5819,
+      "step": 739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138932235134582e-05,
+      "loss": 0.5494,
+      "step": 740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133291250214024e-05,
+      "loss": 0.6202,
+      "step": 741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812764260910973e-05,
+      "loss": 0.699,
+      "step": 742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121986317138997e-05,
+      "loss": 0.6092,
+      "step": 743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116322379626304e-05,
+      "loss": 0.5802,
+      "step": 744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811065080190334e-05,
+      "loss": 0.6396,
+      "step": 745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810497158930898e-05,
+      "loss": 0.548,
+      "step": 746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099284747189285e-05,
+      "loss": 0.542,
+      "step": 747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809359028089751e-05,
+      "loss": 0.565,
+      "step": 748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087888195794065e-05,
+      "loss": 0.4999,
+      "step": 749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8082178497246556e-05,
+      "loss": 0.5353,
+      "step": 750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8076461190629737e-05,
+      "loss": 0.6082,
+      "step": 751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070736281325535e-05,
+      "loss": 0.4916,
+      "step": 752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806500377472303e-05,
+      "loss": 0.5863,
+      "step": 753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059263676218454e-05,
+      "loss": 0.4131,
+      "step": 754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8053515991215187e-05,
+      "loss": 0.6253,
+      "step": 755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047760725123745e-05,
+      "loss": 0.5657,
+      "step": 756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041997883361783e-05,
+      "loss": 0.7178,
+      "step": 757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803622747135409e-05,
+      "loss": 0.5428,
+      "step": 758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030449494532583e-05,
+      "loss": 0.6538,
+      "step": 759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802466395833629e-05,
+      "loss": 0.6536,
+      "step": 760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018870868211365e-05,
+      "loss": 0.4922,
+      "step": 761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013070229611072e-05,
+      "loss": 0.6091,
+      "step": 762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007262047995776e-05,
+      "loss": 0.6139,
+      "step": 763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001446328832945e-05,
+      "loss": 0.5525,
+      "step": 764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7995623077597136e-05,
+      "loss": 0.6641,
+      "step": 765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989792299770008e-05,
+      "loss": 0.4998,
+      "step": 766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7983954000840297e-05,
+      "loss": 0.4605,
+      "step": 767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978108186303824e-05,
+      "loss": 0.6522,
+      "step": 768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797225486166348e-05,
+      "loss": 0.6364,
+      "step": 769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966394032429227e-05,
+      "loss": 0.5125,
+      "step": 770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7960525704118093e-05,
+      "loss": 0.6315,
+      "step": 771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954649882254163e-05,
+      "loss": 0.5576,
+      "step": 772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948766572368576e-05,
+      "loss": 0.4357,
+      "step": 773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794287577999952e-05,
+      "loss": 0.5868,
+      "step": 774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936977510692232e-05,
+      "loss": 0.5688,
+      "step": 775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931071769998976e-05,
+      "loss": 0.5227,
+      "step": 776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792515856347906e-05,
+      "loss": 0.6973,
+      "step": 777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7919237896698816e-05,
+      "loss": 0.5645,
+      "step": 778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79133097752316e-05,
+      "loss": 0.4956,
+      "step": 779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907374204657773e-05,
+      "loss": 0.5782,
+      "step": 780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790143119056473e-05,
+      "loss": 0.5363,
+      "step": 781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895480738546855e-05,
+      "loss": 0.5617,
+      "step": 782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889522854205547e-05,
+      "loss": 0.5005,
+      "step": 783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7883557543149186e-05,
+      "loss": 0.5446,
+      "step": 784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787758481099316e-05,
+      "loss": 0.6228,
+      "step": 785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871604663359824e-05,
+      "loss": 0.6375,
+      "step": 786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865617105878536e-05,
+      "loss": 0.6135,
+      "step": 787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859622144185605e-05,
+      "loss": 0.5115,
+      "step": 788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853619783924334e-05,
+      "loss": 0.5973,
+      "step": 789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847610030744967e-05,
+      "loss": 0.4402,
+      "step": 790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784159289030473e-05,
+      "loss": 0.5859,
+      "step": 791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835568368267775e-05,
+      "loss": 0.5832,
+      "step": 792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782953647030524e-05,
+      "loss": 0.5913,
+      "step": 793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782349720209517e-05,
+      "loss": 0.6233,
+      "step": 794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781745056932257e-05,
+      "loss": 0.5347,
+      "step": 795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811396577679366e-05,
+      "loss": 0.5043,
+      "step": 796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805335232864424e-05,
+      "loss": 0.5238,
+      "step": 797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799266540583516e-05,
+      "loss": 0.6254,
+      "step": 798
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793190506549342e-05,
+      "loss": 0.5928,
+      "step": 799
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778710713648151e-05,
+      "loss": 0.6046,
+      "step": 800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781016436106533e-05,
+      "loss": 0.4849,
+      "step": 801
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777491841115782e-05,
+      "loss": 0.6185,
+      "step": 802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7768813067375685e-05,
+      "loss": 0.7124,
+      "step": 803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776270041050732e-05,
+      "loss": 0.6221,
+      "step": 804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756580446306814e-05,
+      "loss": 0.6057,
+      "step": 805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750453180535125e-05,
+      "loss": 0.6962,
+      "step": 806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744318618960085e-05,
+      "loss": 0.6526,
+      "step": 807
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738176767356393e-05,
+      "loss": 0.6475,
+      "step": 808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773202763150562e-05,
+      "loss": 0.5524,
+      "step": 809
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772587121719618e-05,
+      "loss": 0.3573,
+      "step": 810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7719707530223355e-05,
+      "loss": 0.6487,
+      "step": 811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713536576389252e-05,
+      "loss": 0.5975,
+      "step": 812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707358361502838e-05,
+      "loss": 0.626,
+      "step": 813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701172891379905e-05,
+      "loss": 0.4041,
+      "step": 814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7694980171843074e-05,
+      "loss": 0.5191,
+      "step": 815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76887802087218e-05,
+      "loss": 0.6034,
+      "step": 816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682573007852343e-05,
+      "loss": 0.4985,
+      "step": 817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676358575077784e-05,
+      "loss": 0.5236,
+      "step": 818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7670136916248018e-05,
+      "loss": 0.5398,
+      "step": 819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663908037219723e-05,
+      "loss": 0.5993,
+      "step": 820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657671943856392e-05,
+      "loss": 0.5567,
+      "step": 821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651428642028302e-05,
+      "loss": 0.5165,
+      "step": 822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645178137612515e-05,
+      "loss": 0.4476,
+      "step": 823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638920436492876e-05,
+      "loss": 0.6348,
+      "step": 824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763265554456e-05,
+      "loss": 0.6132,
+      "step": 825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762638346771128e-05,
+      "loss": 0.7144,
+      "step": 826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7620104211850853e-05,
+      "loss": 0.5801,
+      "step": 827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613817782889643e-05,
+      "loss": 0.6691,
+      "step": 828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607524186745296e-05,
+      "loss": 0.502,
+      "step": 829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760122342934223e-05,
+      "loss": 0.5797,
+      "step": 830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759491551661159e-05,
+      "loss": 0.5635,
+      "step": 831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588600454491255e-05,
+      "loss": 0.5319,
+      "step": 832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582278248925845e-05,
+      "loss": 0.5809,
+      "step": 833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575948905866694e-05,
+      "loss": 0.6858,
+      "step": 834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7569612431271865e-05,
+      "loss": 0.5489,
+      "step": 835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563268831106123e-05,
+      "loss": 0.6145,
+      "step": 836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556918111340943e-05,
+      "loss": 0.5806,
+      "step": 837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550560277954514e-05,
+      "loss": 0.5508,
+      "step": 838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544195336931705e-05,
+      "loss": 0.7378,
+      "step": 839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537823294264084e-05,
+      "loss": 0.5612,
+      "step": 840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531444155949906e-05,
+      "loss": 0.6616,
+      "step": 841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75250579279941e-05,
+      "loss": 0.5888,
+      "step": 842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518664616408268e-05,
+      "loss": 0.5253,
+      "step": 843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751226422721069e-05,
+      "loss": 0.6422,
+      "step": 844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750585676642629e-05,
+      "loss": 0.6372,
+      "step": 845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499442240086676e-05,
+      "loss": 0.5622,
+      "step": 846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493020654230083e-05,
+      "loss": 0.6601,
+      "step": 847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486592014901398e-05,
+      "loss": 0.6107,
+      "step": 848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748015632815215e-05,
+      "loss": 0.5637,
+      "step": 849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747371360004051e-05,
+      "loss": 0.596,
+      "step": 850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467263836631254e-05,
+      "loss": 0.6005,
+      "step": 851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7460807043995806e-05,
+      "loss": 0.5673,
+      "step": 852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454343228212195e-05,
+      "loss": 0.4938,
+      "step": 853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447872395365062e-05,
+      "loss": 0.6285,
+      "step": 854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441394551545655e-05,
+      "loss": 0.4032,
+      "step": 855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743490970285181e-05,
+      "loss": 0.5473,
+      "step": 856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428417855387986e-05,
+      "loss": 0.5401,
+      "step": 857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74219190152652e-05,
+      "loss": 0.5936,
+      "step": 858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415413188601062e-05,
+      "loss": 0.5818,
+      "step": 859
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408900381519766e-05,
+      "loss": 0.5573,
+      "step": 860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740238060015207e-05,
+      "loss": 0.5999,
+      "step": 861
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395853850635292e-05,
+      "loss": 0.604,
+      "step": 862
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389320139113318e-05,
+      "loss": 0.5809,
+      "step": 863
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7382779471736594e-05,
+      "loss": 0.6859,
+      "step": 864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376231854662095e-05,
+      "loss": 0.5573,
+      "step": 865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736967729405335e-05,
+      "loss": 0.6041,
+      "step": 866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736311579608042e-05,
+      "loss": 0.6411,
+      "step": 867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73565473669199e-05,
+      "loss": 0.5727,
+      "step": 868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7349972012754915e-05,
+      "loss": 0.4976,
+      "step": 869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343389739775094e-05,
+      "loss": 0.6651,
+      "step": 870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733680055417659e-05,
+      "loss": 0.5579,
+      "step": 871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330204462162057e-05,
+      "loss": 0.5968,
+      "step": 872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323601469940656e-05,
+      "loss": 0.5318,
+      "step": 873
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731699158372804e-05,
+      "loss": 0.6045,
+      "step": 874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310374809746352e-05,
+      "loss": 0.5511,
+      "step": 875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730375115422422e-05,
+      "loss": 0.6536,
+      "step": 876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297120623396746e-05,
+      "loss": 0.6166,
+      "step": 877
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729048322350551e-05,
+      "loss": 0.561,
+      "step": 878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283838960798555e-05,
+      "loss": 0.5039,
+      "step": 879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727718784153038e-05,
+      "loss": 0.5567,
+      "step": 880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270529871961946e-05,
+      "loss": 0.6459,
+      "step": 881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726386505836066e-05,
+      "loss": 0.4886,
+      "step": 882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725719340700037e-05,
+      "loss": 0.5563,
+      "step": 883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725051492416136e-05,
+      "loss": 0.6977,
+      "step": 884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7243829616130345e-05,
+      "loss": 0.4394,
+      "step": 885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7237137489200466e-05,
+      "loss": 0.6819,
+      "step": 886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230438549671285e-05,
+      "loss": 0.4881,
+      "step": 887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722373280384877e-05,
+      "loss": 0.593,
+      "step": 888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7217020258045307e-05,
+      "loss": 0.5671,
+      "step": 889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210300918579676e-05,
+      "loss": 0.608,
+      "step": 890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203574791777046e-05,
+      "loss": 0.5355,
+      "step": 891
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7196841883968988e-05,
+      "loss": 0.5573,
+      "step": 892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190102201493446e-05,
+      "loss": 0.6918,
+      "step": 893
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183355750694753e-05,
+      "loss": 0.563,
+      "step": 894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7176602537923597e-05,
+      "loss": 0.455,
+      "step": 895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7169842569537047e-05,
+      "loss": 0.5772,
+      "step": 896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7163075851898518e-05,
+      "loss": 0.6559,
+      "step": 897
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156302391377793e-05,
+      "loss": 0.6433,
+      "step": 898
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7149522194350984e-05,
+      "loss": 0.5346,
+      "step": 899
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142735267200565e-05,
+      "loss": 0.675,
+      "step": 900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135941616315327e-05,
+      "loss": 0.5134,
+      "step": 901
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71291412480904e-05,
+      "loss": 0.7184,
+      "step": 902
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712233416892724e-05,
+      "loss": 0.5908,
+      "step": 903
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7115520385233614e-05,
+      "loss": 0.6182,
+      "step": 904
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710869990342359e-05,
+      "loss": 0.5525,
+      "step": 905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710187272991757e-05,
+      "loss": 0.6614,
+      "step": 906
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7095038871142228e-05,
+      "loss": 0.6183,
+      "step": 907
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088198333530547e-05,
+      "loss": 0.5551,
+      "step": 908
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081351123521786e-05,
+      "loss": 0.6261,
+      "step": 909
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70744972475615e-05,
+      "loss": 0.554,
+      "step": 910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7067636712101497e-05,
+      "loss": 0.6083,
+      "step": 911
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706076952359987e-05,
+      "loss": 0.5609,
+      "step": 912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053895688520976e-05,
+      "loss": 0.5598,
+      "step": 913
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7047015213335418e-05,
+      "loss": 0.5289,
+      "step": 914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704012810452006e-05,
+      "loss": 0.4696,
+      "step": 915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033234368558e-05,
+      "loss": 0.4686,
+      "step": 916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7026334011938583e-05,
+      "loss": 0.6502,
+      "step": 917
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019427041157387e-05,
+      "loss": 0.5776,
+      "step": 918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012513462716205e-05,
+      "loss": 0.5587,
+      "step": 919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005593283123066e-05,
+      "loss": 0.4074,
+      "step": 920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998666508892198e-05,
+      "loss": 0.4839,
+      "step": 921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6991733146544044e-05,
+      "loss": 0.5157,
+      "step": 922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698479320260525e-05,
+      "loss": 0.6761,
+      "step": 923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977846683608658e-05,
+      "loss": 0.5305,
+      "step": 924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970893596093292e-05,
+      "loss": 0.5912,
+      "step": 925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963933946604363e-05,
+      "loss": 0.6518,
+      "step": 926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956967741693264e-05,
+      "loss": 0.558,
+      "step": 927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949994987917554e-05,
+      "loss": 0.7185,
+      "step": 928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943015691840958e-05,
+      "loss": 0.572,
+      "step": 929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6936029860033356e-05,
+      "loss": 0.6179,
+      "step": 930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6929037499070787e-05,
+      "loss": 0.6181,
+      "step": 931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692203861553543e-05,
+      "loss": 0.4946,
+      "step": 932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691503321601561e-05,
+      "loss": 0.659,
+      "step": 933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690802130710578e-05,
+      "loss": 0.6136,
+      "step": 934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901002895406523e-05,
+      "loss": 0.5625,
+      "step": 935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6893977987524544e-05,
+      "loss": 0.6777,
+      "step": 936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688694659007266e-05,
+      "loss": 0.6021,
+      "step": 937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68799087096698e-05,
+      "loss": 0.5496,
+      "step": 938
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872864352940998e-05,
+      "loss": 0.5722,
+      "step": 939
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686581352651738e-05,
+      "loss": 0.5465,
+      "step": 940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685875623703616e-05,
+      "loss": 0.5542,
+      "step": 941
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851692491140645e-05,
+      "loss": 0.5753,
+      "step": 942
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844622295480205e-05,
+      "loss": 0.4621,
+      "step": 943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68375456567103e-05,
+      "loss": 0.5729,
+      "step": 944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830462581492445e-05,
+      "loss": 0.5628,
+      "step": 945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823373076494206e-05,
+      "loss": 0.4689,
+      "step": 946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6816277148389223e-05,
+      "loss": 0.4192,
+      "step": 947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809174803857154e-05,
+      "loss": 0.5846,
+      "step": 948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6802066049583726e-05,
+      "loss": 0.5742,
+      "step": 949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794950892260676e-05,
+      "loss": 0.5218,
+      "step": 950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6787829338585783e-05,
+      "loss": 0.5258,
+      "step": 951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6780701395262846e-05,
+      "loss": 0.5125,
+      "step": 952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677356706900167e-05,
+      "loss": 0.6176,
+      "step": 953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766426366518082e-05,
+      "loss": 0.436,
+      "step": 954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675927929453389e-05,
+      "loss": 0.5429,
+      "step": 955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752125859776924e-05,
+      "loss": 0.6152,
+      "step": 956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744966068980984e-05,
+      "loss": 0.6369,
+      "step": 957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737799928885862e-05,
+      "loss": 0.5653,
+      "step": 958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730627446237325e-05,
+      "loss": 0.5907,
+      "step": 959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672344862778711e-05,
+      "loss": 0.5399,
+      "step": 960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716263480292916e-05,
+      "loss": 0.5666,
+      "step": 961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709072010518407e-05,
+      "loss": 0.6127,
+      "step": 962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701874225233185e-05,
+      "loss": 0.5687,
+      "step": 963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6694670131212816e-05,
+      "loss": 0.5343,
+      "step": 964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687459735238785e-05,
+      "loss": 0.6458,
+      "step": 965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680243044098525e-05,
+      "loss": 0.5378,
+      "step": 966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6673020064585387e-05,
+      "loss": 0.5076,
+      "step": 967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665790803498637e-05,
+      "loss": 0.4329,
+      "step": 968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665855526764347e-05,
+      "loss": 0.5801,
+      "step": 969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651313463830966e-05,
+      "loss": 0.5775,
+      "step": 970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644065398878128e-05,
+      "loss": 0.4988,
+      "step": 971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663681107960784e-05,
+      "loss": 0.5192,
+      "step": 972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662955051284887e-05,
+      "loss": 0.6774,
+      "step": 973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6622283705435868e-05,
+      "loss": 0.5846,
+      "step": 974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6615010664209376e-05,
+      "loss": 0.4476,
+      "step": 975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607731396015782e-05,
+      "loss": 0.6126,
+      "step": 976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600445907707345e-05,
+      "loss": 0.5861,
+      "step": 977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659315420614218e-05,
+      "loss": 0.4985,
+      "step": 978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6585856298184254e-05,
+      "loss": 0.5452,
+      "step": 979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657855219070337e-05,
+      "loss": 0.5855,
+      "step": 980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6571241890575157e-05,
+      "loss": 0.4424,
+      "step": 981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65639254046811e-05,
+      "loss": 0.5902,
+      "step": 982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655660273990849e-05,
+      "loss": 0.4971,
+      "step": 983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654927390315043e-05,
+      "loss": 0.6003,
+      "step": 984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6541938901305844e-05,
+      "loss": 0.6421,
+      "step": 985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653459774127945e-05,
+      "loss": 0.5163,
+      "step": 986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527250429981774e-05,
+      "loss": 0.6142,
+      "step": 987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6519896974329124e-05,
+      "loss": 0.5777,
+      "step": 988
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6512537381243593e-05,
+      "loss": 0.6148,
+      "step": 989
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6505171657653053e-05,
+      "loss": 0.4984,
+      "step": 990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497799810491147e-05,
+      "loss": 0.6247,
+      "step": 991
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490421846697283e-05,
+      "loss": 0.5899,
+      "step": 992
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648303777321662e-05,
+      "loss": 0.4666,
+      "step": 993
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647564759700008e-05,
+      "loss": 0.4866,
+      "step": 994
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646825132500432e-05,
+      "loss": 0.5389,
+      "step": 995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460848964191744e-05,
+      "loss": 0.5135,
+      "step": 996
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453440521530475e-05,
+      "loss": 0.5153,
+      "step": 997
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446026003994373e-05,
+      "loss": 0.5555,
+      "step": 998
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643860541856301e-05,
+      "loss": 0.6321,
+      "step": 999
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6431178772221675e-05,
+      "loss": 0.52,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6423746071961357e-05,
+      "loss": 0.5903,
+      "step": 1001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6416307324778747e-05,
+      "loss": 0.506,
+      "step": 1002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6408862537676228e-05,
+      "loss": 0.4624,
+      "step": 1003
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6401411717661867e-05,
+      "loss": 0.5158,
+      "step": 1004
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639395487174941e-05,
+      "loss": 0.5308,
+      "step": 1005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386492006958275e-05,
+      "loss": 0.5004,
+      "step": 1006
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379023130313556e-05,
+      "loss": 0.5398,
+      "step": 1007
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6371548248845986e-05,
+      "loss": 0.5547,
+      "step": 1008
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636406736959197e-05,
+      "loss": 0.642,
+      "step": 1009
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6356580499593545e-05,
+      "loss": 0.5697,
+      "step": 1010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349087645898398e-05,
+      "loss": 0.6295,
+      "step": 1011
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6341588815559842e-05,
+      "loss": 0.5229,
+      "step": 1012
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6334084015636817e-05,
+      "loss": 0.5874,
+      "step": 1013
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6326573253193886e-05,
+      "loss": 0.4976,
+      "step": 1014
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6319056535301218e-05,
+      "loss": 0.5453,
+      "step": 1015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311533869034596e-05,
+      "loss": 0.4516,
+      "step": 1016
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304005261475397e-05,
+      "loss": 0.5835,
+      "step": 1017
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6296470719710594e-05,
+      "loss": 0.4342,
+      "step": 1018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288930250832745e-05,
+      "loss": 0.5714,
+      "step": 1019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6281383861939985e-05,
+      "loss": 0.6426,
+      "step": 1020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273831560136022e-05,
+      "loss": 0.5369,
+      "step": 1021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266273352530132e-05,
+      "loss": 0.5072,
+      "step": 1022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258709246237155e-05,
+      "loss": 0.529,
+      "step": 1023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251139248377478e-05,
+      "loss": 0.4865,
+      "step": 1024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243563366077025e-05,
+      "loss": 0.5057,
+      "step": 1025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6235981606467282e-05,
+      "loss": 0.4414,
+      "step": 1026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228393976685248e-05,
+      "loss": 0.5519,
+      "step": 1027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220800483873452e-05,
+      "loss": 0.5753,
+      "step": 1028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213201135179955e-05,
+      "loss": 0.6094,
+      "step": 1029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205595937758307e-05,
+      "loss": 0.7094,
+      "step": 1030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619798489876758e-05,
+      "loss": 0.5084,
+      "step": 1031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619036802537235e-05,
+      "loss": 0.774,
+      "step": 1032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6182745324742667e-05,
+      "loss": 0.6858,
+      "step": 1033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175116804054076e-05,
+      "loss": 0.4843,
+      "step": 1034
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167482470487606e-05,
+      "loss": 0.597,
+      "step": 1035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615984233122975e-05,
+      "loss": 0.5346,
+      "step": 1036
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152196393472466e-05,
+      "loss": 0.5945,
+      "step": 1037
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144544664413174e-05,
+      "loss": 0.6371,
+      "step": 1038
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613688715125475e-05,
+      "loss": 0.5275,
+      "step": 1039
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61292238612055e-05,
+      "loss": 0.6511,
+      "step": 1040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612155480147919e-05,
+      "loss": 0.563,
+      "step": 1041
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6113879979294988e-05,
+      "loss": 0.6541,
+      "step": 1042
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610619940187752e-05,
+      "loss": 0.5578,
+      "step": 1043
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098513076456796e-05,
+      "loss": 0.6756,
+      "step": 1044
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090821010268268e-05,
+      "loss": 0.7033,
+      "step": 1045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083123210552768e-05,
+      "loss": 0.514,
+      "step": 1046
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075419684556538e-05,
+      "loss": 0.602,
+      "step": 1047
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067710439531206e-05,
+      "loss": 0.5943,
+      "step": 1048
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059995482733785e-05,
+      "loss": 0.6667,
+      "step": 1049
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6052274821426665e-05,
+      "loss": 0.5713,
+      "step": 1050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6044548462877607e-05,
+      "loss": 0.4648,
+      "step": 1051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6036816414359728e-05,
+      "loss": 0.5668,
+      "step": 1052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6029078683151508e-05,
+      "loss": 0.4588,
+      "step": 1053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021335276536777e-05,
+      "loss": 0.5508,
+      "step": 1054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013586201804706e-05,
+      "loss": 0.5199,
+      "step": 1055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005831466249797e-05,
+      "loss": 0.5081,
+      "step": 1056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998071077171884e-05,
+      "loss": 0.4931,
+      "step": 1057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990305041876125e-05,
+      "loss": 0.6573,
+      "step": 1058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982533367672996e-05,
+      "loss": 0.5802,
+      "step": 1059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597475606187827e-05,
+      "loss": 0.5609,
+      "step": 1060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596697313181303e-05,
+      "loss": 0.5489,
+      "step": 1061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595918458480365e-05,
+      "loss": 0.6869,
+      "step": 1062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951390428181796e-05,
+      "loss": 0.6782,
+      "step": 1063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943590669284405e-05,
+      "loss": 0.6342,
+      "step": 1064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59357853154537e-05,
+      "loss": 0.634,
+      "step": 1065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927974374037165e-05,
+      "loss": 0.6576,
+      "step": 1066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5920157852387537e-05,
+      "loss": 0.4877,
+      "step": 1067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912335757862817e-05,
+      "loss": 0.6025,
+      "step": 1068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5904508097826243e-05,
+      "loss": 0.6248,
+      "step": 1069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58966748796463e-05,
+      "loss": 0.5412,
+      "step": 1070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888836110696704e-05,
+      "loss": 0.5147,
+      "step": 1071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880991798356384e-05,
+      "loss": 0.6432,
+      "step": 1072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873141950009505e-05,
+      "loss": 0.6206,
+      "step": 1073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5865286573045433e-05,
+      "loss": 0.6464,
+      "step": 1074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585742567485874e-05,
+      "loss": 0.444,
+      "step": 1075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584955926284919e-05,
+      "loss": 0.5802,
+      "step": 1076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584168734442175e-05,
+      "loss": 0.5916,
+      "step": 1077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833809926986558e-05,
+      "loss": 0.6746,
+      "step": 1078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825927017958943e-05,
+      "loss": 0.5172,
+      "step": 1079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5818038624759387e-05,
+      "loss": 0.5209,
+      "step": 1080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581014475481354e-05,
+      "loss": 0.5648,
+      "step": 1081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580224541555221e-05,
+      "loss": 0.5115,
+      "step": 1082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579434061441136e-05,
+      "loss": 0.4516,
+      "step": 1083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5786430358832078e-05,
+      "loss": 0.4658,
+      "step": 1084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778514656260595e-05,
+      "loss": 0.606,
+      "step": 1085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577059351414828e-05,
+      "loss": 0.5811,
+      "step": 1086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762666939951605e-05,
+      "loss": 0.56,
+      "step": 1087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754734941132165e-05,
+      "loss": 0.4271,
+      "step": 1088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746797525156658e-05,
+      "loss": 0.5999,
+      "step": 1089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573885469949688e-05,
+      "loss": 0.5938,
+      "step": 1090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5730906471629732e-05,
+      "loss": 0.5754,
+      "step": 1091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572295284903718e-05,
+      "loss": 0.5923,
+      "step": 1092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5714993839206284e-05,
+      "loss": 0.5256,
+      "step": 1093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707029449629168e-05,
+      "loss": 0.6908,
+      "step": 1094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5699059687803023e-05,
+      "loss": 0.5643,
+      "step": 1095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569108456123009e-05,
+      "loss": 0.5738,
+      "step": 1096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568310407741768e-05,
+      "loss": 0.5895,
+      "step": 1097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567511824387812e-05,
+      "loss": 0.5315,
+      "step": 1098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667127068128793e-05,
+      "loss": 0.5387,
+      "step": 1099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56591305576921e-05,
+      "loss": 0.6552,
+      "step": 1100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5651128720095468e-05,
+      "loss": 0.5272,
+      "step": 1101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5643121562871344e-05,
+      "loss": 0.5466,
+      "step": 1102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563510909355717e-05,
+      "loss": 0.5429,
+      "step": 1103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56270913196954e-05,
+      "loss": 0.4077,
+      "step": 1104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619068248833472e-05,
+      "loss": 0.6056,
+      "step": 1105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561103988852382e-05,
+      "loss": 0.5033,
+      "step": 1106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603006246323845e-05,
+      "loss": 0.5435,
+      "step": 1107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559496732979594e-05,
+      "loss": 0.579,
+      "step": 1108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586923146507434e-05,
+      "loss": 0.603,
+      "step": 1109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557887370403064e-05,
+      "loss": 0.5558,
+      "step": 1110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557081900994281e-05,
+      "loss": 0.5831,
+      "step": 1111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5562759071826145e-05,
+      "loss": 0.5736,
+      "step": 1112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5554693897267767e-05,
+      "loss": 0.5512,
+      "step": 1113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554662349385975e-05,
+      "loss": 0.5061,
+      "step": 1114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538547869199077e-05,
+      "loss": 0.5529,
+      "step": 1115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5530467030887642e-05,
+      "loss": 0.5382,
+      "step": 1116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552238098653226e-05,
+      "loss": 0.6215,
+      "step": 1117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514289743744636e-05,
+      "loss": 0.4878,
+      "step": 1118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550619331014137e-05,
+      "loss": 0.4986,
+      "step": 1119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498091693343954e-05,
+      "loss": 0.4478,
+      "step": 1120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5489984900978753e-05,
+      "loss": 0.574,
+      "step": 1121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5481872940677005e-05,
+      "loss": 0.5275,
+      "step": 1122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5473755820074817e-05,
+      "loss": 0.5326,
+      "step": 1123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546563354681315e-05,
+      "loss": 0.617,
+      "step": 1124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5457506128537818e-05,
+      "loss": 0.5129,
+      "step": 1125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544937357289947e-05,
+      "loss": 0.5192,
+      "step": 1126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441235887553604e-05,
+      "loss": 0.6329,
+      "step": 1127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543309308016054e-05,
+      "loss": 0.6574,
+      "step": 1128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5424945158385415e-05,
+      "loss": 0.525,
+      "step": 1129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5416792129898184e-05,
+      "loss": 0.479,
+      "step": 1130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408634002373613e-05,
+      "loss": 0.5501,
+      "step": 1131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5400470783491267e-05,
+      "loss": 0.5446,
+      "step": 1132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5392302480935503e-05,
+      "loss": 0.6212,
+      "step": 1133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538412910239545e-05,
+      "loss": 0.5296,
+      "step": 1134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375950655565046e-05,
+      "loss": 0.6451,
+      "step": 1135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367767148142968e-05,
+      "loss": 0.4573,
+      "step": 1136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359578587832674e-05,
+      "loss": 0.5265,
+      "step": 1137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351384982342378e-05,
+      "loss": 0.588,
+      "step": 1138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343186339385035e-05,
+      "loss": 0.6342,
+      "step": 1139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5334982666678347e-05,
+      "loss": 0.6418,
+      "step": 1140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5326773971944757e-05,
+      "loss": 0.594,
+      "step": 1141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531856026291142e-05,
+      "loss": 0.4931,
+      "step": 1142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5310341547310227e-05,
+      "loss": 0.5795,
+      "step": 1143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530211783287777e-05,
+      "loss": 0.6283,
+      "step": 1144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5293889127355353e-05,
+      "loss": 0.5049,
+      "step": 1145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528565543848898e-05,
+      "loss": 0.667,
+      "step": 1146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277416774029337e-05,
+      "loss": 0.5132,
+      "step": 1147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5269173141731806e-05,
+      "loss": 0.5926,
+      "step": 1148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526092454935644e-05,
+      "loss": 0.55,
+      "step": 1149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252671004667949e-05,
+      "loss": 0.7083,
+      "step": 1150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5244412515435732e-05,
+      "loss": 0.5471,
+      "step": 1151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5236149089433813e-05,
+      "loss": 0.5968,
+      "step": 1152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522788073444088e-05,
+      "loss": 0.5074,
+      "step": 1153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219607458240262e-05,
+      "loss": 0.6212,
+      "step": 1154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211329268619911e-05,
+      "loss": 0.4394,
+      "step": 1155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203046173372414e-05,
+      "loss": 0.5596,
+      "step": 1156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194758180294967e-05,
+      "loss": 0.5611,
+      "step": 1157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518646529718938e-05,
+      "loss": 0.5163,
+      "step": 1158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178167531862074e-05,
+      "loss": 0.7018,
+      "step": 1159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169864892124052e-05,
+      "loss": 0.6343,
+      "step": 1160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161557385790908e-05,
+      "loss": 0.5736,
+      "step": 1161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153245020682833e-05,
+      "loss": 0.5178,
+      "step": 1162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514492780462457e-05,
+      "loss": 0.6095,
+      "step": 1163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136605745445444e-05,
+      "loss": 0.433,
+      "step": 1164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512827885097933e-05,
+      "loss": 0.5968,
+      "step": 1165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119947129064656e-05,
+      "loss": 0.6897,
+      "step": 1166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111610587544402e-05,
+      "loss": 0.6304,
+      "step": 1167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103269234266077e-05,
+      "loss": 0.6689,
+      "step": 1168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094923077081722e-05,
+      "loss": 0.4813,
+      "step": 1169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086572123847898e-05,
+      "loss": 0.4889,
+      "step": 1170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078216382425682e-05,
+      "loss": 0.5816,
+      "step": 1171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506985586068066e-05,
+      "loss": 0.5241,
+      "step": 1172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061490566482917e-05,
+      "loss": 0.5745,
+      "step": 1173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505312050770703e-05,
+      "loss": 0.5832,
+      "step": 1174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5044745692232065e-05,
+      "loss": 0.6398,
+      "step": 1175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036366127941553e-05,
+      "loss": 0.5914,
+      "step": 1176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502798182272351e-05,
+      "loss": 0.4363,
+      "step": 1177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019592784470408e-05,
+      "loss": 0.4913,
+      "step": 1178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5011199021079174e-05,
+      "loss": 0.4863,
+      "step": 1179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002800540451188e-05,
+      "loss": 0.4985,
+      "step": 1180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994397350492262e-05,
+      "loss": 0.4832,
+      "step": 1181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498598945911265e-05,
+      "loss": 0.6375,
+      "step": 1182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977576874227026e-05,
+      "loss": 0.5274,
+      "step": 1183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4969159603754483e-05,
+      "loss": 0.5892,
+      "step": 1184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4960737655618528e-05,
+      "loss": 0.5919,
+      "step": 1185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952311037747065e-05,
+      "loss": 0.5755,
+      "step": 1186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943879758072399e-05,
+      "loss": 0.4386,
+      "step": 1187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935443824531221e-05,
+      "loss": 0.5733,
+      "step": 1188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927003245064607e-05,
+      "loss": 0.6624,
+      "step": 1189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918558027618e-05,
+      "loss": 0.5889,
+      "step": 1190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910108180141212e-05,
+      "loss": 0.5421,
+      "step": 1191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901653710588418e-05,
+      "loss": 0.5227,
+      "step": 1192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893194626918131e-05,
+      "loss": 0.5408,
+      "step": 1193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884730937093221e-05,
+      "loss": 0.5877,
+      "step": 1194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487626264908089e-05,
+      "loss": 0.3671,
+      "step": 1195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4867789770852666e-05,
+      "loss": 0.529,
+      "step": 1196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4859312310384395e-05,
+      "loss": 0.4599,
+      "step": 1197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850830275656245e-05,
+      "loss": 0.5352,
+      "step": 1198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842343674652689e-05,
+      "loss": 0.5058,
+      "step": 1199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833852515362488e-05,
+      "loss": 0.5213,
+      "step": 1200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825356805778704e-05,
+      "loss": 0.5788,
+      "step": 1201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481685655389868e-05,
+      "loss": 0.5629,
+      "step": 1202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4808351767724029e-05,
+      "loss": 0.4991,
+      "step": 1203
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799842455260644e-05,
+      "loss": 0.6147,
+      "step": 1204
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791328624518668e-05,
+      "loss": 0.5689,
+      "step": 1205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782810283512503e-05,
+      "loss": 0.4694,
+      "step": 1206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774287440260797e-05,
+      "loss": 0.6012,
+      "step": 1207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476576010278643e-05,
+      "loss": 0.4317,
+      "step": 1208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475722827911652e-05,
+      "loss": 0.5802,
+      "step": 1209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4748691977282403e-05,
+      "loss": 0.4316,
+      "step": 1210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740151205319635e-05,
+      "loss": 0.5692,
+      "step": 1211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4731605971267971e-05,
+      "loss": 0.6463,
+      "step": 1212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723056283171379e-05,
+      "loss": 0.5922,
+      "step": 1213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4714502149078009e-05,
+      "loss": 0.5954,
+      "step": 1214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705943577040205e-05,
+      "loss": 0.597,
+      "step": 1215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697380575114481e-05,
+      "loss": 0.6807,
+      "step": 1216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688813151361524e-05,
+      "loss": 0.4334,
+      "step": 1217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680241313846184e-05,
+      "loss": 0.5413,
+      "step": 1218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4671665070637467e-05,
+      "loss": 0.5884,
+      "step": 1219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4663084429808522e-05,
+      "loss": 0.5492,
+      "step": 1220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654499399436642e-05,
+      "loss": 0.6498,
+      "step": 1221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4645909987603248e-05,
+      "loss": 0.5257,
+      "step": 1222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637316202393895e-05,
+      "loss": 0.4497,
+      "step": 1223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628718051898236e-05,
+      "loss": 0.5541,
+      "step": 1224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4620115544210056e-05,
+      "loss": 0.4509,
+      "step": 1225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611508687427223e-05,
+      "loss": 0.5889,
+      "step": 1226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602897489651709e-05,
+      "loss": 0.6248,
+      "step": 1227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594281958989571e-05,
+      "loss": 0.6342,
+      "step": 1228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585662103550941e-05,
+      "loss": 0.5527,
+      "step": 1229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4577037931450024e-05,
+      "loss": 0.6366,
+      "step": 1230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568409450805094e-05,
+      "loss": 0.7101,
+      "step": 1231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559776669738472e-05,
+      "loss": 0.6646,
+      "step": 1232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551139596376532e-05,
+      "loss": 0.5505,
+      "step": 1233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542498238849686e-05,
+      "loss": 0.5906,
+      "step": 1234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4533852605292382e-05,
+      "loss": 0.4909,
+      "step": 1235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525202703843091e-05,
+      "loss": 0.6272,
+      "step": 1236
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4516548542644299e-05,
+      "loss": 0.5609,
+      "step": 1237
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507890129842509e-05,
+      "loss": 0.5511,
+      "step": 1238
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449922747358822e-05,
+      "loss": 0.5854,
+      "step": 1239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449056058203593e-05,
+      "loss": 0.4321,
+      "step": 1240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481889463344118e-05,
+      "loss": 0.5088,
+      "step": 1241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447321412567525e-05,
+      "loss": 0.6008,
+      "step": 1242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464534577195757e-05,
+      "loss": 0.6564,
+      "step": 1243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455850826076034e-05,
+      "loss": 0.6226,
+      "step": 1244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447162880490436e-05,
+      "loss": 0.5446,
+      "step": 1245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438470748617264e-05,
+      "loss": 0.5826,
+      "step": 1246
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4429774438638759e-05,
+      "loss": 0.5809,
+      "step": 1247
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4421073958741098e-05,
+      "loss": 0.6398,
+      "step": 1248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412369317114383e-05,
+      "loss": 0.6122,
+      "step": 1249
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403660521952627e-05,
+      "loss": 0.6022,
+      "step": 1250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394947581453761e-05,
+      "loss": 0.5908,
+      "step": 1251
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4386230503819619e-05,
+      "loss": 0.526,
+      "step": 1252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377509297255918e-05,
+      "loss": 0.5913,
+      "step": 1253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368783969972274e-05,
+      "loss": 0.4626,
+      "step": 1254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4360054530182175e-05,
+      "loss": 0.6162,
+      "step": 1255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351320986102983e-05,
+      "loss": 0.5624,
+      "step": 1256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342583345955927e-05,
+      "loss": 0.548,
+      "step": 1257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4333841617966084e-05,
+      "loss": 0.5177,
+      "step": 1258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4325095810362385e-05,
+      "loss": 0.6082,
+      "step": 1259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316345931377595e-05,
+      "loss": 0.6457,
+      "step": 1260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307591989248321e-05,
+      "loss": 0.6171,
+      "step": 1261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298833992214987e-05,
+      "loss": 0.4334,
+      "step": 1262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290071948521833e-05,
+      "loss": 0.5123,
+      "step": 1263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428130586641692e-05,
+      "loss": 0.4846,
+      "step": 1264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272535754152097e-05,
+      "loss": 0.6117,
+      "step": 1265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263761619983013e-05,
+      "loss": 0.5823,
+      "step": 1266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254983472169104e-05,
+      "loss": 0.5497,
+      "step": 1267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424620131897358e-05,
+      "loss": 0.5505,
+      "step": 1268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237415168663424e-05,
+      "loss": 0.4479,
+      "step": 1269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4228625029509382e-05,
+      "loss": 0.5249,
+      "step": 1270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219830909785955e-05,
+      "loss": 0.4743,
+      "step": 1271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421103281777139e-05,
+      "loss": 0.5277,
+      "step": 1272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202230761747671e-05,
+      "loss": 0.4926,
+      "step": 1273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419342475000052e-05,
+      "loss": 0.549,
+      "step": 1274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184614790819375e-05,
+      "loss": 0.6296,
+      "step": 1275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175800892497397e-05,
+      "loss": 0.5331,
+      "step": 1276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416698306333145e-05,
+      "loss": 0.6601,
+      "step": 1277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4158161311622098e-05,
+      "loss": 0.6018,
+      "step": 1278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4149335645673603e-05,
+      "loss": 0.5481,
+      "step": 1279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4140506073793908e-05,
+      "loss": 0.5885,
+      "step": 1280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131672604294627e-05,
+      "loss": 0.5023,
+      "step": 1281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122835245491053e-05,
+      "loss": 0.5452,
+      "step": 1282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411399400570214e-05,
+      "loss": 0.5266,
+      "step": 1283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105148893250486e-05,
+      "loss": 0.4613,
+      "step": 1284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096299916462339e-05,
+      "loss": 0.6174,
+      "step": 1285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4087447083667589e-05,
+      "loss": 0.5268,
+      "step": 1286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407859040319975e-05,
+      "loss": 0.4977,
+      "step": 1287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406972988339596e-05,
+      "loss": 0.4881,
+      "step": 1288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060865532596972e-05,
+      "loss": 0.6436,
+      "step": 1289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051997359147145e-05,
+      "loss": 0.5491,
+      "step": 1290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4043125371394435e-05,
+      "loss": 0.6691,
+      "step": 1291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034249577690391e-05,
+      "loss": 0.5137,
+      "step": 1292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402536998639014e-05,
+      "loss": 0.5019,
+      "step": 1293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016486605852391e-05,
+      "loss": 0.4637,
+      "step": 1294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4007599444439417e-05,
+      "loss": 0.5507,
+      "step": 1295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998708510517039e-05,
+      "loss": 0.55,
+      "step": 1296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3989813812454652e-05,
+      "loss": 0.5872,
+      "step": 1297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3980915358625174e-05,
+      "loss": 0.5131,
+      "step": 1298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397201315740507e-05,
+      "loss": 0.575,
+      "step": 1299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3963107217174322e-05,
+      "loss": 0.5933,
+      "step": 1300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954197546316448e-05,
+      "loss": 0.4937,
+      "step": 1301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945284153218461e-05,
+      "loss": 0.5856,
+      "step": 1302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936367046270888e-05,
+      "loss": 0.4832,
+      "step": 1303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927446233867749e-05,
+      "loss": 0.551,
+      "step": 1304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391852172440655e-05,
+      "loss": 0.5937,
+      "step": 1305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390959352628828e-05,
+      "loss": 0.5063,
+      "step": 1306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3900661647917401e-05,
+      "loss": 0.4647,
+      "step": 1307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891726097701836e-05,
+      "loss": 0.5895,
+      "step": 1308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388278688405297e-05,
+      "loss": 0.5399,
+      "step": 1309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873844015385626e-05,
+      "loss": 0.489,
+      "step": 1310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386489750011808e-05,
+      "loss": 0.6412,
+      "step": 1311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855947346672034e-05,
+      "loss": 0.5171,
+      "step": 1312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846993563472614e-05,
+      "loss": 0.5419,
+      "step": 1313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838036158948367e-05,
+      "loss": 0.4719,
+      "step": 1314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3829075141531244e-05,
+      "loss": 0.4875,
+      "step": 1315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820110519656606e-05,
+      "loss": 0.6015,
+      "step": 1316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811142301763194e-05,
+      "loss": 0.5675,
+      "step": 1317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802170496293144e-05,
+      "loss": 0.5889,
+      "step": 1318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793195111691967e-05,
+      "loss": 0.5666,
+      "step": 1319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3784216156408535e-05,
+      "loss": 0.5826,
+      "step": 1320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3775233638895094e-05,
+      "loss": 0.6722,
+      "step": 1321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3766247567607236e-05,
+      "loss": 0.5355,
+      "step": 1322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3757257951003902e-05,
+      "loss": 0.5668,
+      "step": 1323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3748264797547363e-05,
+      "loss": 0.6478,
+      "step": 1324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373926811570323e-05,
+      "loss": 0.6413,
+      "step": 1325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3730267913940426e-05,
+      "loss": 0.5315,
+      "step": 1326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721264200731196e-05,
+      "loss": 0.5896,
+      "step": 1327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712256984551083e-05,
+      "loss": 0.5533,
+      "step": 1328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703246273878929e-05,
+      "loss": 0.524,
+      "step": 1329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694232077196868e-05,
+      "loss": 0.4608,
+      "step": 1330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3685214402990313e-05,
+      "loss": 0.5469,
+      "step": 1331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676193259747954e-05,
+      "loss": 0.6955,
+      "step": 1332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667168655961745e-05,
+      "loss": 0.472,
+      "step": 1333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3658140600126892e-05,
+      "loss": 0.5545,
+      "step": 1334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3649109100741863e-05,
+      "loss": 0.5501,
+      "step": 1335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3640074166308352e-05,
+      "loss": 0.5138,
+      "step": 1336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36310358053313e-05,
+      "loss": 0.3948,
+      "step": 1337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621994026318868e-05,
+      "loss": 0.6288,
+      "step": 1338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3612948837782429e-05,
+      "loss": 0.488,
+      "step": 1339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603900248236578e-05,
+      "loss": 0.652,
+      "step": 1340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594848266199103e-05,
+      "loss": 0.5972,
+      "step": 1341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3585792900190986e-05,
+      "loss": 0.5861,
+      "step": 1342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3576734158736398e-05,
+      "loss": 0.6183,
+      "step": 1343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567672050362679e-05,
+      "loss": 0.5218,
+      "step": 1344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558606583600348e-05,
+      "loss": 0.4503,
+      "step": 1345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3549537766983085e-05,
+      "loss": 0.5021,
+      "step": 1346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3540465609047715e-05,
+      "loss": 0.4972,
+      "step": 1347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3531390118334216e-05,
+      "loss": 0.5317,
+      "step": 1348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35223113033857e-05,
+      "loss": 0.6279,
+      "step": 1349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3513229172748413e-05,
+      "loss": 0.5368,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350414373497171e-05,
+      "loss": 0.5858,
+      "step": 1351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495054998608075e-05,
+      "loss": 0.4067,
+      "step": 1352
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485962972213087e-05,
+      "loss": 0.5231,
+      "step": 1353
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476867664345421e-05,
+      "loss": 0.6598,
+      "step": 1354
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467769083566851e-05,
+      "loss": 0.6266,
+      "step": 1355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345866723844222e-05,
+      "loss": 0.6451,
+      "step": 1356
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3449562137539452e-05,
+      "loss": 0.5579,
+      "step": 1357
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344045378942953e-05,
+      "loss": 0.5624,
+      "step": 1358
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34313422026865e-05,
+      "loss": 0.5952,
+      "step": 1359
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3422227385887449e-05,
+      "loss": 0.595,
+      "step": 1360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3413109347612516e-05,
+      "loss": 0.511,
+      "step": 1361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340398809644486e-05,
+      "loss": 0.5344,
+      "step": 1362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3394863640970671e-05,
+      "loss": 0.6125,
+      "step": 1363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3385735989779156e-05,
+      "loss": 0.5406,
+      "step": 1364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376605151462522e-05,
+      "loss": 0.471,
+      "step": 1365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367471134615993e-05,
+      "loss": 0.4236,
+      "step": 1366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335833394783777e-05,
+      "loss": 0.5429,
+      "step": 1367
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349193599729037e-05,
+      "loss": 0.5326,
+      "step": 1368
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334005009889397e-05,
+      "loss": 0.4503,
+      "step": 1369
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330903453939693e-05,
+      "loss": 0.5319,
+      "step": 1370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321753673476301e-05,
+      "loss": 0.627,
+      "step": 1371
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331260076611684e-05,
+      "loss": 0.4156,
+      "step": 1372
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3303444740477303e-05,
+      "loss": 0.5429,
+      "step": 1373
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294285605176609e-05,
+      "loss": 0.484,
+      "step": 1374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285123368836608e-05,
+      "loss": 0.5133,
+      "step": 1375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3275958040082068e-05,
+      "loss": 0.5557,
+      "step": 1376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3266789627540674e-05,
+      "loss": 0.5875,
+      "step": 1377
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325761813984301e-05,
+      "loss": 0.5507,
+      "step": 1378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248443585622553e-05,
+      "loss": 0.6345,
+      "step": 1379
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239265973515662e-05,
+      "loss": 0.4525,
+      "step": 1380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230085312161588e-05,
+      "loss": 0.562,
+      "step": 1381
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220901610202446e-05,
+      "loss": 0.4425,
+      "step": 1382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3211714876283208e-05,
+      "loss": 0.5752,
+      "step": 1383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202525119051705e-05,
+      "loss": 0.6677,
+      "step": 1384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3193332347158615e-05,
+      "loss": 0.5729,
+      "step": 1385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318413656925745e-05,
+      "loss": 0.607,
+      "step": 1386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3174937794004559e-05,
+      "loss": 0.5344,
+      "step": 1387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31657360300591e-05,
+      "loss": 0.5126,
+      "step": 1388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3156531286083054e-05,
+      "loss": 0.5177,
+      "step": 1389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147323570741206e-05,
+      "loss": 0.4042,
+      "step": 1390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313811289270114e-05,
+      "loss": 0.4953,
+      "step": 1391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3128899260633218e-05,
+      "loss": 0.5793,
+      "step": 1392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31196826832106e-05,
+      "loss": 0.5539,
+      "step": 1393
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110463169109203e-05,
+      "loss": 0.5584,
+      "step": 1394
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3101240727007713e-05,
+      "loss": 0.489,
+      "step": 1395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309201536558758e-05,
+      "loss": 0.4761,
+      "step": 1396
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082787093532992e-05,
+      "loss": 0.586,
+      "step": 1397
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307355591953088e-05,
+      "loss": 0.5171,
+      "step": 1398
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064321852270911e-05,
+      "loss": 0.4342,
+      "step": 1399
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055084900445469e-05,
+      "loss": 0.4914,
+      "step": 1400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045845072749659e-05,
+      "loss": 0.5227,
+      "step": 1401
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036602377881286e-05,
+      "loss": 0.5209,
+      "step": 1402
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302735682454087e-05,
+      "loss": 0.5796,
+      "step": 1403
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018108421431594e-05,
+      "loss": 0.5938,
+      "step": 1404
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3008857177259348e-05,
+      "loss": 0.5992,
+      "step": 1405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299960310073269e-05,
+      "loss": 0.4581,
+      "step": 1406
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990346200562842e-05,
+      "loss": 0.6171,
+      "step": 1407
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2981086485463683e-05,
+      "loss": 0.5646,
+      "step": 1408
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2971823964151744e-05,
+      "loss": 0.4479,
+      "step": 1409
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962558645346195e-05,
+      "loss": 0.4551,
+      "step": 1410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953290537768846e-05,
+      "loss": 0.4921,
+      "step": 1411
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944019650144122e-05,
+      "loss": 0.5237,
+      "step": 1412
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934745991199074e-05,
+      "loss": 0.4408,
+      "step": 1413
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2925469569663356e-05,
+      "loss": 0.5025,
+      "step": 1414
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2916190394269228e-05,
+      "loss": 0.3863,
+      "step": 1415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906908473751537e-05,
+      "loss": 0.5699,
+      "step": 1416
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897623816847715e-05,
+      "loss": 0.6132,
+      "step": 1417
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888336432297774e-05,
+      "loss": 0.4848,
+      "step": 1418
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287904632884429e-05,
+      "loss": 0.5926,
+      "step": 1419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869753515232395e-05,
+      "loss": 0.4746,
+      "step": 1420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860458000209781e-05,
+      "loss": 0.4916,
+      "step": 1421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2851159792526678e-05,
+      "loss": 0.4894,
+      "step": 1422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841858900935847e-05,
+      "loss": 0.512,
+      "step": 1423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2832555334192582e-05,
+      "loss": 0.642,
+      "step": 1424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2823249101054695e-05,
+      "loss": 0.4678,
+      "step": 1425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813940210282503e-05,
+      "loss": 0.5704,
+      "step": 1426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280462867063883e-05,
+      "loss": 0.5065,
+      "step": 1427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795314490888986e-05,
+      "loss": 0.5647,
+      "step": 1428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785997679800772e-05,
+      "loss": 0.4939,
+      "step": 1429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776678246144467e-05,
+      "loss": 0.541,
+      "step": 1430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767356198692815e-05,
+      "loss": 0.5497,
+      "step": 1431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2758031546221026e-05,
+      "loss": 0.5318,
+      "step": 1432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748704297506755e-05,
+      "loss": 0.6613,
+      "step": 1433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739374461330106e-05,
+      "loss": 0.4853,
+      "step": 1434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730042046473616e-05,
+      "loss": 0.4729,
+      "step": 1435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272070706172225e-05,
+      "loss": 0.6324,
+      "step": 1436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711369515863394e-05,
+      "loss": 0.4914,
+      "step": 1437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702029417686845e-05,
+      "loss": 0.4448,
+      "step": 1438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692686775984797e-05,
+      "loss": 0.5271,
+      "step": 1439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683341599551846e-05,
+      "loss": 0.4646,
+      "step": 1440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673993897184967e-05,
+      "loss": 0.5219,
+      "step": 1441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2664643677683519e-05,
+      "loss": 0.6189,
+      "step": 1442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655290949849227e-05,
+      "loss": 0.4929,
+      "step": 1443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645935722486176e-05,
+      "loss": 0.6157,
+      "step": 1444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636578004400807e-05,
+      "loss": 0.6165,
+      "step": 1445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627217804401908e-05,
+      "loss": 0.4954,
+      "step": 1446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2617855131300593e-05,
+      "loss": 0.6255,
+      "step": 1447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2608489993910315e-05,
+      "loss": 0.4193,
+      "step": 1448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2599122401046842e-05,
+      "loss": 0.6346,
+      "step": 1449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2589752361528254e-05,
+      "loss": 0.4636,
+      "step": 1450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580379884174935e-05,
+      "loss": 0.5775,
+      "step": 1451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257100497780956e-05,
+      "loss": 0.6198,
+      "step": 1452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561627651257096e-05,
+      "loss": 0.6254,
+      "step": 1453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552247913344783e-05,
+      "loss": 0.4878,
+      "step": 1454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542865772902136e-05,
+      "loss": 0.5805,
+      "step": 1455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533481238760925e-05,
+      "loss": 0.5991,
+      "step": 1456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524094319755183e-05,
+      "loss": 0.4524,
+      "step": 1457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2514705024721181e-05,
+      "loss": 0.5267,
+      "step": 1458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505313362497424e-05,
+      "loss": 0.5432,
+      "step": 1459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2495919341924653e-05,
+      "loss": 0.4977,
+      "step": 1460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486522971845827e-05,
+      "loss": 0.5614,
+      "step": 1461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2477124261106109e-05,
+      "loss": 0.5663,
+      "step": 1462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2467723218552878e-05,
+      "loss": 0.5588,
+      "step": 1463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458319853035695e-05,
+      "loss": 0.5009,
+      "step": 1464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448914173406318e-05,
+      "loss": 0.5149,
+      "step": 1465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.243950618851868e-05,
+      "loss": 0.5863,
+      "step": 1466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2430095907228883e-05,
+      "loss": 0.5436,
+      "step": 1467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420683338395188e-05,
+      "loss": 0.4916,
+      "step": 1468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2411268490878017e-05,
+      "loss": 0.5513,
+      "step": 1469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401851373539929e-05,
+      "loss": 0.4569,
+      "step": 1470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392431995245624e-05,
+      "loss": 0.4484,
+      "step": 1471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383010364861931e-05,
+      "loss": 0.4135,
+      "step": 1472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23735864912578e-05,
+      "loss": 0.4828,
+      "step": 1473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236416038330428e-05,
+      "loss": 0.4859,
+      "step": 1474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2354732049874546e-05,
+      "loss": 0.4586,
+      "step": 1475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2345301499843849e-05,
+      "loss": 0.5372,
+      "step": 1476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335868742089533e-05,
+      "loss": 0.4147,
+      "step": 1477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2326433785491021e-05,
+      "loss": 0.6714,
+      "step": 1478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316996638929805e-05,
+      "loss": 0.5096,
+      "step": 1479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307557311289437e-05,
+      "loss": 0.5379,
+      "step": 1480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298115811455527e-05,
+      "loss": 0.4897,
+      "step": 1481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228867214831572e-05,
+      "loss": 0.5141,
+      "step": 1482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227922633075971e-05,
+      "loss": 0.5534,
+      "step": 1483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269778367679212e-05,
+      "loss": 0.5014,
+      "step": 1484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2260328267967963e-05,
+      "loss": 0.5376,
+      "step": 1485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2250876040521705e-05,
+      "loss": 0.4658,
+      "step": 1486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241421694238192e-05,
+      "loss": 0.5872,
+      "step": 1487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2231965238017163e-05,
+      "loss": 0.5344,
+      "step": 1488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222506680760351e-05,
+      "loss": 0.5151,
+      "step": 1489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2213046031371467e-05,
+      "loss": 0.5041,
+      "step": 1490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203583298756188e-05,
+      "loss": 0.5548,
+      "step": 1491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194118491822149e-05,
+      "loss": 0.5584,
+      "step": 1492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184651619478945e-05,
+      "loss": 0.5406,
+      "step": 1493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2175182690638108e-05,
+      "loss": 0.6101,
+      "step": 1494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165711714213114e-05,
+      "loss": 0.6036,
+      "step": 1495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156238699119358e-05,
+      "loss": 0.4181,
+      "step": 1496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146763654274158e-05,
+      "loss": 0.4172,
+      "step": 1497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2137286588596746e-05,
+      "loss": 0.4537,
+      "step": 1498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127807511008252e-05,
+      "loss": 0.5571,
+      "step": 1499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118326430431698e-05,
+      "loss": 0.5416,
+      "step": 1500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2108843355791994e-05,
+      "loss": 0.5992,
+      "step": 1501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099358296015933e-05,
+      "loss": 0.5714,
+      "step": 1502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2089871260032164e-05,
+      "loss": 0.5034,
+      "step": 1503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208038225677121e-05,
+      "loss": 0.4634,
+      "step": 1504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2070891295165434e-05,
+      "loss": 0.5249,
+      "step": 1505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206139838414905e-05,
+      "loss": 0.6044,
+      "step": 1506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051903532658104e-05,
+      "loss": 0.4504,
+      "step": 1507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204240674963047e-05,
+      "loss": 0.4761,
+      "step": 1508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203290804400584e-05,
+      "loss": 0.5016,
+      "step": 1509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023407424725715e-05,
+      "loss": 0.4185,
+      "step": 1510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2013904900733402e-05,
+      "loss": 0.5402,
+      "step": 1511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2004400480973988e-05,
+      "loss": 0.4874,
+      "step": 1512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994894174394358e-05,
+      "loss": 0.5645,
+      "step": 1513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198538598994317e-05,
+      "loss": 0.5331,
+      "step": 1514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1975875936570842e-05,
+      "loss": 0.526,
+      "step": 1515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966364023229566e-05,
+      "loss": 0.5486,
+      "step": 1516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1956850258873274e-05,
+      "loss": 0.3877,
+      "step": 1517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194733465245764e-05,
+      "loss": 0.5503,
+      "step": 1518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1937817212940077e-05,
+      "loss": 0.5533,
+      "step": 1519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192829794927972e-05,
+      "loss": 0.5626,
+      "step": 1520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918776870437423e-05,
+      "loss": 0.546,
+      "step": 1521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909253985375748e-05,
+      "loss": 0.3593,
+      "step": 1522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189972930305896e-05,
+      "loss": 0.636,
+      "step": 1523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189020283245301e-05,
+      "loss": 0.5757,
+      "step": 1524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1880674582525537e-05,
+      "loss": 0.5319,
+      "step": 1525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1871144562245853e-05,
+      "loss": 0.4621,
+      "step": 1526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861612780584938e-05,
+      "loss": 0.5196,
+      "step": 1527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852079246515431e-05,
+      "loss": 0.5117,
+      "step": 1528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842543969011616e-05,
+      "loss": 0.5988,
+      "step": 1529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183300695704942e-05,
+      "loss": 0.5836,
+      "step": 1530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823468219606407e-05,
+      "loss": 0.5548,
+      "step": 1531
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813927765661758e-05,
+      "loss": 0.4707,
+      "step": 1532
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1804385604196276e-05,
+      "loss": 0.4658,
+      "step": 1533
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794841744192371e-05,
+      "loss": 0.5904,
+      "step": 1534
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178529619463405e-05,
+      "loss": 0.5959,
+      "step": 1535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775748964506908e-05,
+      "loss": 0.4742,
+      "step": 1536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1766200062798126e-05,
+      "loss": 0.537,
+      "step": 1537
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1756649498496453e-05,
+      "loss": 0.4655,
+      "step": 1538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1747097280592212e-05,
+      "loss": 0.5833,
+      "step": 1539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737543418077274e-05,
+      "loss": 0.3917,
+      "step": 1540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727987919945065e-05,
+      "loss": 0.5265,
+      "step": 1541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1718430795190548e-05,
+      "loss": 0.5499,
+      "step": 1542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1708872052810211e-05,
+      "loss": 0.5742,
+      "step": 1543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699311701802079e-05,
+      "loss": 0.5934,
+      "step": 1544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689749751165674e-05,
+      "loss": 0.6171,
+      "step": 1545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1680186209902037e-05,
+      "loss": 0.5769,
+      "step": 1546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670621087013703e-05,
+      "loss": 0.5033,
+      "step": 1547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166105439150469e-05,
+      "loss": 0.551,
+      "step": 1548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651486132380503e-05,
+      "loss": 0.5658,
+      "step": 1549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164191631864812e-05,
+      "loss": 0.5155,
+      "step": 1550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1632344959315976e-05,
+      "loss": 0.4809,
+      "step": 1551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1622772063393963e-05,
+      "loss": 0.5587,
+      "step": 1552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1613197639893423e-05,
+      "loss": 0.4179,
+      "step": 1553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1603621697827134e-05,
+      "loss": 0.4428,
+      "step": 1554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1594044246209304e-05,
+      "loss": 0.5217,
+      "step": 1555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584465294055558e-05,
+      "loss": 0.6361,
+      "step": 1556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574884850382937e-05,
+      "loss": 0.4417,
+      "step": 1557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565302924209885e-05,
+      "loss": 0.6105,
+      "step": 1558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1555719524556248e-05,
+      "loss": 0.5323,
+      "step": 1559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546134660443248e-05,
+      "loss": 0.4316,
+      "step": 1560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1536548340893492e-05,
+      "loss": 0.4621,
+      "step": 1561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1526960574930954e-05,
+      "loss": 0.5468,
+      "step": 1562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1517371371580974e-05,
+      "loss": 0.5505,
+      "step": 1563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1507780739870242e-05,
+      "loss": 0.6466,
+      "step": 1564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149818868882679e-05,
+      "loss": 0.6105,
+      "step": 1565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488595227479995e-05,
+      "loss": 0.6162,
+      "step": 1566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479000364860551e-05,
+      "loss": 0.6582,
+      "step": 1567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1469404110000479e-05,
+      "loss": 0.5212,
+      "step": 1568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459806471933104e-05,
+      "loss": 0.5567,
+      "step": 1569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145020745969306e-05,
+      "loss": 0.6478,
+      "step": 1570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1440607082316269e-05,
+      "loss": 0.5504,
+      "step": 1571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431005348839938e-05,
+      "loss": 0.5664,
+      "step": 1572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1421402268302552e-05,
+      "loss": 0.5869,
+      "step": 1573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411797849743871e-05,
+      "loss": 0.4913,
+      "step": 1574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402192102204905e-05,
+      "loss": 0.5501,
+      "step": 1575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392585034727914e-05,
+      "loss": 0.4763,
+      "step": 1576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382976656356412e-05,
+      "loss": 0.6064,
+      "step": 1577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1373366976135135e-05,
+      "loss": 0.4748,
+      "step": 1578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1363756003110048e-05,
+      "loss": 0.4722,
+      "step": 1579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354143746328337e-05,
+      "loss": 0.4845,
+      "step": 1580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1344530214838394e-05,
+      "loss": 0.5362,
+      "step": 1581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1334915417689807e-05,
+      "loss": 0.6133,
+      "step": 1582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325299363933363e-05,
+      "loss": 0.669,
+      "step": 1583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1315682062621025e-05,
+      "loss": 0.4991,
+      "step": 1584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1306063522805934e-05,
+      "loss": 0.6133,
+      "step": 1585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1296443753542397e-05,
+      "loss": 0.4545,
+      "step": 1586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1286822763885876e-05,
+      "loss": 0.6176,
+      "step": 1587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277200562892985e-05,
+      "loss": 0.5638,
+      "step": 1588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1267577159621472e-05,
+      "loss": 0.501,
+      "step": 1589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257952563130228e-05,
+      "loss": 0.5305,
+      "step": 1590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248326782479252e-05,
+      "loss": 0.4944,
+      "step": 1591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238699826729673e-05,
+      "loss": 0.6549,
+      "step": 1592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1229071704943719e-05,
+      "loss": 0.5206,
+      "step": 1593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219442426184712e-05,
+      "loss": 0.6658,
+      "step": 1594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1209811999517073e-05,
+      "loss": 0.5446,
+      "step": 1595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1200180434006289e-05,
+      "loss": 0.5681,
+      "step": 1596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1190547738718935e-05,
+      "loss": 0.4395,
+      "step": 1597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1180913922722638e-05,
+      "loss": 0.6209,
+      "step": 1598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171278995086084e-05,
+      "loss": 0.6148,
+      "step": 1599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116164296487901e-05,
+      "loss": 0.4103,
+      "step": 1600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152005841172182e-05,
+      "loss": 0.5186,
+      "step": 1601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142367633037402e-05,
+      "loss": 0.5694,
+      "step": 1602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132728349547488e-05,
+      "loss": 0.5954,
+      "step": 1603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123087999776276e-05,
+      "loss": 0.4506,
+      "step": 1604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113446592798602e-05,
+      "loss": 0.5368,
+      "step": 1605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103804137690298e-05,
+      "loss": 0.4682,
+      "step": 1606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109416064352818e-05,
+      "loss": 0.5217,
+      "step": 1607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1084516119390045e-05,
+      "loss": 0.4329,
+      "step": 1608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074870574354662e-05,
+      "loss": 0.5748,
+      "step": 1609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065224017501755e-05,
+      "loss": 0.5551,
+      "step": 1610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055576457912006e-05,
+      "loss": 0.6276,
+      "step": 1611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045927904667035e-05,
+      "loss": 0.5116,
+      "step": 1612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1036278366849407e-05,
+      "loss": 0.5358,
+      "step": 1613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026627853542601e-05,
+      "loss": 0.6445,
+      "step": 1614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1016976373831026e-05,
+      "loss": 0.4735,
+      "step": 1615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007323936799988e-05,
+      "loss": 0.5648,
+      "step": 1616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997670551535711e-05,
+      "loss": 0.6034,
+      "step": 1617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0988016227125298e-05,
+      "loss": 0.4771,
+      "step": 1618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0978360972656743e-05,
+      "loss": 0.5164,
+      "step": 1619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096870479721891e-05,
+      "loss": 0.4874,
+      "step": 1620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959047709901533e-05,
+      "loss": 0.5215,
+      "step": 1621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0949389719795207e-05,
+      "loss": 0.5451,
+      "step": 1622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939730835991371e-05,
+      "loss": 0.4989,
+      "step": 1623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0930071067582307e-05,
+      "loss": 0.5965,
+      "step": 1624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920410423661137e-05,
+      "loss": 0.6091,
+      "step": 1625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0910748913321798e-05,
+      "loss": 0.4835,
+      "step": 1626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901086545659046e-05,
+      "loss": 0.6287,
+      "step": 1627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0891423329768443e-05,
+      "loss": 0.5141,
+      "step": 1628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088175927474635e-05,
+      "loss": 0.5482,
+      "step": 1629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872094389689922e-05,
+      "loss": 0.5872,
+      "step": 1630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862428683697093e-05,
+      "loss": 0.3915,
+      "step": 1631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852762165866557e-05,
+      "loss": 0.4087,
+      "step": 1632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843094845297795e-05,
+      "loss": 0.396,
+      "step": 1633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833426731091028e-05,
+      "loss": 0.5215,
+      "step": 1634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0823757832347228e-05,
+      "loss": 0.6428,
+      "step": 1635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814088158168108e-05,
+      "loss": 0.5838,
+      "step": 1636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0804417717656106e-05,
+      "loss": 0.4848,
+      "step": 1637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794746519914386e-05,
+      "loss": 0.5321,
+      "step": 1638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785074574046824e-05,
+      "loss": 0.6477,
+      "step": 1639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0775401889157994e-05,
+      "loss": 0.6279,
+      "step": 1640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765728474353178e-05,
+      "loss": 0.6694,
+      "step": 1641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0756054338738331e-05,
+      "loss": 0.5318,
+      "step": 1642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746379491420103e-05,
+      "loss": 0.6164,
+      "step": 1643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0736703941505798e-05,
+      "loss": 0.6076,
+      "step": 1644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0727027698103386e-05,
+      "loss": 0.4493,
+      "step": 1645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0717350770321493e-05,
+      "loss": 0.4055,
+      "step": 1646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707673167269392e-05,
+      "loss": 0.5621,
+      "step": 1647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697994898056982e-05,
+      "loss": 0.5841,
+      "step": 1648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688315971794798e-05,
+      "loss": 0.5237,
+      "step": 1649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.067863639759399e-05,
+      "loss": 0.5417,
+      "step": 1650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066895618456632e-05,
+      "loss": 0.5955,
+      "step": 1651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659275341824145e-05,
+      "loss": 0.5028,
+      "step": 1652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0649593878480419e-05,
+      "loss": 0.5883,
+      "step": 1653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0639911803648687e-05,
+      "loss": 0.457,
+      "step": 1654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0630229126443059e-05,
+      "loss": 0.5437,
+      "step": 1655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0620545855978212e-05,
+      "loss": 0.6506,
+      "step": 1656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061086200136939e-05,
+      "loss": 0.5436,
+      "step": 1657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601177571732382e-05,
+      "loss": 0.458,
+      "step": 1658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591492576183517e-05,
+      "loss": 0.6221,
+      "step": 1659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0581807023839661e-05,
+      "loss": 0.5047,
+      "step": 1660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0572120923818201e-05,
+      "loss": 0.4183,
+      "step": 1661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056243428523704e-05,
+      "loss": 0.5275,
+      "step": 1662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552747117214586e-05,
+      "loss": 0.4502,
+      "step": 1663
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054305942886975e-05,
+      "loss": 0.5867,
+      "step": 1664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533371229321926e-05,
+      "loss": 0.5354,
+      "step": 1665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0523682527690999e-05,
+      "loss": 0.4966,
+      "step": 1666
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0513993333097318e-05,
+      "loss": 0.6719,
+      "step": 1667
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05043036546617e-05,
+      "loss": 0.3987,
+      "step": 1668
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494613501505414e-05,
+      "loss": 0.5483,
+      "step": 1669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484922882750186e-05,
+      "loss": 0.5945,
+      "step": 1670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0475231807518163e-05,
+      "loss": 0.5201,
+      "step": 1671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0465540284931939e-05,
+      "loss": 0.4592,
+      "step": 1672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455848324114517e-05,
+      "loss": 0.5384,
+      "step": 1673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0446155934189317e-05,
+      "loss": 0.6019,
+      "step": 1674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0436463124280165e-05,
+      "loss": 0.5374,
+      "step": 1675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042676990351128e-05,
+      "loss": 0.5237,
+      "step": 1676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041707628100727e-05,
+      "loss": 0.5625,
+      "step": 1677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407382265893117e-05,
+      "loss": 0.6133,
+      "step": 1678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397687867294177e-05,
+      "loss": 0.6429,
+      "step": 1679
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0387993094336162e-05,
+      "loss": 0.5276,
+      "step": 1680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037829795614514e-05,
+      "loss": 0.516,
+      "step": 1681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036860246184752e-05,
+      "loss": 0.6549,
+      "step": 1682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358906620570052e-05,
+      "loss": 0.5765,
+      "step": 1683
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034921044143981e-05,
+      "loss": 0.5241,
+      "step": 1684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033951393358418e-05,
+      "loss": 0.4418,
+      "step": 1685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0329817106130866e-05,
+      "loss": 0.4701,
+      "step": 1686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.032011996820787e-05,
+      "loss": 0.5123,
+      "step": 1687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310422528943482e-05,
+      "loss": 0.5156,
+      "step": 1688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0300724797466282e-05,
+      "loss": 0.5978,
+      "step": 1689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291026782905121e-05,
+      "loss": 0.6034,
+      "step": 1690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281328494389115e-05,
+      "loss": 0.4956,
+      "step": 1691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0271629941047647e-05,
+      "loss": 0.5032,
+      "step": 1692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261931132010336e-05,
+      "loss": 0.6594,
+      "step": 1693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252232076407057e-05,
+      "loss": 0.5117,
+      "step": 1694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242532783367898e-05,
+      "loss": 0.5879,
+      "step": 1695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023283326202319e-05,
+      "loss": 0.5799,
+      "step": 1696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0223133521503465e-05,
+      "loss": 0.6023,
+      "step": 1697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0213433570939471e-05,
+      "loss": 0.4979,
+      "step": 1698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203733419462146e-05,
+      "loss": 0.5955,
+      "step": 1699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194033076202616e-05,
+      "loss": 0.6395,
+      "step": 1700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184332550292199e-05,
+      "loss": 0.5224,
+      "step": 1701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174631850862377e-05,
+      "loss": 0.476,
+      "step": 1702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164930987044797e-05,
+      "loss": 0.598,
+      "step": 1703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155229967971253e-05,
+      "loss": 0.5338,
+      "step": 1704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145528802773697e-05,
+      "loss": 0.505,
+      "step": 1705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135827500584213e-05,
+      "loss": 0.616,
+      "step": 1706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0126126070535013e-05,
+      "loss": 0.4713,
+      "step": 1707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011642452175843e-05,
+      "loss": 0.5018,
+      "step": 1708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106722863386913e-05,
+      "loss": 0.5529,
+      "step": 1709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0097021104553011e-05,
+      "loss": 0.4404,
+      "step": 1710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087319254389366e-05,
+      "loss": 0.5561,
+      "step": 1711
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077617322028706e-05,
+      "loss": 0.5458,
+      "step": 1712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067915316603838e-05,
+      "loss": 0.4525,
+      "step": 1713
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058213247247639e-05,
+      "loss": 0.5595,
+      "step": 1714
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048511123093046e-05,
+      "loss": 0.5516,
+      "step": 1715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003880895327304e-05,
+      "loss": 0.4061,
+      "step": 1716
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029106746920656e-05,
+      "loss": 0.5078,
+      "step": 1717
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001940451316896e-05,
+      "loss": 0.5456,
+      "step": 1718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000970226115104e-05,
+      "loss": 0.5373,
+      "step": 1719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.6005,
+      "step": 1720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990297738848966e-06,
+      "loss": 0.4524,
+      "step": 1721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980595486831043e-06,
+      "loss": 0.3452,
+      "step": 1722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970893253079345e-06,
+      "loss": 0.575,
+      "step": 1723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961191046726963e-06,
+      "loss": 0.522,
+      "step": 1724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.951488876906959e-06,
+      "loss": 0.6229,
+      "step": 1725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941786752752361e-06,
+      "loss": 0.5494,
+      "step": 1726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932084683396166e-06,
+      "loss": 0.5959,
+      "step": 1727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922382677971296e-06,
+      "loss": 0.5233,
+      "step": 1728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912680745610638e-06,
+      "loss": 0.5906,
+      "step": 1729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90297889544699e-06,
+      "loss": 0.5145,
+      "step": 1730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.893277136613088e-06,
+      "loss": 0.4225,
+      "step": 1731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883575478241575e-06,
+      "loss": 0.5025,
+      "step": 1732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87387392946499e-06,
+      "loss": 0.5976,
+      "step": 1733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864172499415794e-06,
+      "loss": 0.464,
+      "step": 1734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854471197226307e-06,
+      "loss": 0.4944,
+      "step": 1735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844770032028752e-06,
+      "loss": 0.4351,
+      "step": 1736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835069012955207e-06,
+      "loss": 0.6253,
+      "step": 1737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825368149137626e-06,
+      "loss": 0.4489,
+      "step": 1738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815667449707801e-06,
+      "loss": 0.4999,
+      "step": 1739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805966923797388e-06,
+      "loss": 0.5797,
+      "step": 1740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79626658053786e-06,
+      "loss": 0.6045,
+      "step": 1741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786566429060532e-06,
+      "loss": 0.4884,
+      "step": 1742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.776866478496535e-06,
+      "loss": 0.5263,
+      "step": 1743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767166737976814e-06,
+      "loss": 0.4987,
+      "step": 1744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757467216632106e-06,
+      "loss": 0.4758,
+      "step": 1745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747767923592948e-06,
+      "loss": 0.4836,
+      "step": 1746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738068867989667e-06,
+      "loss": 0.4326,
+      "step": 1747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728370058952356e-06,
+      "loss": 0.5649,
+      "step": 1748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718671505610888e-06,
+      "loss": 0.5098,
+      "step": 1749
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708973217094884e-06,
+      "loss": 0.5438,
+      "step": 1750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.699275202533721e-06,
+      "loss": 0.4879,
+      "step": 1751
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689577471056518e-06,
+      "loss": 0.5506,
+      "step": 1752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679880031792134e-06,
+      "loss": 0.5024,
+      "step": 1753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670182893869133e-06,
+      "loss": 0.4075,
+      "step": 1754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660486066415821e-06,
+      "loss": 0.4741,
+      "step": 1755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650789558560194e-06,
+      "loss": 0.5218,
+      "step": 1756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64109337942995e-06,
+      "loss": 0.5447,
+      "step": 1757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.631397538152482e-06,
+      "loss": 0.5744,
+      "step": 1758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621702043854863e-06,
+      "loss": 0.4535,
+      "step": 1759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.612006905663843e-06,
+      "loss": 0.5055,
+      "step": 1760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602312132705825e-06,
+      "loss": 0.6092,
+      "step": 1761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592617734106886e-06,
+      "loss": 0.468,
+      "step": 1762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58292371899273e-06,
+      "loss": 0.5074,
+      "step": 1763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573230096488722e-06,
+      "loss": 0.4651,
+      "step": 1764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563536875719835e-06,
+      "loss": 0.5208,
+      "step": 1765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553844065810686e-06,
+      "loss": 0.6315,
+      "step": 1766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.544151675885487e-06,
+      "loss": 0.5788,
+      "step": 1767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534459715068065e-06,
+      "loss": 0.5472,
+      "step": 1768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.524768192481842e-06,
+      "loss": 0.5085,
+      "step": 1769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515077117249818e-06,
+      "loss": 0.5017,
+      "step": 1770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505386498494589e-06,
+      "loss": 0.542,
+      "step": 1771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495696345338303e-06,
+      "loss": 0.4687,
+      "step": 1772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.486006666902687e-06,
+      "loss": 0.4445,
+      "step": 1773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476317472309004e-06,
+      "loss": 0.5253,
+      "step": 1774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466628770678077e-06,
+      "loss": 0.5251,
+      "step": 1775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456940571130252e-06,
+      "loss": 0.482,
+      "step": 1776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447252882785417e-06,
+      "loss": 0.532,
+      "step": 1777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437565714762962e-06,
+      "loss": 0.3921,
+      "step": 1778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.427879076181802e-06,
+      "loss": 0.5,
+      "step": 1779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418192976160342e-06,
+      "loss": 0.5012,
+      "step": 1780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408507423816485e-06,
+      "loss": 0.4649,
+      "step": 1781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398822428267623e-06,
+      "loss": 0.6604,
+      "step": 1782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389137998630614e-06,
+      "loss": 0.5075,
+      "step": 1783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379454144021793e-06,
+      "loss": 0.534,
+      "step": 1784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.369770873556945e-06,
+      "loss": 0.5499,
+      "step": 1785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360088196351317e-06,
+      "loss": 0.4958,
+      "step": 1786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350406121519581e-06,
+      "loss": 0.5173,
+      "step": 1787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34072465817586e-06,
+      "loss": 0.4371,
+      "step": 1788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.331043815433682e-06,
+      "loss": 0.6547,
+      "step": 1789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321363602406013e-06,
+      "loss": 0.5108,
+      "step": 1790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311684028205202e-06,
+      "loss": 0.5352,
+      "step": 1791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30200510194302e-06,
+      "loss": 0.4621,
+      "step": 1792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292326832730613e-06,
+      "loss": 0.5127,
+      "step": 1793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.282649229678509e-06,
+      "loss": 0.5864,
+      "step": 1794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27297230189662e-06,
+      "loss": 0.5848,
+      "step": 1795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263296058494207e-06,
+      "loss": 0.5649,
+      "step": 1796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253620508579902e-06,
+      "loss": 0.403,
+      "step": 1797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243945661261669e-06,
+      "loss": 0.6117,
+      "step": 1798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234271525646825e-06,
+      "loss": 0.4879,
+      "step": 1799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224598110842007e-06,
+      "loss": 0.5028,
+      "step": 1800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214925425953181e-06,
+      "loss": 0.6166,
+      "step": 1801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205253480085616e-06,
+      "loss": 0.3953,
+      "step": 1802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195582282343897e-06,
+      "loss": 0.4022,
+      "step": 1803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185911841831899e-06,
+      "loss": 0.4879,
+      "step": 1804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176242167652776e-06,
+      "loss": 0.5446,
+      "step": 1805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166573268908979e-06,
+      "loss": 0.5076,
+      "step": 1806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156905154702208e-06,
+      "loss": 0.6236,
+      "step": 1807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147237834133446e-06,
+      "loss": 0.5534,
+      "step": 1808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137571316302912e-06,
+      "loss": 0.5037,
+      "step": 1809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12790561031008e-06,
+      "loss": 0.5041,
+      "step": 1810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118240725253648e-06,
+      "loss": 0.5749,
+      "step": 1811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10857667023156e-06,
+      "loss": 0.5243,
+      "step": 1812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098913454340955e-06,
+      "loss": 0.6797,
+      "step": 1813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.089251086678205e-06,
+      "loss": 0.4581,
+      "step": 1814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.079589576338865e-06,
+      "loss": 0.4995,
+      "step": 1815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069928932417696e-06,
+      "loss": 0.5757,
+      "step": 1816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060269164008635e-06,
+      "loss": 0.4983,
+      "step": 1817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050610280204798e-06,
+      "loss": 0.4712,
+      "step": 1818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040952290098472e-06,
+      "loss": 0.5117,
+      "step": 1819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.031295202781093e-06,
+      "loss": 0.5432,
+      "step": 1820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02163902734326e-06,
+      "loss": 0.5358,
+      "step": 1821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.011983772874702e-06,
+      "loss": 0.574,
+      "step": 1822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002329448464292e-06,
+      "loss": 0.4749,
+      "step": 1823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992676063200014e-06,
+      "loss": 0.6096,
+      "step": 1824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.983023626168979e-06,
+      "loss": 0.4586,
+      "step": 1825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973372146457402e-06,
+      "loss": 0.4668,
+      "step": 1826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963721633150597e-06,
+      "loss": 0.5765,
+      "step": 1827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954072095332968e-06,
+      "loss": 0.5694,
+      "step": 1828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944423542087996e-06,
+      "loss": 0.5518,
+      "step": 1829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93477598249825e-06,
+      "loss": 0.5268,
+      "step": 1830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925129425645341e-06,
+      "loss": 0.6068,
+      "step": 1831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91548388060996e-06,
+      "loss": 0.5518,
+      "step": 1832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905839356471824e-06,
+      "loss": 0.5793,
+      "step": 1833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896195862309706e-06,
+      "loss": 0.4701,
+      "step": 1834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886553407201398e-06,
+      "loss": 0.4746,
+      "step": 1835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876912000223726e-06,
+      "loss": 0.608,
+      "step": 1836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867271650452512e-06,
+      "loss": 0.5571,
+      "step": 1837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857632366962602e-06,
+      "loss": 0.5404,
+      "step": 1838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84799415882782e-06,
+      "loss": 0.4892,
+      "step": 1839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838357035120994e-06,
+      "loss": 0.6792,
+      "step": 1840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828721004913917e-06,
+      "loss": 0.6926,
+      "step": 1841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.819086077277365e-06,
+      "loss": 0.5447,
+      "step": 1842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80945226128107e-06,
+      "loss": 0.4776,
+      "step": 1843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799819565993713e-06,
+      "loss": 0.6431,
+      "step": 1844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790188000482934e-06,
+      "loss": 0.6491,
+      "step": 1845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.780557573815288e-06,
+      "loss": 0.4977,
+      "step": 1846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.770928295056284e-06,
+      "loss": 0.5756,
+      "step": 1847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761300173270327e-06,
+      "loss": 0.6096,
+      "step": 1848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75167321752075e-06,
+      "loss": 0.4767,
+      "step": 1849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742047436869776e-06,
+      "loss": 0.5348,
+      "step": 1850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73242284037853e-06,
+      "loss": 0.5276,
+      "step": 1851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.722799437107022e-06,
+      "loss": 0.5904,
+      "step": 1852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.713177236114126e-06,
+      "loss": 0.5638,
+      "step": 1853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703556246457608e-06,
+      "loss": 0.5271,
+      "step": 1854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693936477194068e-06,
+      "loss": 0.4454,
+      "step": 1855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68431793737898e-06,
+      "loss": 0.4979,
+      "step": 1856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674700636066639e-06,
+      "loss": 0.4902,
+      "step": 1857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.665084582310195e-06,
+      "loss": 0.5415,
+      "step": 1858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65546978516161e-06,
+      "loss": 0.5806,
+      "step": 1859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645856253671664e-06,
+      "loss": 0.5209,
+      "step": 1860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.636243996889953e-06,
+      "loss": 0.5133,
+      "step": 1861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.626633023864869e-06,
+      "loss": 0.6473,
+      "step": 1862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61702334364359e-06,
+      "loss": 0.4172,
+      "step": 1863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607414965272087e-06,
+      "loss": 0.6746,
+      "step": 1864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5978078977951e-06,
+      "loss": 0.4625,
+      "step": 1865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58820215025613e-06,
+      "loss": 0.65,
+      "step": 1866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57859773169745e-06,
+      "loss": 0.5237,
+      "step": 1867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.568994651160065e-06,
+      "loss": 0.4746,
+      "step": 1868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.559392917683736e-06,
+      "loss": 0.5585,
+      "step": 1869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.549792540306942e-06,
+      "loss": 0.5968,
+      "step": 1870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540193528066898e-06,
+      "loss": 0.5521,
+      "step": 1871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.530595889999523e-06,
+      "loss": 0.4214,
+      "step": 1872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52099963513945e-06,
+      "loss": 0.6191,
+      "step": 1873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511404772520005e-06,
+      "loss": 0.4945,
+      "step": 1874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.501811311173212e-06,
+      "loss": 0.5477,
+      "step": 1875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.492219260129763e-06,
+      "loss": 0.6049,
+      "step": 1876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.482628628419028e-06,
+      "loss": 0.4923,
+      "step": 1877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.473039425069051e-06,
+      "loss": 0.4706,
+      "step": 1878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463451659106511e-06,
+      "loss": 0.4823,
+      "step": 1879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453865339556757e-06,
+      "loss": 0.457,
+      "step": 1880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.444280475443754e-06,
+      "loss": 0.4997,
+      "step": 1881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434697075790117e-06,
+      "loss": 0.4826,
+      "step": 1882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.425115149617065e-06,
+      "loss": 0.5669,
+      "step": 1883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415534705944445e-06,
+      "loss": 0.6461,
+      "step": 1884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.405955753790697e-06,
+      "loss": 0.4393,
+      "step": 1885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396378302172868e-06,
+      "loss": 0.4148,
+      "step": 1886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.386802360106577e-06,
+      "loss": 0.5904,
+      "step": 1887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37722793660604e-06,
+      "loss": 0.4782,
+      "step": 1888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36765504068403e-06,
+      "loss": 0.4471,
+      "step": 1889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358083681351882e-06,
+      "loss": 0.6068,
+      "step": 1890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348513867619499e-06,
+      "loss": 0.5553,
+      "step": 1891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338945608495314e-06,
+      "loss": 0.61,
+      "step": 1892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3293789129863e-06,
+      "loss": 0.5316,
+      "step": 1893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319813790097964e-06,
+      "loss": 0.525,
+      "step": 1894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310250248834328e-06,
+      "loss": 0.4921,
+      "step": 1895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300688298197922e-06,
+      "loss": 0.5546,
+      "step": 1896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29112794718979e-06,
+      "loss": 0.5692,
+      "step": 1897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.281569204809454e-06,
+      "loss": 0.4331,
+      "step": 1898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272012080054938e-06,
+      "loss": 0.4957,
+      "step": 1899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262456581922728e-06,
+      "loss": 0.5162,
+      "step": 1900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25290271940779e-06,
+      "loss": 0.4578,
+      "step": 1901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24335050150355e-06,
+      "loss": 0.5056,
+      "step": 1902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23379993720188e-06,
+      "loss": 0.578,
+      "step": 1903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224251035493096e-06,
+      "loss": 0.4304,
+      "step": 1904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.214703805365951e-06,
+      "loss": 0.6104,
+      "step": 1905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205158255807632e-06,
+      "loss": 0.5179,
+      "step": 1906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195614395803724e-06,
+      "loss": 0.5593,
+      "step": 1907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186072234338246e-06,
+      "loss": 0.4793,
+      "step": 1908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176531780393598e-06,
+      "loss": 0.5857,
+      "step": 1909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.166993042950583e-06,
+      "loss": 0.4628,
+      "step": 1910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157456030988386e-06,
+      "loss": 0.5095,
+      "step": 1911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147920753484572e-06,
+      "loss": 0.4966,
+      "step": 1912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138387219415067e-06,
+      "loss": 0.4799,
+      "step": 1913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.128855437754148e-06,
+      "loss": 0.4713,
+      "step": 1914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119325417474468e-06,
+      "loss": 0.4991,
+      "step": 1915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109797167546993e-06,
+      "loss": 0.4624,
+      "step": 1916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100270696941042e-06,
+      "loss": 0.5377,
+      "step": 1917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090746014624254e-06,
+      "loss": 0.4085,
+      "step": 1918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.081223129562579e-06,
+      "loss": 0.5347,
+      "step": 1919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07170205072028e-06,
+      "loss": 0.5649,
+      "step": 1920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062182787059928e-06,
+      "loss": 0.4985,
+      "step": 1921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.052665347542361e-06,
+      "loss": 0.3873,
+      "step": 1922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043149741126729e-06,
+      "loss": 0.489,
+      "step": 1923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033635976770437e-06,
+      "loss": 0.5284,
+      "step": 1924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02412406342916e-06,
+      "loss": 0.4556,
+      "step": 1925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014614010056835e-06,
+      "loss": 0.4777,
+      "step": 1926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005105825605644e-06,
+      "loss": 0.4826,
+      "step": 1927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995599519026019e-06,
+      "loss": 0.4836,
+      "step": 1928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986095099266603e-06,
+      "loss": 0.4129,
+      "step": 1929
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976592575274287e-06,
+      "loss": 0.5818,
+      "step": 1930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96709195599416e-06,
+      "loss": 0.5255,
+      "step": 1931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957593250369533e-06,
+      "loss": 0.5599,
+      "step": 1932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948096467341897e-06,
+      "loss": 0.5659,
+      "step": 1933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938601615850952e-06,
+      "loss": 0.6023,
+      "step": 1934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.929108704834569e-06,
+      "loss": 0.5505,
+      "step": 1935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919617743228793e-06,
+      "loss": 0.6003,
+      "step": 1936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91012873996784e-06,
+      "loss": 0.576,
+      "step": 1937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90064170398407e-06,
+      "loss": 0.5925,
+      "step": 1938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891156644208009e-06,
+      "loss": 0.426,
+      "step": 1939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881673569568305e-06,
+      "loss": 0.429,
+      "step": 1940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872192488991753e-06,
+      "loss": 0.6407,
+      "step": 1941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862713411403256e-06,
+      "loss": 0.5087,
+      "step": 1942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853236345725843e-06,
+      "loss": 0.5992,
+      "step": 1943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843761300880644e-06,
+      "loss": 0.4287,
+      "step": 1944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83428828578689e-06,
+      "loss": 0.4283,
+      "step": 1945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.824817309361892e-06,
+      "loss": 0.5363,
+      "step": 1946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81534838052106e-06,
+      "loss": 0.5175,
+      "step": 1947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805881508177856e-06,
+      "loss": 0.5521,
+      "step": 1948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796416701243816e-06,
+      "loss": 0.5201,
+      "step": 1949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786953968628535e-06,
+      "loss": 0.5512,
+      "step": 1950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777493319239652e-06,
+      "loss": 0.4424,
+      "step": 1951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76803476198284e-06,
+      "loss": 0.5115,
+      "step": 1952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.758578305761813e-06,
+      "loss": 0.5644,
+      "step": 1953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.7491239594783e-06,
+      "loss": 0.5521,
+      "step": 1954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.739671732032038e-06,
+      "loss": 0.5331,
+      "step": 1955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73022163232079e-06,
+      "loss": 0.6665,
+      "step": 1956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72077366924029e-06,
+      "loss": 0.5011,
+      "step": 1957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711327851684282e-06,
+      "loss": 0.4858,
+      "step": 1958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701884188544477e-06,
+      "loss": 0.5945,
+      "step": 1959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692442688710566e-06,
+      "loss": 0.5107,
+      "step": 1960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6830033610702e-06,
+      "loss": 0.5087,
+      "step": 1961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673566214508982e-06,
+      "loss": 0.6731,
+      "step": 1962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664131257910472e-06,
+      "loss": 0.5005,
+      "step": 1963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.654698500156153e-06,
+      "loss": 0.5079,
+      "step": 1964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.645267950125457e-06,
+      "loss": 0.4627,
+      "step": 1965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63583961669572e-06,
+      "loss": 0.6592,
+      "step": 1966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.626413508742204e-06,
+      "loss": 0.4721,
+      "step": 1967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6169896351380705e-06,
+      "loss": 0.4597,
+      "step": 1968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607568004754378e-06,
+      "loss": 0.5089,
+      "step": 1969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598148626460071e-06,
+      "loss": 0.5812,
+      "step": 1970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588731509121987e-06,
+      "loss": 0.6447,
+      "step": 1971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579316661604816e-06,
+      "loss": 0.4845,
+      "step": 1972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.569904092771121e-06,
+      "loss": 0.4409,
+      "step": 1973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.560493811481324e-06,
+      "loss": 0.4919,
+      "step": 1974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551085826593683e-06,
+      "loss": 0.609,
+      "step": 1975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541680146964307e-06,
+      "loss": 0.4678,
+      "step": 1976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532276781447124e-06,
+      "loss": 0.5306,
+      "step": 1977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522875738893894e-06,
+      "loss": 0.4494,
+      "step": 1978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513477028154175e-06,
+      "loss": 0.4505,
+      "step": 1979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504080658075349e-06,
+      "loss": 0.597,
+      "step": 1980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494686637502576e-06,
+      "loss": 0.5148,
+      "step": 1981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.485294975278823e-06,
+      "loss": 0.434,
+      "step": 1982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475905680244817e-06,
+      "loss": 0.5484,
+      "step": 1983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466518761239077e-06,
+      "loss": 0.632,
+      "step": 1984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.457134227097869e-06,
+      "loss": 0.6305,
+      "step": 1985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447752086655219e-06,
+      "loss": 0.5921,
+      "step": 1986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.438372348742908e-06,
+      "loss": 0.5163,
+      "step": 1987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428995022190441e-06,
+      "loss": 0.6025,
+      "step": 1988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419620115825069e-06,
+      "loss": 0.5335,
+      "step": 1989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410247638471746e-06,
+      "loss": 0.4445,
+      "step": 1990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40087759895316e-06,
+      "loss": 0.5501,
+      "step": 1991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.391510006089686e-06,
+      "loss": 0.4972,
+      "step": 1992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.382144868699409e-06,
+      "loss": 0.5404,
+      "step": 1993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372782195598096e-06,
+      "loss": 0.4576,
+      "step": 1994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3634219955991944e-06,
+      "loss": 0.58,
+      "step": 1995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354064277513829e-06,
+      "loss": 0.5033,
+      "step": 1996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3447090501507775e-06,
+      "loss": 0.397,
+      "step": 1997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.335356322316487e-06,
+      "loss": 0.4284,
+      "step": 1998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3260061028150364e-06,
+      "loss": 0.5154,
+      "step": 1999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316658400448158e-06,
+      "loss": 0.5166,
+      "step": 2000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.307313224015206e-06,
+      "loss": 0.4698,
+      "step": 2001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297970582313159e-06,
+      "loss": 0.5613,
+      "step": 2002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288630484136605e-06,
+      "loss": 0.5262,
+      "step": 2003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279292938277752e-06,
+      "loss": 0.3554,
+      "step": 2004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269957953526385e-06,
+      "loss": 0.4552,
+      "step": 2005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.260625538669897e-06,
+      "loss": 0.6507,
+      "step": 2006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251295702493246e-06,
+      "loss": 0.4246,
+      "step": 2007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241968453778977e-06,
+      "loss": 0.504,
+      "step": 2008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2326438013071875e-06,
+      "loss": 0.4708,
+      "step": 2009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223321753855534e-06,
+      "loss": 0.5451,
+      "step": 2010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214002320199234e-06,
+      "loss": 0.5403,
+      "step": 2011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204685509111018e-06,
+      "loss": 0.5582,
+      "step": 2012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.195371329361176e-06,
+      "loss": 0.5923,
+      "step": 2013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186059789717498e-06,
+      "loss": 0.59,
+      "step": 2014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.176750898945308e-06,
+      "loss": 0.459,
+      "step": 2015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.167444665807418e-06,
+      "loss": 0.6196,
+      "step": 2016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158141099064158e-06,
+      "loss": 0.5176,
+      "step": 2017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148840207473325e-06,
+      "loss": 0.6235,
+      "step": 2018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139541999790221e-06,
+      "loss": 0.5086,
+      "step": 2019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130246484767609e-06,
+      "loss": 0.4905,
+      "step": 2020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120953671155715e-06,
+      "loss": 0.5263,
+      "step": 2021
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1116635677022315e-06,
+      "loss": 0.4695,
+      "step": 2022
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102376183152288e-06,
+      "loss": 0.4941,
+      "step": 2023
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093091526248468e-06,
+      "loss": 0.4142,
+      "step": 2024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083809605730775e-06,
+      "loss": 0.5032,
+      "step": 2025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.074530430336646e-06,
+      "loss": 0.4822,
+      "step": 2026
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065254008800931e-06,
+      "loss": 0.6637,
+      "step": 2027
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055980349855882e-06,
+      "loss": 0.4244,
+      "step": 2028
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.046709462231156e-06,
+      "loss": 0.6193,
+      "step": 2029
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037441354653808e-06,
+      "loss": 0.5104,
+      "step": 2030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.028176035848257e-06,
+      "loss": 0.5121,
+      "step": 2031
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0189135145363204e-06,
+      "loss": 0.4968,
+      "step": 2032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.009653799437162e-06,
+      "loss": 0.4802,
+      "step": 2033
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000396899267312e-06,
+      "loss": 0.3873,
+      "step": 2034
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991142822740655e-06,
+      "loss": 0.4784,
+      "step": 2035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98189157856841e-06,
+      "loss": 0.5464,
+      "step": 2036
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.972643175459137e-06,
+      "loss": 0.5241,
+      "step": 2037
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963397622118713e-06,
+      "loss": 0.5708,
+      "step": 2038
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.954154927250345e-06,
+      "loss": 0.5156,
+      "step": 2039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.944915099554532e-06,
+      "loss": 0.6041,
+      "step": 2040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935678147729092e-06,
+      "loss": 0.5204,
+      "step": 2041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926444080469122e-06,
+      "loss": 0.6241,
+      "step": 2042
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91721290646701e-06,
+      "loss": 0.5617,
+      "step": 2043
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.907984634412424e-06,
+      "loss": 0.5222,
+      "step": 2044
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8987592729922885e-06,
+      "loss": 0.5941,
+      "step": 2045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889536830890803e-06,
+      "loss": 0.4707,
+      "step": 2046
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.880317316789403e-06,
+      "loss": 0.5335,
+      "step": 2047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871100739366785e-06,
+      "loss": 0.4574,
+      "step": 2048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861887107298864e-06,
+      "loss": 0.5794,
+      "step": 2049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852676429258798e-06,
+      "loss": 0.5661,
+      "step": 2050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843468713916947e-06,
+      "loss": 0.4997,
+      "step": 2051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8342639699409045e-06,
+      "loss": 0.5367,
+      "step": 2052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.825062205995443e-06,
+      "loss": 0.4814,
+      "step": 2053
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8158634307425505e-06,
+      "loss": 0.5413,
+      "step": 2054
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8066676528413845e-06,
+      "loss": 0.4754,
+      "step": 2055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797474880948296e-06,
+      "loss": 0.5455,
+      "step": 2056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.788285123716796e-06,
+      "loss": 0.5106,
+      "step": 2057
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779098389797556e-06,
+      "loss": 0.4592,
+      "step": 2058
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7699146878384125e-06,
+      "loss": 0.4986,
+      "step": 2059
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760734026484341e-06,
+      "loss": 0.5985,
+      "step": 2060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7515564143774535e-06,
+      "loss": 0.5248,
+      "step": 2061
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7423818601569915e-06,
+      "loss": 0.4867,
+      "step": 2062
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7332103724593274e-06,
+      "loss": 0.471,
+      "step": 2063
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.724041959917932e-06,
+      "loss": 0.5947,
+      "step": 2064
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714876631163396e-06,
+      "loss": 0.5078,
+      "step": 2065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705714394823393e-06,
+      "loss": 0.5781,
+      "step": 2066
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.696555259522699e-06,
+      "loss": 0.5293,
+      "step": 2067
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.68739923388316e-06,
+      "loss": 0.4112,
+      "step": 2068
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6782463265237005e-06,
+      "loss": 0.5843,
+      "step": 2069
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669096546060314e-06,
+      "loss": 0.5183,
+      "step": 2070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.659949901106035e-06,
+      "loss": 0.4005,
+      "step": 2071
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650806400270967e-06,
+      "loss": 0.4834,
+      "step": 2072
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641666052162234e-06,
+      "loss": 0.6208,
+      "step": 2073
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632528865384011e-06,
+      "loss": 0.5151,
+      "step": 2074
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623394848537479e-06,
+      "loss": 0.4985,
+      "step": 2075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614264010220848e-06,
+      "loss": 0.425,
+      "step": 2076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.605136359029332e-06,
+      "loss": 0.6069,
+      "step": 2077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5960119035551426e-06,
+      "loss": 0.4742,
+      "step": 2078
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5868906523874845e-06,
+      "loss": 0.6159,
+      "step": 2079
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.577772614112552e-06,
+      "loss": 0.596,
+      "step": 2080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568657797313506e-06,
+      "loss": 0.399,
+      "step": 2081
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5595462105704735e-06,
+      "loss": 0.5626,
+      "step": 2082
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550437862460555e-06,
+      "loss": 0.5278,
+      "step": 2083
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541332761557784e-06,
+      "loss": 0.4821,
+      "step": 2084
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532230916433152e-06,
+      "loss": 0.413,
+      "step": 2085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523132335654578e-06,
+      "loss": 0.5485,
+      "step": 2086
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.514037027786917e-06,
+      "loss": 0.5365,
+      "step": 2087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504945001391925e-06,
+      "loss": 0.4225,
+      "step": 2088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.495856265028292e-06,
+      "loss": 0.4209,
+      "step": 2089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48677082725159e-06,
+      "loss": 0.5288,
+      "step": 2090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477688696614302e-06,
+      "loss": 0.5084,
+      "step": 2091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.468609881665787e-06,
+      "loss": 0.5606,
+      "step": 2092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.459534390952289e-06,
+      "loss": 0.5898,
+      "step": 2093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4504622330169195e-06,
+      "loss": 0.491,
+      "step": 2094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441393416399653e-06,
+      "loss": 0.6222,
+      "step": 2095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432327949637324e-06,
+      "loss": 0.4586,
+      "step": 2096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423265841263606e-06,
+      "loss": 0.5182,
+      "step": 2097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414207099809018e-06,
+      "loss": 0.501,
+      "step": 2098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405151733800897e-06,
+      "loss": 0.5265,
+      "step": 2099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396099751763424e-06,
+      "loss": 0.5263,
+      "step": 2100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387051162217571e-06,
+      "loss": 0.5261,
+      "step": 2101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.378005973681136e-06,
+      "loss": 0.5563,
+      "step": 2102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368964194668703e-06,
+      "loss": 0.6151,
+      "step": 2103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3599258336916506e-06,
+      "loss": 0.5341,
+      "step": 2104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350890899258144e-06,
+      "loss": 0.434,
+      "step": 2105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3418593998731106e-06,
+      "loss": 0.5577,
+      "step": 2106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332831344038261e-06,
+      "loss": 0.5464,
+      "step": 2107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323806740252049e-06,
+      "loss": 0.3988,
+      "step": 2108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314785597009692e-06,
+      "loss": 0.5334,
+      "step": 2109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3057679228031365e-06,
+      "loss": 0.5812,
+      "step": 2110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2967537261210745e-06,
+      "loss": 0.5722,
+      "step": 2111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287743015448918e-06,
+      "loss": 0.5532,
+      "step": 2112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2787357992688056e-06,
+      "loss": 0.4582,
+      "step": 2113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.269732086059573e-06,
+      "loss": 0.4838,
+      "step": 2114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2607318842967725e-06,
+      "loss": 0.5528,
+      "step": 2115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.251735202452641e-06,
+      "loss": 0.4623,
+      "step": 2116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242742048996102e-06,
+      "loss": 0.5379,
+      "step": 2117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233752432392768e-06,
+      "loss": 0.5761,
+      "step": 2118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224766361104908e-06,
+      "loss": 0.5724,
+      "step": 2119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215783843591469e-06,
+      "loss": 0.5157,
+      "step": 2120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206804888308037e-06,
+      "loss": 0.4863,
+      "step": 2121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1978295037068585e-06,
+      "loss": 0.4381,
+      "step": 2122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188857698236806e-06,
+      "loss": 0.5462,
+      "step": 2123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179889480343396e-06,
+      "loss": 0.6023,
+      "step": 2124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170924858468755e-06,
+      "loss": 0.5467,
+      "step": 2125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161963841051636e-06,
+      "loss": 0.4863,
+      "step": 2126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153006436527387e-06,
+      "loss": 0.4379,
+      "step": 2127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144052653327969e-06,
+      "loss": 0.5024,
+      "step": 2128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1351024998819246e-06,
+      "loss": 0.5644,
+      "step": 2129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126155984614377e-06,
+      "loss": 0.532,
+      "step": 2130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117213115947036e-06,
+      "loss": 0.5455,
+      "step": 2131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108273902298165e-06,
+      "loss": 0.5071,
+      "step": 2132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099338352082602e-06,
+      "loss": 0.5819,
+      "step": 2133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0904064737117205e-06,
+      "loss": 0.5769,
+      "step": 2134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.081478275593452e-06,
+      "loss": 0.5122,
+      "step": 2135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0725537661322535e-06,
+      "loss": 0.4606,
+      "step": 2136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063632953729113e-06,
+      "loss": 0.4734,
+      "step": 2137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054715846781538e-06,
+      "loss": 0.5667,
+      "step": 2138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045802453683553e-06,
+      "loss": 0.5686,
+      "step": 2139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036892782825681e-06,
+      "loss": 0.509,
+      "step": 2140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027986842594935e-06,
+      "loss": 0.3648,
+      "step": 2141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.019084641374831e-06,
+      "loss": 0.4366,
+      "step": 2142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.010186187545352e-06,
+      "loss": 0.5159,
+      "step": 2143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001291489482963e-06,
+      "loss": 0.4647,
+      "step": 2144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.992400555560587e-06,
+      "loss": 0.4947,
+      "step": 2145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983513394147611e-06,
+      "loss": 0.5299,
+      "step": 2146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.97463001360986e-06,
+      "loss": 0.4688,
+      "step": 2147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965750422309613e-06,
+      "loss": 0.4837,
+      "step": 2148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956874628605565e-06,
+      "loss": 0.5071,
+      "step": 2149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948002640852859e-06,
+      "loss": 0.5373,
+      "step": 2150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9391344674030295e-06,
+      "loss": 0.4883,
+      "step": 2151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930270116604043e-06,
+      "loss": 0.5209,
+      "step": 2152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.921409596800254e-06,
+      "loss": 0.4855,
+      "step": 2153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.912552916332413e-06,
+      "loss": 0.5229,
+      "step": 2154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903700083537665e-06,
+      "loss": 0.4623,
+      "step": 2155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894851106749516e-06,
+      "loss": 0.5492,
+      "step": 2156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886005994297863e-06,
+      "loss": 0.5317,
+      "step": 2157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877164754508946e-06,
+      "loss": 0.5984,
+      "step": 2158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8683273957053774e-06,
+      "loss": 0.4697,
+      "step": 2159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.859493926206096e-06,
+      "loss": 0.55,
+      "step": 2160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.850664354326399e-06,
+      "loss": 0.5318,
+      "step": 2161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.841838688377902e-06,
+      "loss": 0.5765,
+      "step": 2162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833016936668554e-06,
+      "loss": 0.4798,
+      "step": 2163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824199107502606e-06,
+      "loss": 0.5004,
+      "step": 2164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8153852091806275e-06,
+      "loss": 0.5288,
+      "step": 2165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8065752499994835e-06,
+      "loss": 0.5039,
+      "step": 2166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797769238252331e-06,
+      "loss": 0.4501,
+      "step": 2167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788967182228613e-06,
+      "loss": 0.5913,
+      "step": 2168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780169090214047e-06,
+      "loss": 0.5,
+      "step": 2169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77137497049062e-06,
+      "loss": 0.5517,
+      "step": 2170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762584831336577e-06,
+      "loss": 0.5671,
+      "step": 2171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753798681026422e-06,
+      "loss": 0.5229,
+      "step": 2172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745016527830898e-06,
+      "loss": 0.5203,
+      "step": 2173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736238380016988e-06,
+      "loss": 0.5094,
+      "step": 2174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727464245847902e-06,
+      "loss": 0.4169,
+      "step": 2175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718694133583083e-06,
+      "loss": 0.4517,
+      "step": 2176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709928051478168e-06,
+      "loss": 0.589,
+      "step": 2177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701166007785017e-06,
+      "loss": 0.5963,
+      "step": 2178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.692408010751683e-06,
+      "loss": 0.5585,
+      "step": 2179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683654068622408e-06,
+      "loss": 0.534,
+      "step": 2180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67490418963762e-06,
+      "loss": 0.388,
+      "step": 2181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6661583820339195e-06,
+      "loss": 0.3662,
+      "step": 2182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.657416654044077e-06,
+      "loss": 0.4259,
+      "step": 2183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.648679013897015e-06,
+      "loss": 0.5041,
+      "step": 2184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6399454698178294e-06,
+      "loss": 0.4579,
+      "step": 2185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631216030027727e-06,
+      "loss": 0.4936,
+      "step": 2186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622490702744087e-06,
+      "loss": 0.4413,
+      "step": 2187
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613769496180388e-06,
+      "loss": 0.4387,
+      "step": 2188
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605052418546242e-06,
+      "loss": 0.4796,
+      "step": 2189
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.596339478047377e-06,
+      "loss": 0.445,
+      "step": 2190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5876306828856185e-06,
+      "loss": 0.5144,
+      "step": 2191
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.578926041258907e-06,
+      "loss": 0.6325,
+      "step": 2192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.570225561361242e-06,
+      "loss": 0.5919,
+      "step": 2193
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5615292513827415e-06,
+      "loss": 0.491,
+      "step": 2194
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552837119509566e-06,
+      "loss": 0.4479,
+      "step": 2195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.544149173923972e-06,
+      "loss": 0.406,
+      "step": 2196
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535465422804245e-06,
+      "loss": 0.4604,
+      "step": 2197
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5267858743247494e-06,
+      "loss": 0.4738,
+      "step": 2198
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51811053665588e-06,
+      "loss": 0.6284,
+      "step": 2199
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.509439417964071e-06,
+      "loss": 0.5102,
+      "step": 2200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500772526411784e-06,
+      "loss": 0.5643,
+      "step": 2201
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492109870157491e-06,
+      "loss": 0.4042,
+      "step": 2202
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.483451457355706e-06,
+      "loss": 0.5271,
+      "step": 2203
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474797296156913e-06,
+      "loss": 0.4355,
+      "step": 2204
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.466147394707625e-06,
+      "loss": 0.5381,
+      "step": 2205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457501761150317e-06,
+      "loss": 0.5172,
+      "step": 2206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4488604036234705e-06,
+      "loss": 0.499,
+      "step": 2207
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44022333026153e-06,
+      "loss": 0.5786,
+      "step": 2208
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4315905491949075e-06,
+      "loss": 0.4683,
+      "step": 2209
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422962068549976e-06,
+      "loss": 0.6066,
+      "step": 2210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414337896449062e-06,
+      "loss": 0.5067,
+      "step": 2211
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405718041010436e-06,
+      "loss": 0.5335,
+      "step": 2212
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3971025103482936e-06,
+      "loss": 0.451,
+      "step": 2213
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38849131257278e-06,
+      "loss": 0.5022,
+      "step": 2214
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379884455789948e-06,
+      "loss": 0.5311,
+      "step": 2215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371281948101765e-06,
+      "loss": 0.4992,
+      "step": 2216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36268379760611e-06,
+      "loss": 0.5504,
+      "step": 2217
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3540900123967535e-06,
+      "loss": 0.5796,
+      "step": 2218
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.34550060056336e-06,
+      "loss": 0.5776,
+      "step": 2219
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.336915570191481e-06,
+      "loss": 0.5262,
+      "step": 2220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3283349293625354e-06,
+      "loss": 0.4191,
+      "step": 2221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3197586861538175e-06,
+      "loss": 0.5824,
+      "step": 2222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311186848638479e-06,
+      "loss": 0.4693,
+      "step": 2223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302619424885522e-06,
+      "loss": 0.5796,
+      "step": 2224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294056422959797e-06,
+      "loss": 0.5318,
+      "step": 2225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.285497850921992e-06,
+      "loss": 0.4857,
+      "step": 2226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276943716828624e-06,
+      "loss": 0.4751,
+      "step": 2227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2683940287320315e-06,
+      "loss": 0.4467,
+      "step": 2228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25984879468037e-06,
+      "loss": 0.522,
+      "step": 2229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251308022717601e-06,
+      "loss": 0.3704,
+      "step": 2230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242771720883484e-06,
+      "loss": 0.53,
+      "step": 2231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234239897213573e-06,
+      "loss": 0.5705,
+      "step": 2232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225712559739207e-06,
+      "loss": 0.4672,
+      "step": 2233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217189716487496e-06,
+      "loss": 0.4621,
+      "step": 2234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208671375481336e-06,
+      "loss": 0.4399,
+      "step": 2235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200157544739359e-06,
+      "loss": 0.3977,
+      "step": 2236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191648232275974e-06,
+      "loss": 0.5422,
+      "step": 2237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183143446101326e-06,
+      "loss": 0.5273,
+      "step": 2238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1746431942213e-06,
+      "loss": 0.5652,
+      "step": 2239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.166147484637516e-06,
+      "loss": 0.4926,
+      "step": 2240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157656325347311e-06,
+      "loss": 0.4555,
+      "step": 2241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149169724343757e-06,
+      "loss": 0.5439,
+      "step": 2242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140687689615604e-06,
+      "loss": 0.501,
+      "step": 2243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.132210229147341e-06,
+      "loss": 0.5208,
+      "step": 2244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.123737350919111e-06,
+      "loss": 0.5498,
+      "step": 2245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.115269062906782e-06,
+      "loss": 0.5102,
+      "step": 2246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10680537308187e-06,
+      "loss": 0.5065,
+      "step": 2247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098346289411589e-06,
+      "loss": 0.3999,
+      "step": 2248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089891819858792e-06,
+      "loss": 0.4879,
+      "step": 2249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0814419723819995e-06,
+      "loss": 0.6783,
+      "step": 2250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072996754935397e-06,
+      "loss": 0.5177,
+      "step": 2251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064556175468777e-06,
+      "loss": 0.5026,
+      "step": 2252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056120241927604e-06,
+      "loss": 0.5318,
+      "step": 2253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0476889622529354e-06,
+      "loss": 0.5224,
+      "step": 2254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.039262344381477e-06,
+      "loss": 0.5371,
+      "step": 2255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030840396245516e-06,
+      "loss": 0.3786,
+      "step": 2256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022423125772973e-06,
+      "loss": 0.541,
+      "step": 2257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0140105408873495e-06,
+      "loss": 0.4759,
+      "step": 2258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.005602649507737e-06,
+      "loss": 0.5248,
+      "step": 2259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997199459548817e-06,
+      "loss": 0.5324,
+      "step": 2260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.988800978920826e-06,
+      "loss": 0.5011,
+      "step": 2261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980407215529598e-06,
+      "loss": 0.5017,
+      "step": 2262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972018177276492e-06,
+      "loss": 0.5793,
+      "step": 2263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963633872058453e-06,
+      "loss": 0.4103,
+      "step": 2264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.955254307767938e-06,
+      "loss": 0.487,
+      "step": 2265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94687949229297e-06,
+      "loss": 0.6277,
+      "step": 2266
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938509433517083e-06,
+      "loss": 0.5742,
+      "step": 2267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9301441393193405e-06,
+      "loss": 0.5369,
+      "step": 2268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921783617574319e-06,
+      "loss": 0.5702,
+      "step": 2269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913427876152104e-06,
+      "loss": 0.5257,
+      "step": 2270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.90507692291828e-06,
+      "loss": 0.4472,
+      "step": 2271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.896730765733925e-06,
+      "loss": 0.4765,
+      "step": 2272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8883894124555985e-06,
+      "loss": 0.5685,
+      "step": 2273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.880052870935346e-06,
+      "loss": 0.4513,
+      "step": 2274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871721149020674e-06,
+      "loss": 0.4646,
+      "step": 2275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86339425455456e-06,
+      "loss": 0.4605,
+      "step": 2276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855072195375433e-06,
+      "loss": 0.4917,
+      "step": 2277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.846754979317171e-06,
+      "loss": 0.3818,
+      "step": 2278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838442614209093e-06,
+      "loss": 0.4825,
+      "step": 2279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8301351078759525e-06,
+      "loss": 0.5327,
+      "step": 2280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821832468137929e-06,
+      "loss": 0.5481,
+      "step": 2281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81353470281062e-06,
+      "loss": 0.5611,
+      "step": 2282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805241819705036e-06,
+      "loss": 0.5533,
+      "step": 2283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796953826627589e-06,
+      "loss": 0.4786,
+      "step": 2284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788670731380091e-06,
+      "loss": 0.4895,
+      "step": 2285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.780392541759741e-06,
+      "loss": 0.4816,
+      "step": 2286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772119265559123e-06,
+      "loss": 0.6283,
+      "step": 2287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.763850910566193e-06,
+      "loss": 0.5872,
+      "step": 2288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.755587484564274e-06,
+      "loss": 0.5061,
+      "step": 2289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747328995332053e-06,
+      "loss": 0.507,
+      "step": 2290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739075450643565e-06,
+      "loss": 0.4206,
+      "step": 2291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.730826858268196e-06,
+      "loss": 0.509,
+      "step": 2292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7225832259706616e-06,
+      "loss": 0.5711,
+      "step": 2293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714344561511025e-06,
+      "loss": 0.5007,
+      "step": 2294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706110872644648e-06,
+      "loss": 0.5729,
+      "step": 2295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697882167122236e-06,
+      "loss": 0.541,
+      "step": 2296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689658452689779e-06,
+      "loss": 0.5237,
+      "step": 2297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.681439737088585e-06,
+      "loss": 0.5023,
+      "step": 2298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673226028055249e-06,
+      "loss": 0.4694,
+      "step": 2299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665017333321652e-06,
+      "loss": 0.5095,
+      "step": 2300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65681366061497e-06,
+      "loss": 0.588,
+      "step": 2301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648615017657622e-06,
+      "loss": 0.5552,
+      "step": 2302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640421412167329e-06,
+      "loss": 0.5194,
+      "step": 2303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6322328518570315e-06,
+      "loss": 0.5231,
+      "step": 2304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624049344434959e-06,
+      "loss": 0.4935,
+      "step": 2305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615870897604548e-06,
+      "loss": 0.5058,
+      "step": 2306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607697519064505e-06,
+      "loss": 0.4593,
+      "step": 2307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599529216508738e-06,
+      "loss": 0.4488,
+      "step": 2308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591365997626388e-06,
+      "loss": 0.5792,
+      "step": 2309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.583207870101822e-06,
+      "loss": 0.4797,
+      "step": 2310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575054841614589e-06,
+      "loss": 0.5599,
+      "step": 2311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566906919839467e-06,
+      "loss": 0.6006,
+      "step": 2312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558764112446395e-06,
+      "loss": 0.5668,
+      "step": 2313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.550626427100533e-06,
+      "loss": 0.492,
+      "step": 2314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.542493871462182e-06,
+      "loss": 0.6218,
+      "step": 2315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534366453186849e-06,
+      "loss": 0.5199,
+      "step": 2316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.526244179925182e-06,
+      "loss": 0.4971,
+      "step": 2317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518127059322996e-06,
+      "loss": 0.5283,
+      "step": 2318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5100150990212485e-06,
+      "loss": 0.5918,
+      "step": 2319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5019083066560485e-06,
+      "loss": 0.5333,
+      "step": 2320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493806689858635e-06,
+      "loss": 0.5037,
+      "step": 2321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485710256255367e-06,
+      "loss": 0.538,
+      "step": 2322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477619013467742e-06,
+      "loss": 0.4478,
+      "step": 2323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.469532969112359e-06,
+      "loss": 0.4964,
+      "step": 2324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461452130800927e-06,
+      "loss": 0.4212,
+      "step": 2325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453376506140251e-06,
+      "loss": 0.503,
+      "step": 2326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4453061027322345e-06,
+      "loss": 0.4754,
+      "step": 2327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437240928173859e-06,
+      "loss": 0.5854,
+      "step": 2328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429180990057191e-06,
+      "loss": 0.4688,
+      "step": 2329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421126295969362e-06,
+      "loss": 0.6373,
+      "step": 2330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413076853492569e-06,
+      "loss": 0.5409,
+      "step": 2331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405032670204066e-06,
+      "loss": 0.4393,
+      "step": 2332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.396993753676157e-06,
+      "loss": 0.4889,
+      "step": 2333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388960111476185e-06,
+      "loss": 0.4011,
+      "step": 2334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38093175116653e-06,
+      "loss": 0.536,
+      "step": 2335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372908680304604e-06,
+      "loss": 0.5103,
+      "step": 2336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364890906442832e-06,
+      "loss": 0.5234,
+      "step": 2337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3568784371286586e-06,
+      "loss": 0.4529,
+      "step": 2338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348871279904532e-06,
+      "loss": 0.4881,
+      "step": 2339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3408694423079025e-06,
+      "loss": 0.6024,
+      "step": 2340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3328729318712105e-06,
+      "loss": 0.4889,
+      "step": 2341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324881756121882e-06,
+      "loss": 0.4455,
+      "step": 2342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31689592258232e-06,
+      "loss": 0.5945,
+      "step": 2343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3089154387699104e-06,
+      "loss": 0.43,
+      "step": 2344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300940312196982e-06,
+      "loss": 0.6054,
+      "step": 2345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2929705503708364e-06,
+      "loss": 0.4104,
+      "step": 2346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28500616079372e-06,
+      "loss": 0.5515,
+      "step": 2347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277047150962824e-06,
+      "loss": 0.4846,
+      "step": 2348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2690935283702726e-06,
+      "loss": 0.5117,
+      "step": 2349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2611453005031225e-06,
+      "loss": 0.5147,
+      "step": 2350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253202474843348e-06,
+      "loss": 0.5167,
+      "step": 2351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245265058867837e-06,
+      "loss": 0.5597,
+      "step": 2352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2373330600484e-06,
+      "loss": 0.4927,
+      "step": 2353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229406485851721e-06,
+      "loss": 0.5806,
+      "step": 2354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221485343739407e-06,
+      "loss": 0.4362,
+      "step": 2355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213569641167929e-06,
+      "loss": 0.4988,
+      "step": 2356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205659385588646e-06,
+      "loss": 0.6285,
+      "step": 2357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197754584447793e-06,
+      "loss": 0.5186,
+      "step": 2358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189855245186461e-06,
+      "loss": 0.5496,
+      "step": 2359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181961375240619e-06,
+      "loss": 0.5198,
+      "step": 2360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174072982041057e-06,
+      "loss": 0.483,
+      "step": 2361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166190073013443e-06,
+      "loss": 0.4975,
+      "step": 2362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15831265557825e-06,
+      "loss": 0.5351,
+      "step": 2363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150440737150815e-06,
+      "loss": 0.5172,
+      "step": 2364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142574325141265e-06,
+      "loss": 0.5417,
+      "step": 2365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134713426954569e-06,
+      "loss": 0.5862,
+      "step": 2366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126858049990495e-06,
+      "loss": 0.5331,
+      "step": 2367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119008201643616e-06,
+      "loss": 0.4559,
+      "step": 2368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1111638893033025e-06,
+      "loss": 0.6395,
+      "step": 2369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1033251203537e-06,
+      "loss": 0.4186,
+      "step": 2370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.095491902173761e-06,
+      "loss": 0.422,
+      "step": 2371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087664242137185e-06,
+      "loss": 0.5548,
+      "step": 2372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079842147612468e-06,
+      "loss": 0.5344,
+      "step": 2373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072025625962838e-06,
+      "loss": 0.5363,
+      "step": 2374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064214684546299e-06,
+      "loss": 0.444,
+      "step": 2375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056409330715595e-06,
+      "loss": 0.568,
+      "step": 2376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048609571818207e-06,
+      "loss": 0.6008,
+      "step": 2377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040815415196352e-06,
+      "loss": 0.4801,
+      "step": 2378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033026868186973e-06,
+      "loss": 0.4954,
+      "step": 2379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025243938121736e-06,
+      "loss": 0.6231,
+      "step": 2380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017466632327007e-06,
+      "loss": 0.472,
+      "step": 2381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009694958123876e-06,
+      "loss": 0.5166,
+      "step": 2382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001928922828118e-06,
+      "loss": 0.4801,
+      "step": 2383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994168533750207e-06,
+      "loss": 0.5344,
+      "step": 2384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986413798195297e-06,
+      "loss": 0.5407,
+      "step": 2385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9786647234632234e-06,
+      "loss": 0.6319,
+      "step": 2386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970921316848493e-06,
+      "loss": 0.5314,
+      "step": 2387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963183585640274e-06,
+      "loss": 0.4439,
+      "step": 2388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.955451537122396e-06,
+      "loss": 0.613,
+      "step": 2389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947725178573335e-06,
+      "loss": 0.4269,
+      "step": 2390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940004517266216e-06,
+      "loss": 0.565,
+      "step": 2391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932289560468796e-06,
+      "loss": 0.4227,
+      "step": 2392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9245803154434645e-06,
+      "loss": 0.5811,
+      "step": 2393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916876789447236e-06,
+      "loss": 0.5304,
+      "step": 2394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909178989731735e-06,
+      "loss": 0.6028,
+      "step": 2395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9014869235432056e-06,
+      "loss": 0.5209,
+      "step": 2396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893800598122486e-06,
+      "loss": 0.4227,
+      "step": 2397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886120020705013e-06,
+      "loss": 0.5746,
+      "step": 2398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878445198520815e-06,
+      "loss": 0.3898,
+      "step": 2399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8707761387945e-06,
+      "loss": 0.5002,
+      "step": 2400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863112848745253e-06,
+      "loss": 0.5754,
+      "step": 2401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855455335586824e-06,
+      "loss": 0.5196,
+      "step": 2402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847803606527537e-06,
+      "loss": 0.6152,
+      "step": 2403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840157668770255e-06,
+      "loss": 0.457,
+      "step": 2404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832517529512397e-06,
+      "loss": 0.5624,
+      "step": 2405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824883195945926e-06,
+      "loss": 0.3973,
+      "step": 2406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817254675257338e-06,
+      "loss": 0.5478,
+      "step": 2407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8096319746276544e-06,
+      "loss": 0.5152,
+      "step": 2408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8020151012324168e-06,
+      "loss": 0.4943,
+      "step": 2409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7944040622416968e-06,
+      "loss": 0.5995,
+      "step": 2410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.786798864820047e-06,
+      "loss": 0.4977,
+      "step": 2411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7791995161265494e-06,
+      "loss": 0.4513,
+      "step": 2412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7716060233147523e-06,
+      "loss": 0.5263,
+      "step": 2413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.764018393532721e-06,
+      "loss": 0.5889,
+      "step": 2414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.756436633922974e-06,
+      "loss": 0.4349,
+      "step": 2415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7488607516225297e-06,
+      "loss": 0.6294,
+      "step": 2416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7412907537628496e-06,
+      "loss": 0.4065,
+      "step": 2417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7337266474698675e-06,
+      "loss": 0.5687,
+      "step": 2418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726168439863984e-06,
+      "loss": 0.496,
+      "step": 2419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7186161380600184e-06,
+      "loss": 0.4961,
+      "step": 2420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.71106974916726e-06,
+      "loss": 0.5324,
+      "step": 2421
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7035292802894063e-06,
+      "loss": 0.4008,
+      "step": 2422
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6959947385246065e-06,
+      "loss": 0.5716,
+      "step": 2423
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6884661309654046e-06,
+      "loss": 0.4862,
+      "step": 2424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680943464698784e-06,
+      "loss": 0.5288,
+      "step": 2425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6734267468061167e-06,
+      "loss": 0.4633,
+      "step": 2426
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.665915984363185e-06,
+      "loss": 0.4322,
+      "step": 2427
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.658411184440165e-06,
+      "loss": 0.5169,
+      "step": 2428
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6509123541016043e-06,
+      "loss": 0.4445,
+      "step": 2429
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.643419500406461e-06,
+      "loss": 0.3821,
+      "step": 2430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6359326304080346e-06,
+      "loss": 0.4808,
+      "step": 2431
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.628451751154016e-06,
+      "loss": 0.4424,
+      "step": 2432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6209768696864478e-06,
+      "loss": 0.4789,
+      "step": 2433
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.613507993041725e-06,
+      "loss": 0.5985,
+      "step": 2434
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6060451282505927e-06,
+      "loss": 0.5156,
+      "step": 2435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5985882823381358e-06,
+      "loss": 0.3918,
+      "step": 2436
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.591137462323775e-06,
+      "loss": 0.5525,
+      "step": 2437
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.583692675221254e-06,
+      "loss": 0.5972,
+      "step": 2438
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5762539280386443e-06,
+      "loss": 0.4557,
+      "step": 2439
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.568821227778327e-06,
+      "loss": 0.5088,
+      "step": 2440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.561394581436991e-06,
+      "loss": 0.5569,
+      "step": 2441
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5539739960056297e-06,
+      "loss": 0.6864,
+      "step": 2442
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546559478469528e-06,
+      "loss": 0.4083,
+      "step": 2443
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5391510358082603e-06,
+      "loss": 0.3987,
+      "step": 2444
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531748674995682e-06,
+      "loss": 0.5083,
+      "step": 2445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5243524029999232e-06,
+      "loss": 0.4076,
+      "step": 2446
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5169622267833826e-06,
+      "loss": 0.4948,
+      "step": 2447
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5095781533027216e-06,
+      "loss": 0.4468,
+      "step": 2448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5022001895088553e-06,
+      "loss": 0.4358,
+      "step": 2449
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49482834234695e-06,
+      "loss": 0.55,
+      "step": 2450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.48746261875641e-06,
+      "loss": 0.4575,
+      "step": 2451
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.480103025670879e-06,
+      "loss": 0.426,
+      "step": 2452
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472749570018228e-06,
+      "loss": 0.5431,
+      "step": 2453
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4654022587205517e-06,
+      "loss": 0.5335,
+      "step": 2454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458061098694161e-06,
+      "loss": 0.5092,
+      "step": 2455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4507260968495737e-06,
+      "loss": 0.4697,
+      "step": 2456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4433972600915144e-06,
+      "loss": 0.4679,
+      "step": 2457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.436074595318901e-06,
+      "loss": 0.5788,
+      "step": 2458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4287581094248447e-06,
+      "loss": 0.616,
+      "step": 2459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4214478092966374e-06,
+      "loss": 0.5246,
+      "step": 2460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4141437018157453e-06,
+      "loss": 0.4498,
+      "step": 2461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4068457938578205e-06,
+      "loss": 0.525,
+      "step": 2462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.399554092292654e-06,
+      "loss": 0.3182,
+      "step": 2463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3922686039842225e-06,
+      "loss": 0.5128,
+      "step": 2464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3849893357906283e-06,
+      "loss": 0.4871,
+      "step": 2465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.377716294564134e-06,
+      "loss": 0.5018,
+      "step": 2466
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3704494871511374e-06,
+      "loss": 0.3756,
+      "step": 2467
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3631889203921617e-06,
+      "loss": 0.5223,
+      "step": 2468
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355934601121874e-06,
+      "loss": 0.4817,
+      "step": 2469
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3486865361690323e-06,
+      "loss": 0.4449,
+      "step": 2470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3414447323565356e-06,
+      "loss": 0.458,
+      "step": 2471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3342091965013636e-06,
+      "loss": 0.4298,
+      "step": 2472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3269799354146195e-06,
+      "loss": 0.6215,
+      "step": 2473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.319756955901476e-06,
+      "loss": 0.5548,
+      "step": 2474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.312540264761216e-06,
+      "loss": 0.4908,
+      "step": 2475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3053298687871904e-06,
+      "loss": 0.4799,
+      "step": 2476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.298125774766816e-06,
+      "loss": 0.5181,
+      "step": 2477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290927989481599e-06,
+      "loss": 0.5332,
+      "step": 2478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283736519707085e-06,
+      "loss": 0.4988,
+      "step": 2479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.276551372212895e-06,
+      "loss": 0.4817,
+      "step": 2480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.269372553762675e-06,
+      "loss": 0.4899,
+      "step": 2481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262200071114141e-06,
+      "loss": 0.5358,
+      "step": 2482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2550339310190173e-06,
+      "loss": 0.5482,
+      "step": 2483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247874140223077e-06,
+      "loss": 0.471,
+      "step": 2484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2407207054661106e-06,
+      "loss": 0.4941,
+      "step": 2485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233573633481922e-06,
+      "loss": 0.3761,
+      "step": 2486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22643293099833e-06,
+      "loss": 0.5256,
+      "step": 2487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2192986047371557e-06,
+      "loss": 0.4933,
+      "step": 2488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2121706614142213e-06,
+      "loss": 0.6113,
+      "step": 2489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2050491077393274e-06,
+      "loss": 0.4595,
+      "step": 2490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1979339504162786e-06,
+      "loss": 0.4939,
+      "step": 2491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1908251961428482e-06,
+      "loss": 0.6181,
+      "step": 2492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.183722851610782e-06,
+      "loss": 0.4508,
+      "step": 2493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1766269235057944e-06,
+      "loss": 0.446,
+      "step": 2494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1695374185075577e-06,
+      "loss": 0.6075,
+      "step": 2495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.162454343289699e-06,
+      "loss": 0.4577,
+      "step": 2496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1553777045197944e-06,
+      "loss": 0.5973,
+      "step": 2497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1483075088593582e-06,
+      "loss": 0.4974,
+      "step": 2498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1412437629638416e-06,
+      "loss": 0.5041,
+      "step": 2499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1341864734826224e-06,
+      "loss": 0.5927,
+      "step": 2500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.127135647059003e-06,
+      "loss": 0.5423,
+      "step": 2501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1200912903302005e-06,
+      "loss": 0.4989,
+      "step": 2502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1130534099273424e-06,
+      "loss": 0.5096,
+      "step": 2503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1060220124754593e-06,
+      "loss": 0.4256,
+      "step": 2504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0989971045934795e-06,
+      "loss": 0.3961,
+      "step": 2505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091978692894222e-06,
+      "loss": 0.5021,
+      "step": 2506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.084966783984392e-06,
+      "loss": 0.4663,
+      "step": 2507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.077961384464572e-06,
+      "loss": 0.4913,
+      "step": 2508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0709625009292153e-06,
+      "loss": 0.4575,
+      "step": 2509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0639701399666467e-06,
+      "loss": 0.588,
+      "step": 2510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0569843081590435e-06,
+      "loss": 0.4973,
+      "step": 2511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050005012082449e-06,
+      "loss": 0.4861,
+      "step": 2512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0430322583067384e-06,
+      "loss": 0.6309,
+      "step": 2513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.03606605339564e-06,
+      "loss": 0.4804,
+      "step": 2514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.029106403906713e-06,
+      "loss": 0.3869,
+      "step": 2515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0221533163913454e-06,
+      "loss": 0.4779,
+      "step": 2516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0152067973947507e-06,
+      "loss": 0.4213,
+      "step": 2517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0082668534559544e-06,
+      "loss": 0.4016,
+      "step": 2518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0013334911078053e-06,
+      "loss": 0.589,
+      "step": 2519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9944067168769344e-06,
+      "loss": 0.4995,
+      "step": 2520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.987486537283796e-06,
+      "loss": 0.421,
+      "step": 2521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9805729588426134e-06,
+      "loss": 0.5068,
+      "step": 2522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9736659880614182e-06,
+      "loss": 0.4332,
+      "step": 2523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9667656314420035e-06,
+      "loss": 0.5005,
+      "step": 2524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.959871895479944e-06,
+      "loss": 0.5423,
+      "step": 2525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9529847866645855e-06,
+      "loss": 0.4261,
+      "step": 2526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9461043114790256e-06,
+      "loss": 0.6425,
+      "step": 2527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9392304764001335e-06,
+      "loss": 0.5031,
+      "step": 2528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9323632878985054e-06,
+      "loss": 0.4592,
+      "step": 2529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.925502752438506e-06,
+      "loss": 0.4864,
+      "step": 2530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9186488764782117e-06,
+      "loss": 0.5721,
+      "step": 2531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.911801666469456e-06,
+      "loss": 0.5534,
+      "step": 2532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9049611288577705e-06,
+      "loss": 0.5232,
+      "step": 2533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89812727008243e-06,
+      "loss": 0.5866,
+      "step": 2534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8913000965764094e-06,
+      "loss": 0.5692,
+      "step": 2535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.884479614766391e-06,
+      "loss": 0.4822,
+      "step": 2536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8776658310727633e-06,
+      "loss": 0.5112,
+      "step": 2537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.870858751909599e-06,
+      "loss": 0.486,
+      "step": 2538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.864058383684676e-06,
+      "loss": 0.3907,
+      "step": 2539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.857264732799436e-06,
+      "loss": 0.4987,
+      "step": 2540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8504778056490177e-06,
+      "loss": 0.4379,
+      "step": 2541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.843697608622209e-06,
+      "loss": 0.3759,
+      "step": 2542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.836924148101482e-06,
+      "loss": 0.5365,
+      "step": 2543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.830157430462955e-06,
+      "loss": 0.4968,
+      "step": 2544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.823397462076404e-06,
+      "loss": 0.4527,
+      "step": 2545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8166442493052492e-06,
+      "loss": 0.4799,
+      "step": 2546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8098977985065543e-06,
+      "loss": 0.4921,
+      "step": 2547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8031581160310173e-06,
+      "loss": 0.5652,
+      "step": 2548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.796425208222956e-06,
+      "loss": 0.4895,
+      "step": 2549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7896990814203275e-06,
+      "loss": 0.4234,
+      "step": 2550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7829797419546933e-06,
+      "loss": 0.5022,
+      "step": 2551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77626719615123e-06,
+      "loss": 0.5909,
+      "step": 2552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7695614503287172e-06,
+      "loss": 0.5469,
+      "step": 2553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7628625107995356e-06,
+      "loss": 0.5659,
+      "step": 2554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7561703838696574e-06,
+      "loss": 0.4846,
+      "step": 2555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7494850758386425e-06,
+      "loss": 0.4906,
+      "step": 2556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.742806592999632e-06,
+      "loss": 0.5381,
+      "step": 2557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.73613494163934e-06,
+      "loss": 0.5109,
+      "step": 2558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729470128038054e-06,
+      "loss": 0.5437,
+      "step": 2559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.722812158469622e-06,
+      "loss": 0.5452,
+      "step": 2560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7161610392014483e-06,
+      "loss": 0.5547,
+      "step": 2561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.709516776494492e-06,
+      "loss": 0.5222,
+      "step": 2562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.702879376603257e-06,
+      "loss": 0.4824,
+      "step": 2563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696248845775784e-06,
+      "loss": 0.4446,
+      "step": 2564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.689625190253651e-06,
+      "loss": 0.5593,
+      "step": 2565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.683008416271963e-06,
+      "loss": 0.4893,
+      "step": 2566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6763985300593464e-06,
+      "loss": 0.5526,
+      "step": 2567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669795537837946e-06,
+      "loss": 0.4516,
+      "step": 2568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6631994458234145e-06,
+      "loss": 0.4832,
+      "step": 2569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6566102602249055e-06,
+      "loss": 0.5813,
+      "step": 2570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6500279872450873e-06,
+      "loss": 0.4622,
+      "step": 2571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6434526330801003e-06,
+      "loss": 0.6247,
+      "step": 2572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.636884203919584e-06,
+      "loss": 0.4931,
+      "step": 2573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6303227059466564e-06,
+      "loss": 0.4219,
+      "step": 2574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6237681453379104e-06,
+      "loss": 0.4936,
+      "step": 2575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.617220528263411e-06,
+      "loss": 0.467,
+      "step": 2576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61067986088668e-06,
+      "loss": 0.5377,
+      "step": 2577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6041461493647123e-06,
+      "loss": 0.4513,
+      "step": 2578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597619399847934e-06,
+      "loss": 0.3962,
+      "step": 2579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.591099618480238e-06,
+      "loss": 0.4917,
+      "step": 2580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584586811398938e-06,
+      "loss": 0.5292,
+      "step": 2581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5780809847348053e-06,
+      "loss": 0.4846,
+      "step": 2582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.571582144612015e-06,
+      "loss": 0.4607,
+      "step": 2583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5650902971481906e-06,
+      "loss": 0.4575,
+      "step": 2584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.558605448454351e-06,
+      "loss": 0.4857,
+      "step": 2585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5521276046349387e-06,
+      "loss": 0.6071,
+      "step": 2586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5456567717878076e-06,
+      "loss": 0.5061,
+      "step": 2587
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5391929560041937e-06,
+      "loss": 0.4089,
+      "step": 2588
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5327361633687498e-06,
+      "loss": 0.5072,
+      "step": 2589
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5262863999594933e-06,
+      "loss": 0.3819,
+      "step": 2590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5198436718478526e-06,
+      "loss": 0.4425,
+      "step": 2591
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5134079850986035e-06,
+      "loss": 0.4691,
+      "step": 2592
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.506979345769919e-06,
+      "loss": 0.5327,
+      "step": 2593
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5005577599133235e-06,
+      "loss": 0.4935,
+      "step": 2594
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494143233573708e-06,
+      "loss": 0.4129,
+      "step": 2595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4877357727893158e-06,
+      "loss": 0.5988,
+      "step": 2596
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481335383591733e-06,
+      "loss": 0.4042,
+      "step": 2597
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.474942072005906e-06,
+      "loss": 0.6317,
+      "step": 2598
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.468555844050097e-06,
+      "loss": 0.3697,
+      "step": 2599
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.462176705735917e-06,
+      "loss": 0.4276,
+      "step": 2600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4558046630682975e-06,
+      "loss": 0.4948,
+      "step": 2601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.449439722045488e-06,
+      "loss": 0.4809,
+      "step": 2602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.443081888659058e-06,
+      "loss": 0.4915,
+      "step": 2603
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.436731168893881e-06,
+      "loss": 0.4249,
+      "step": 2604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4303875687281373e-06,
+      "loss": 0.3684,
+      "step": 2605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.424051094133306e-06,
+      "loss": 0.5129,
+      "step": 2606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.417721751074157e-06,
+      "loss": 0.6245,
+      "step": 2607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.411399545508747e-06,
+      "loss": 0.4221,
+      "step": 2608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4050844833884135e-06,
+      "loss": 0.4777,
+      "step": 2609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3987765706577716e-06,
+      "loss": 0.5399,
+      "step": 2610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3924758132547043e-06,
+      "loss": 0.4744,
+      "step": 2611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.386182217110361e-06,
+      "loss": 0.5073,
+      "step": 2612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.379895788149148e-06,
+      "loss": 0.4843,
+      "step": 2613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.373616532288725e-06,
+      "loss": 0.6145,
+      "step": 2614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3673444554400004e-06,
+      "loss": 0.4664,
+      "step": 2615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.361079563507126e-06,
+      "loss": 0.5005,
+      "step": 2616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3548218623874864e-06,
+      "loss": 0.4868,
+      "step": 2617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3485713579716997e-06,
+      "loss": 0.4125,
+      "step": 2618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3423280561436103e-06,
+      "loss": 0.4378,
+      "step": 2619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3360919627802794e-06,
+      "loss": 0.4734,
+      "step": 2620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3298630837519865e-06,
+      "loss": 0.5523,
+      "step": 2621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323641424922216e-06,
+      "loss": 0.4174,
+      "step": 2622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3174269921476587e-06,
+      "loss": 0.565,
+      "step": 2623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.311219791278203e-06,
+      "loss": 0.494,
+      "step": 2624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.305019828156927e-06,
+      "loss": 0.4492,
+      "step": 2625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2988271086200984e-06,
+      "loss": 0.4661,
+      "step": 2626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2926416384971663e-06,
+      "loss": 0.4993,
+      "step": 2627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.286463423610752e-06,
+      "loss": 0.4451,
+      "step": 2628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.280292469776648e-06,
+      "loss": 0.4939,
+      "step": 2629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2741287828038215e-06,
+      "loss": 0.4429,
+      "step": 2630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.267972368494382e-06,
+      "loss": 0.4542,
+      "step": 2631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.261823232643611e-06,
+      "loss": 0.4782,
+      "step": 2632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25568138103992e-06,
+      "loss": 0.489,
+      "step": 2633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2495468194648796e-06,
+      "loss": 0.4841,
+      "step": 2634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.243419553693189e-06,
+      "loss": 0.5119,
+      "step": 2635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2372995894926787e-06,
+      "loss": 0.4257,
+      "step": 2636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2311869326243196e-06,
+      "loss": 0.4365,
+      "step": 2637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2250815888421805e-06,
+      "loss": 0.4983,
+      "step": 2638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.218983563893472e-06,
+      "loss": 0.6078,
+      "step": 2639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2128928635184897e-06,
+      "loss": 0.5666,
+      "step": 2640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2068094934506603e-06,
+      "loss": 0.5109,
+      "step": 2641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2007334594164843e-06,
+      "loss": 0.5814,
+      "step": 2642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194664767135577e-06,
+      "loss": 0.4563,
+      "step": 2643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1886034223206355e-06,
+      "loss": 0.5632,
+      "step": 2644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.182549430677431e-06,
+      "loss": 0.4234,
+      "step": 2645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1765027979048338e-06,
+      "loss": 0.5441,
+      "step": 2646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.170463529694762e-06,
+      "loss": 0.4632,
+      "step": 2647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1644316317322257e-06,
+      "loss": 0.3515,
+      "step": 2648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.158407109695274e-06,
+      "loss": 0.5569,
+      "step": 2649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1523899692550364e-06,
+      "loss": 0.5452,
+      "step": 2650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1463802160756685e-06,
+      "loss": 0.5424,
+      "step": 2651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.140377855814395e-06,
+      "loss": 0.5263,
+      "step": 2652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1343828941214673e-06,
+      "loss": 0.3471,
+      "step": 2653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1283953366401768e-06,
+      "loss": 0.5011,
+      "step": 2654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1224151890068446e-06,
+      "loss": 0.4536,
+      "step": 2655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116442456850816e-06,
+      "loss": 0.4623,
+      "step": 2656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1104771457944597e-06,
+      "loss": 0.4639,
+      "step": 2657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1045192614531463e-06,
+      "loss": 0.5548,
+      "step": 2658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098568809435273e-06,
+      "loss": 0.4813,
+      "step": 2659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092625795342229e-06,
+      "loss": 0.6651,
+      "step": 2660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.086690224768405e-06,
+      "loss": 0.5005,
+      "step": 2661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.080762103301186e-06,
+      "loss": 0.492,
+      "step": 2662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0748414365209413e-06,
+      "loss": 0.4331,
+      "step": 2663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.068928230001026e-06,
+      "loss": 0.5028,
+      "step": 2664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.063022489307772e-06,
+      "loss": 0.4739,
+      "step": 2665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0571242200004828e-06,
+      "loss": 0.5212,
+      "step": 2666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.051233427631428e-06,
+      "loss": 0.3976,
+      "step": 2667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0453501177458415e-06,
+      "loss": 0.5243,
+      "step": 2668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03947429588191e-06,
+      "loss": 0.4736,
+      "step": 2669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0336059675707753e-06,
+      "loss": 0.5131,
+      "step": 2670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027745138336522e-06,
+      "loss": 0.4236,
+      "step": 2671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0218918136961784e-06,
+      "loss": 0.4239,
+      "step": 2672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0160459991597046e-06,
+      "loss": 0.5296,
+      "step": 2673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.010207700229996e-06,
+      "loss": 0.4454,
+      "step": 2674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0043769224028685e-06,
+      "loss": 0.4711,
+      "step": 2675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9985536711670607e-06,
+      "loss": 0.3716,
+      "step": 2676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992737952004228e-06,
+      "loss": 0.4625,
+      "step": 2677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98692977038893e-06,
+      "loss": 0.5112,
+      "step": 2678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981129131788633e-06,
+      "loss": 0.5376,
+      "step": 2679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9753360416637123e-06,
+      "loss": 0.4984,
+      "step": 2680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9695505054674212e-06,
+      "loss": 0.5744,
+      "step": 2681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.963772528645913e-06,
+      "loss": 0.4721,
+      "step": 2682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9580021166382213e-06,
+      "loss": 0.531,
+      "step": 2683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95223927487626e-06,
+      "loss": 0.4423,
+      "step": 2684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9464840087848158e-06,
+      "loss": 0.4431,
+      "step": 2685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9407363237815434e-06,
+      "loss": 0.5866,
+      "step": 2686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9349962252769703e-06,
+      "loss": 0.6087,
+      "step": 2687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9292637186744646e-06,
+      "loss": 0.4143,
+      "step": 2688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9235388093702656e-06,
+      "loss": 0.5851,
+      "step": 2689
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9178215027534463e-06,
+      "loss": 0.4171,
+      "step": 2690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912111804205938e-06,
+      "loss": 0.5187,
+      "step": 2691
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9064097191024966e-06,
+      "loss": 0.4692,
+      "step": 2692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9007152528107176e-06,
+      "loss": 0.4616,
+      "step": 2693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8950284106910243e-06,
+      "loss": 0.4951,
+      "step": 2694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8893491980966617e-06,
+      "loss": 0.5884,
+      "step": 2695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8836776203736984e-06,
+      "loss": 0.3972,
+      "step": 2696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.878013682861003e-06,
+      "loss": 0.5265,
+      "step": 2697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8723573908902715e-06,
+      "loss": 0.5452,
+      "step": 2698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8667087497859792e-06,
+      "loss": 0.4381,
+      "step": 2699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8610677648654242e-06,
+      "loss": 0.5441,
+      "step": 2700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8554344414386726e-06,
+      "loss": 0.5866,
+      "step": 2701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8498087848086022e-06,
+      "loss": 0.5396,
+      "step": 2702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8441908002708576e-06,
+      "loss": 0.4044,
+      "step": 2703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8385804931138661e-06,
+      "loss": 0.5599,
+      "step": 2704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8329778686188326e-06,
+      "loss": 0.4623,
+      "step": 2705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8273829320597169e-06,
+      "loss": 0.547,
+      "step": 2706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8217956887032606e-06,
+      "loss": 0.5874,
+      "step": 2707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8162161438089443e-06,
+      "loss": 0.4184,
+      "step": 2708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8106443026290177e-06,
+      "loss": 0.4237,
+      "step": 2709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8050801704084687e-06,
+      "loss": 0.5039,
+      "step": 2710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7995237523850329e-06,
+      "loss": 0.4924,
+      "step": 2711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.793975053789183e-06,
+      "loss": 0.4267,
+      "step": 2712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7884340798441236e-06,
+      "loss": 0.4726,
+      "step": 2713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7829008357657907e-06,
+      "loss": 0.548,
+      "step": 2714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7773753267628412e-06,
+      "loss": 0.4957,
+      "step": 2715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7718575580366548e-06,
+      "loss": 0.5253,
+      "step": 2716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7663475347813152e-06,
+      "loss": 0.4998,
+      "step": 2717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.760845262183628e-06,
+      "loss": 0.4893,
+      "step": 2718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7553507454230956e-06,
+      "loss": 0.4784,
+      "step": 2719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7498639896719194e-06,
+      "loss": 0.5893,
+      "step": 2720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7443850000949968e-06,
+      "loss": 0.6388,
+      "step": 2721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7389137818499147e-06,
+      "loss": 0.5767,
+      "step": 2722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7334503400869419e-06,
+      "loss": 0.463,
+      "step": 2723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.72799467994903e-06,
+      "loss": 0.5454,
+      "step": 2724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7225468065718055e-06,
+      "loss": 0.3832,
+      "step": 2725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7171067250835615e-06,
+      "loss": 0.5564,
+      "step": 2726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.71167444060526e-06,
+      "loss": 0.5266,
+      "step": 2727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706249958250521e-06,
+      "loss": 0.6269,
+      "step": 2728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7008332831256214e-06,
+      "loss": 0.5705,
+      "step": 2729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.695424420329489e-06,
+      "loss": 0.4782,
+      "step": 2730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6900233749536964e-06,
+      "loss": 0.5315,
+      "step": 2731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.684630152082457e-06,
+      "loss": 0.5401,
+      "step": 2732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.679244756792624e-06,
+      "loss": 0.5267,
+      "step": 2733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6738671941536765e-06,
+      "loss": 0.6526,
+      "step": 2734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6684974692277257e-06,
+      "loss": 0.5083,
+      "step": 2735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.663135587069502e-06,
+      "loss": 0.5771,
+      "step": 2736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.657781552726354e-06,
+      "loss": 0.5195,
+      "step": 2737
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6524353712382402e-06,
+      "loss": 0.6036,
+      "step": 2738
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6470970476377369e-06,
+      "loss": 0.5291,
+      "step": 2739
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6417665869500089e-06,
+      "loss": 0.3222,
+      "step": 2740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.636443994192828e-06,
+      "loss": 0.4598,
+      "step": 2741
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6311292743765583e-06,
+      "loss": 0.4511,
+      "step": 2742
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6258224325041528e-06,
+      "loss": 0.5164,
+      "step": 2743
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6205234735711472e-06,
+      "loss": 0.476,
+      "step": 2744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6152324025656573e-06,
+      "loss": 0.5315,
+      "step": 2745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6099492244683811e-06,
+      "loss": 0.4074,
+      "step": 2746
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6046739442525693e-06,
+      "loss": 0.4328,
+      "step": 2747
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5994065668840597e-06,
+      "loss": 0.4471,
+      "step": 2748
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5941470973212281e-06,
+      "loss": 0.362,
+      "step": 2749
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5888955405150285e-06,
+      "loss": 0.4589,
+      "step": 2750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5836519014089457e-06,
+      "loss": 0.5248,
+      "step": 2751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.57841618493903e-06,
+      "loss": 0.4534,
+      "step": 2752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5731883960338645e-06,
+      "loss": 0.5292,
+      "step": 2753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5679685396145616e-06,
+      "loss": 0.4661,
+      "step": 2754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5627566205947875e-06,
+      "loss": 0.4669,
+      "step": 2755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5575526438807143e-06,
+      "loss": 0.5978,
+      "step": 2756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5523566143710556e-06,
+      "loss": 0.5139,
+      "step": 2757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5471685369570288e-06,
+      "loss": 0.4943,
+      "step": 2758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5419884165223819e-06,
+      "loss": 0.5622,
+      "step": 2759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5368162579433555e-06,
+      "loss": 0.5282,
+      "step": 2760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5316520660887092e-06,
+      "loss": 0.48,
+      "step": 2761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5264958458196954e-06,
+      "loss": 0.5396,
+      "step": 2762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5213476019900675e-06,
+      "loss": 0.4504,
+      "step": 2763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5162073394460685e-06,
+      "loss": 0.4418,
+      "step": 2764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5110750630264203e-06,
+      "loss": 0.4118,
+      "step": 2765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5059507775623438e-06,
+      "loss": 0.4768,
+      "step": 2766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5008344878775206e-06,
+      "loss": 0.499,
+      "step": 2767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.495726198788119e-06,
+      "loss": 0.443,
+      "step": 2768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.490625915102769e-06,
+      "loss": 0.4609,
+      "step": 2769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4855336416225652e-06,
+      "loss": 0.5601,
+      "step": 2770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4804493831410637e-06,
+      "loss": 0.5282,
+      "step": 2771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4753731444442742e-06,
+      "loss": 0.4269,
+      "step": 2772
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4703049303106597e-06,
+      "loss": 0.5565,
+      "step": 2773
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4652447455111253e-06,
+      "loss": 0.4886,
+      "step": 2774
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4601925948090211e-06,
+      "loss": 0.4076,
+      "step": 2775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4551484829601325e-06,
+      "loss": 0.4436,
+      "step": 2776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4501124147126787e-06,
+      "loss": 0.4851,
+      "step": 2777
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4450843948073078e-06,
+      "loss": 0.5961,
+      "step": 2778
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4400644279770891e-06,
+      "loss": 0.4314,
+      "step": 2779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4350525189475151e-06,
+      "loss": 0.5924,
+      "step": 2780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4300486724364882e-06,
+      "loss": 0.5735,
+      "step": 2781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.425052893154325e-06,
+      "loss": 0.5345,
+      "step": 2782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4200651858037485e-06,
+      "loss": 0.5243,
+      "step": 2783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4150855550798793e-06,
+      "loss": 0.5324,
+      "step": 2784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.410114005670239e-06,
+      "loss": 0.5188,
+      "step": 2785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4051505422547418e-06,
+      "loss": 0.5048,
+      "step": 2786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.400195169505687e-06,
+      "loss": 0.4622,
+      "step": 2787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395247892087761e-06,
+      "loss": 0.4738,
+      "step": 2788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3903087146580285e-06,
+      "loss": 0.5054,
+      "step": 2789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3853776418659292e-06,
+      "loss": 0.5432,
+      "step": 2790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3804546783532747e-06,
+      "loss": 0.4595,
+      "step": 2791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.375539828754241e-06,
+      "loss": 0.5347,
+      "step": 2792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3706330976953685e-06,
+      "loss": 0.4936,
+      "step": 2793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3657344897955539e-06,
+      "loss": 0.5297,
+      "step": 2794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3608440096660458e-06,
+      "loss": 0.4401,
+      "step": 2795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3559616619104488e-06,
+      "loss": 0.5414,
+      "step": 2796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3510874511246995e-06,
+      "loss": 0.4251,
+      "step": 2797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3462213818970903e-06,
+      "loss": 0.3898,
+      "step": 2798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3413634588082325e-06,
+      "loss": 0.4985,
+      "step": 2799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.336513686431089e-06,
+      "loss": 0.4393,
+      "step": 2800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3316720693309304e-06,
+      "loss": 0.6041,
+      "step": 2801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.326838612065363e-06,
+      "loss": 0.443,
+      "step": 2802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3220133191843077e-06,
+      "loss": 0.5344,
+      "step": 2803
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3171961952299984e-06,
+      "loss": 0.4095,
+      "step": 2804
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.312387244736989e-06,
+      "loss": 0.4278,
+      "step": 2805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3075864722321207e-06,
+      "loss": 0.492,
+      "step": 2806
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3027938822345598e-06,
+      "loss": 0.5581,
+      "step": 2807
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2980094792557462e-06,
+      "loss": 0.4695,
+      "step": 2808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2932332677994342e-06,
+      "loss": 0.5063,
+      "step": 2809
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2884652523616502e-06,
+      "loss": 0.5039,
+      "step": 2810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2837054374307178e-06,
+      "loss": 0.4219,
+      "step": 2811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2789538274872359e-06,
+      "loss": 0.5197,
+      "step": 2812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2742104270040755e-06,
+      "loss": 0.4584,
+      "step": 2813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2694752404463907e-06,
+      "loss": 0.4236,
+      "step": 2814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2647482722715888e-06,
+      "loss": 0.4547,
+      "step": 2815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2600295269293584e-06,
+      "loss": 0.517,
+      "step": 2816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2553190088616296e-06,
+      "loss": 0.5159,
+      "step": 2817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2506167225026033e-06,
+      "loss": 0.5791,
+      "step": 2818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2459226722787188e-06,
+      "loss": 0.5166,
+      "step": 2819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2412368626086723e-06,
+      "loss": 0.4124,
+      "step": 2820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2365592979033992e-06,
+      "loss": 0.4834,
+      "step": 2821
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.23188998256607e-06,
+      "loss": 0.5072,
+      "step": 2822
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2272289209920973e-06,
+      "loss": 0.6177,
+      "step": 2823
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2225761175691153e-06,
+      "loss": 0.5832,
+      "step": 2824
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2179315766769951e-06,
+      "loss": 0.5051,
+      "step": 2825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2132953026878147e-06,
+      "loss": 0.5899,
+      "step": 2826
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2086672999658878e-06,
+      "loss": 0.5332,
+      "step": 2827
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2040475728677292e-06,
+      "loss": 0.3867,
+      "step": 2828
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1994361257420694e-06,
+      "loss": 0.5745,
+      "step": 2829
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1948329629298439e-06,
+      "loss": 0.4988,
+      "step": 2830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1902380887641873e-06,
+      "loss": 0.5084,
+      "step": 2831
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1856515075704334e-06,
+      "loss": 0.5364,
+      "step": 2832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1810732236661115e-06,
+      "loss": 0.472,
+      "step": 2833
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1765032413609378e-06,
+      "loss": 0.4779,
+      "step": 2834
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1719415649568132e-06,
+      "loss": 0.5701,
+      "step": 2835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1673881987478242e-06,
+      "loss": 0.4601,
+      "step": 2836
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1628431470202285e-06,
+      "loss": 0.4606,
+      "step": 2837
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1583064140524614e-06,
+      "loss": 0.4556,
+      "step": 2838
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.153778004115128e-06,
+      "loss": 0.3881,
+      "step": 2839
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1492579214709942e-06,
+      "loss": 0.4025,
+      "step": 2840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1447461703749919e-06,
+      "loss": 0.3646,
+      "step": 2841
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1402427550742067e-06,
+      "loss": 0.4527,
+      "step": 2842
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1357476798078793e-06,
+      "loss": 0.504,
+      "step": 2843
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1312609488073989e-06,
+      "loss": 0.5695,
+      "step": 2844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1267825662963006e-06,
+      "loss": 0.4746,
+      "step": 2845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.12231253649026e-06,
+      "loss": 0.4375,
+      "step": 2846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1178508635970886e-06,
+      "loss": 0.653,
+      "step": 2847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1133975518167372e-06,
+      "loss": 0.4699,
+      "step": 2848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.108952605341277e-06,
+      "loss": 0.4652,
+      "step": 2849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1045160283549105e-06,
+      "loss": 0.4945,
+      "step": 2850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.100087825033962e-06,
+      "loss": 0.5111,
+      "step": 2851
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0956679995468688e-06,
+      "loss": 0.565,
+      "step": 2852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0912565560541877e-06,
+      "loss": 0.5301,
+      "step": 2853
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0868534987085778e-06,
+      "loss": 0.4815,
+      "step": 2854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0824588316548156e-06,
+      "loss": 0.4394,
+      "step": 2855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0780725590297647e-06,
+      "loss": 0.4469,
+      "step": 2856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.073694684962402e-06,
+      "loss": 0.4187,
+      "step": 2857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.069325213573782e-06,
+      "loss": 0.5158,
+      "step": 2858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0649641489770668e-06,
+      "loss": 0.4315,
+      "step": 2859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0606114952774905e-06,
+      "loss": 0.4548,
+      "step": 2860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0562672565723774e-06,
+      "loss": 0.5178,
+      "step": 2861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.051931436951129e-06,
+      "loss": 0.5575,
+      "step": 2862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0476040404952182e-06,
+      "loss": 0.3566,
+      "step": 2863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0432850712781972e-06,
+      "loss": 0.3983,
+      "step": 2864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0389745333656743e-06,
+      "loss": 0.5257,
+      "step": 2865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0346724308153333e-06,
+      "loss": 0.4251,
+      "step": 2866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0303787676769018e-06,
+      "loss": 0.3507,
+      "step": 2867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0260935479921818e-06,
+      "loss": 0.5605,
+      "step": 2868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.021816775795008e-06,
+      "loss": 0.4565,
+      "step": 2869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0175484551112801e-06,
+      "loss": 0.5361,
+      "step": 2870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0132885899589318e-06,
+      "loss": 0.5186,
+      "step": 2871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0090371843479395e-06,
+      "loss": 0.4361,
+      "step": 2872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0047942422803192e-06,
+      "loss": 0.4678,
+      "step": 2873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0005597677501121e-06,
+      "loss": 0.5095,
+      "step": 2874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.963337647434012e-07,
+      "loss": 0.4403,
+      "step": 2875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.921162372382809e-07,
+      "loss": 0.4578,
+      "step": 2876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.879071892048774e-07,
+      "loss": 0.5441,
+      "step": 2877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.837066246053328e-07,
+      "loss": 0.4636,
+      "step": 2878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.795145473937994e-07,
+      "loss": 0.5021,
+      "step": 2879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.753309615164442e-07,
+      "loss": 0.5749,
+      "step": 2880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.711558709114387e-07,
+      "loss": 0.4547,
+      "step": 2881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.669892795089586e-07,
+      "loss": 0.5493,
+      "step": 2882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.628311912311793e-07,
+      "loss": 0.4079,
+      "step": 2883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.586816099922725e-07,
+      "loss": 0.4395,
+      "step": 2884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.545405396983942e-07,
+      "loss": 0.5511,
+      "step": 2885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.50407984247701e-07,
+      "loss": 0.4582,
+      "step": 2886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.462839475303276e-07,
+      "loss": 0.5289,
+      "step": 2887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.421684334283887e-07,
+      "loss": 0.419,
+      "step": 2888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.380614458159798e-07,
+      "loss": 0.4111,
+      "step": 2889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.339629885591661e-07,
+      "loss": 0.4785,
+      "step": 2890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.298730655159871e-07,
+      "loss": 0.4229,
+      "step": 2891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.257916805364453e-07,
+      "loss": 0.5868,
+      "step": 2892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.217188374625075e-07,
+      "loss": 0.4739,
+      "step": 2893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.176545401281e-07,
+      "loss": 0.3793,
+      "step": 2894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.135987923591028e-07,
+      "loss": 0.5212,
+      "step": 2895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.095515979733516e-07,
+      "loss": 0.4741,
+      "step": 2896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.055129607806246e-07,
+      "loss": 0.5137,
+      "step": 2897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.014828845826518e-07,
+      "loss": 0.428,
+      "step": 2898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.974613731730985e-07,
+      "loss": 0.5006,
+      "step": 2899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.934484303375712e-07,
+      "loss": 0.524,
+      "step": 2900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.89444059853608e-07,
+      "loss": 0.4248,
+      "step": 2901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.854482654906805e-07,
+      "loss": 0.3998,
+      "step": 2902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.814610510101828e-07,
+      "loss": 0.3559,
+      "step": 2903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.774824201654386e-07,
+      "loss": 0.4873,
+      "step": 2904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.735123767016851e-07,
+      "loss": 0.5146,
+      "step": 2905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.695509243560784e-07,
+      "loss": 0.4611,
+      "step": 2906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.655980668576936e-07,
+      "loss": 0.5029,
+      "step": 2907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.616538079275028e-07,
+      "loss": 0.479,
+      "step": 2908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.577181512783938e-07,
+      "loss": 0.5312,
+      "step": 2909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.537911006151511e-07,
+      "loss": 0.4694,
+      "step": 2910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.498726596344631e-07,
+      "loss": 0.4274,
+      "step": 2911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.459628320249091e-07,
+      "loss": 0.4594,
+      "step": 2912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.420616214669608e-07,
+      "loss": 0.4451,
+      "step": 2913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.381690316329838e-07,
+      "loss": 0.4966,
+      "step": 2914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.342850661872193e-07,
+      "loss": 0.3944,
+      "step": 2915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.304097287858015e-07,
+      "loss": 0.4603,
+      "step": 2916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.265430230767302e-07,
+      "loss": 0.4808,
+      "step": 2917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.226849526998903e-07,
+      "loss": 0.5785,
+      "step": 2918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.188355212870302e-07,
+      "loss": 0.5329,
+      "step": 2919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.149947324617724e-07,
+      "loss": 0.4252,
+      "step": 2920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.111625898396014e-07,
+      "loss": 0.4012,
+      "step": 2921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.073390970278571e-07,
+      "loss": 0.4547,
+      "step": 2922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.035242576257496e-07,
+      "loss": 0.4937,
+      "step": 2923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.997180752243283e-07,
+      "loss": 0.5089,
+      "step": 2924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.959205534065085e-07,
+      "loss": 0.5342,
+      "step": 2925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.921316957470382e-07,
+      "loss": 0.4392,
+      "step": 2926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.883515058125224e-07,
+      "loss": 0.5602,
+      "step": 2927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.845799871613957e-07,
+      "loss": 0.5458,
+      "step": 2928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.808171433439404e-07,
+      "loss": 0.4649,
+      "step": 2929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.770629779022654e-07,
+      "loss": 0.5217,
+      "step": 2930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.733174943703125e-07,
+      "loss": 0.3933,
+      "step": 2931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.69580696273855e-07,
+      "loss": 0.5198,
+      "step": 2932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.65852587130479e-07,
+      "loss": 0.4118,
+      "step": 2933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.621331704496071e-07,
+      "loss": 0.598,
+      "step": 2934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.584224497324633e-07,
+      "loss": 0.3196,
+      "step": 2935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.547204284720988e-07,
+      "loss": 0.4836,
+      "step": 2936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.510271101533684e-07,
+      "loss": 0.4993,
+      "step": 2937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.473424982529376e-07,
+      "loss": 0.473,
+      "step": 2938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.436665962392741e-07,
+      "loss": 0.5444,
+      "step": 2939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.39999407572648e-07,
+      "loss": 0.5643,
+      "step": 2940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.363409357051255e-07,
+      "loss": 0.3686,
+      "step": 2941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.326911840805706e-07,
+      "loss": 0.5125,
+      "step": 2942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.290501561346352e-07,
+      "loss": 0.4498,
+      "step": 2943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.254178552947621e-07,
+      "loss": 0.5218,
+      "step": 2944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.217942849801762e-07,
+      "loss": 0.438,
+      "step": 2945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.181794486018845e-07,
+      "loss": 0.5069,
+      "step": 2946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.145733495626761e-07,
+      "loss": 0.4972,
+      "step": 2947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.109759912571102e-07,
+      "loss": 0.5281,
+      "step": 2948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0738737707152e-07,
+      "loss": 0.4388,
+      "step": 2949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.038075103840092e-07,
+      "loss": 0.4543,
+      "step": 2950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.002363945644453e-07,
+      "loss": 0.4894,
+      "step": 2951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.966740329744581e-07,
+      "loss": 0.4306,
+      "step": 2952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.931204289674376e-07,
+      "loss": 0.4476,
+      "step": 2953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.895755858885289e-07,
+      "loss": 0.5383,
+      "step": 2954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.860395070746328e-07,
+      "loss": 0.4713,
+      "step": 2955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.825121958543957e-07,
+      "loss": 0.4901,
+      "step": 2956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.789936555482146e-07,
+      "loss": 0.5176,
+      "step": 2957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.754838894682269e-07,
+      "loss": 0.5293,
+      "step": 2958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.719829009183143e-07,
+      "loss": 0.4156,
+      "step": 2959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.684906931940915e-07,
+      "loss": 0.3942,
+      "step": 2960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.650072695829102e-07,
+      "loss": 0.4972,
+      "step": 2961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.615326333638538e-07,
+      "loss": 0.4489,
+      "step": 2962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.580667878077296e-07,
+      "loss": 0.5237,
+      "step": 2963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.546097361770787e-07,
+      "loss": 0.4802,
+      "step": 2964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.511614817261513e-07,
+      "loss": 0.4995,
+      "step": 2965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.47722027700931e-07,
+      "loss": 0.5118,
+      "step": 2966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.442913773391024e-07,
+      "loss": 0.4915,
+      "step": 2967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.408695338700766e-07,
+      "loss": 0.4349,
+      "step": 2968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.374565005149647e-07,
+      "loss": 0.6206,
+      "step": 2969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.340522804865878e-07,
+      "loss": 0.4109,
+      "step": 2970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.306568769894705e-07,
+      "loss": 0.4656,
+      "step": 2971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.272702932198394e-07,
+      "loss": 0.4423,
+      "step": 2972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.238925323656197e-07,
+      "loss": 0.4133,
+      "step": 2973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.205235976064239e-07,
+      "loss": 0.4723,
+      "step": 2974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.171634921135683e-07,
+      "loss": 0.5848,
+      "step": 2975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.138122190500428e-07,
+      "loss": 0.4955,
+      "step": 2976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.104697815705407e-07,
+      "loss": 0.5059,
+      "step": 2977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.071361828214195e-07,
+      "loss": 0.4777,
+      "step": 2978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.038114259407313e-07,
+      "loss": 0.4495,
+      "step": 2979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.004955140581981e-07,
+      "loss": 0.4773,
+      "step": 2980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.971884502952141e-07,
+      "loss": 0.3812,
+      "step": 2981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.938902377648525e-07,
+      "loss": 0.4382,
+      "step": 2982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.906008795718421e-07,
+      "loss": 0.4542,
+      "step": 2983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.873203788125914e-07,
+      "loss": 0.5783,
+      "step": 2984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.84048738575157e-07,
+      "loss": 0.4414,
+      "step": 2985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.807859619392653e-07,
+      "loss": 0.5451,
+      "step": 2986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.775320519762939e-07,
+      "loss": 0.433,
+      "step": 2987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.742870117492749e-07,
+      "loss": 0.5002,
+      "step": 2988
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.710508443128904e-07,
+      "loss": 0.558,
+      "step": 2989
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.678235527134723e-07,
+      "loss": 0.4983,
+      "step": 2990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.64605139988994e-07,
+      "loss": 0.3714,
+      "step": 2991
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.613956091690742e-07,
+      "loss": 0.5629,
+      "step": 2992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.581949632749695e-07,
+      "loss": 0.4627,
+      "step": 2993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.550032053195687e-07,
+      "loss": 0.4967,
+      "step": 2994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.518203383074016e-07,
+      "loss": 0.5294,
+      "step": 2995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.486463652346252e-07,
+      "loss": 0.5941,
+      "step": 2996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.454812890890226e-07,
+      "loss": 0.4982,
+      "step": 2997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.423251128500029e-07,
+      "loss": 0.4105,
+      "step": 2998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.391778394885983e-07,
+      "loss": 0.5564,
+      "step": 2999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.360394719674589e-07,
+      "loss": 0.498,
+      "step": 3000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.329100132408538e-07,
+      "loss": 0.5327,
+      "step": 3001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.297894662546632e-07,
+      "loss": 0.5573,
+      "step": 3002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.266778339463796e-07,
+      "loss": 0.333,
+      "step": 3003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.235751192451045e-07,
+      "loss": 0.4063,
+      "step": 3004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.204813250715424e-07,
+      "loss": 0.5234,
+      "step": 3005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.173964543380027e-07,
+      "loss": 0.3987,
+      "step": 3006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.143205099483961e-07,
+      "loss": 0.4921,
+      "step": 3007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.112534947982273e-07,
+      "loss": 0.3794,
+      "step": 3008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.081954117745968e-07,
+      "loss": 0.5388,
+      "step": 3009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.051462637561977e-07,
+      "loss": 0.4329,
+      "step": 3010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.021060536133127e-07,
+      "loss": 0.4797,
+      "step": 3011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.990747842078081e-07,
+      "loss": 0.5766,
+      "step": 3012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.96052458393137e-07,
+      "loss": 0.3975,
+      "step": 3013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.930390790143313e-07,
+      "loss": 0.529,
+      "step": 3014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.900346489080032e-07,
+      "loss": 0.5054,
+      "step": 3015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.870391709023414e-07,
+      "loss": 0.6179,
+      "step": 3016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.840526478171037e-07,
+      "loss": 0.3923,
+      "step": 3017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.810750824636202e-07,
+      "loss": 0.421,
+      "step": 3018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.781064776447908e-07,
+      "loss": 0.4231,
+      "step": 3019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7514683615507815e-07,
+      "loss": 0.4683,
+      "step": 3020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.721961607805081e-07,
+      "loss": 0.4844,
+      "step": 3021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6925445429866435e-07,
+      "loss": 0.5835,
+      "step": 3022
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.66321719478694e-07,
+      "loss": 0.4958,
+      "step": 3023
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.633979590812887e-07,
+      "loss": 0.4734,
+      "step": 3024
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.604831758587047e-07,
+      "loss": 0.5135,
+      "step": 3025
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.57577372554735e-07,
+      "loss": 0.4649,
+      "step": 3026
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.546805519047304e-07,
+      "loss": 0.4041,
+      "step": 3027
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.517927166355773e-07,
+      "loss": 0.4139,
+      "step": 3028
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4891386946570694e-07,
+      "loss": 0.468,
+      "step": 3029
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.460440131050947e-07,
+      "loss": 0.4882,
+      "step": 3030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.431831502552431e-07,
+      "loss": 0.5472,
+      "step": 3031
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4033128360919776e-07,
+      "loss": 0.4967,
+      "step": 3032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.37488415851528e-07,
+      "loss": 0.4194,
+      "step": 3033
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.34654549658341e-07,
+      "loss": 0.4865,
+      "step": 3034
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.318296876972594e-07,
+      "loss": 0.449,
+      "step": 3035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.290138326274407e-07,
+      "loss": 0.6219,
+      "step": 3036
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2620698709955446e-07,
+      "loss": 0.5367,
+      "step": 3037
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.234091537557983e-07,
+      "loss": 0.4589,
+      "step": 3038
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.206203352298788e-07,
+      "loss": 0.5529,
+      "step": 3039
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.178405341470193e-07,
+      "loss": 0.5514,
+      "step": 3040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.150697531239578e-07,
+      "loss": 0.4393,
+      "step": 3041
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.123079947689335e-07,
+      "loss": 0.5989,
+      "step": 3042
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.095552616817022e-07,
+      "loss": 0.5827,
+      "step": 3043
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0681155645351357e-07,
+      "loss": 0.4684,
+      "step": 3044
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.040768816671281e-07,
+      "loss": 0.5418,
+      "step": 3045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.013512398968e-07,
+      "loss": 0.4795,
+      "step": 3046
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.986346337082836e-07,
+      "loss": 0.4635,
+      "step": 3047
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.959270656588243e-07,
+      "loss": 0.5483,
+      "step": 3048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9322853829716126e-07,
+      "loss": 0.4639,
+      "step": 3049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9053905416352454e-07,
+      "loss": 0.4553,
+      "step": 3050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8785861578962914e-07,
+      "loss": 0.5411,
+      "step": 3051
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8518722569867885e-07,
+      "loss": 0.5325,
+      "step": 3052
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.825248864053521e-07,
+      "loss": 0.4886,
+      "step": 3053
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.798716004158165e-07,
+      "loss": 0.5559,
+      "step": 3054
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.772273702277118e-07,
+      "loss": 0.4941,
+      "step": 3055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.745921983301559e-07,
+      "loss": 0.5349,
+      "step": 3056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7196608720373785e-07,
+      "loss": 0.4334,
+      "step": 3057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6934903932051903e-07,
+      "loss": 0.455,
+      "step": 3058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6674105714402883e-07,
+      "loss": 0.5138,
+      "step": 3059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.641421431292602e-07,
+      "loss": 0.4444,
+      "step": 3060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6155229972267504e-07,
+      "loss": 0.4996,
+      "step": 3061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5897152936219206e-07,
+      "loss": 0.444,
+      "step": 3062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5639983447719017e-07,
+      "loss": 0.4135,
+      "step": 3063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.538372174885074e-07,
+      "loss": 0.4077,
+      "step": 3064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.512836808084352e-07,
+      "loss": 0.4296,
+      "step": 3065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.487392268407153e-07,
+      "loss": 0.5121,
+      "step": 3066
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462038579805427e-07,
+      "loss": 0.469,
+      "step": 3067
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4367757661455836e-07,
+      "loss": 0.5775,
+      "step": 3068
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4116038512084893e-07,
+      "loss": 0.538,
+      "step": 3069
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.386522858689456e-07,
+      "loss": 0.4301,
+      "step": 3070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.361532812198198e-07,
+      "loss": 0.4838,
+      "step": 3071
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3366337352587987e-07,
+      "loss": 0.4797,
+      "step": 3072
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3118256513097635e-07,
+      "loss": 0.4974,
+      "step": 3073
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2871085837038795e-07,
+      "loss": 0.622,
+      "step": 3074
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.262482555708302e-07,
+      "loss": 0.5094,
+      "step": 3075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2379475905044644e-07,
+      "loss": 0.482,
+      "step": 3076
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.213503711188082e-07,
+      "loss": 0.5309,
+      "step": 3077
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.189150940769126e-07,
+      "loss": 0.4724,
+      "step": 3078
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.164889302171814e-07,
+      "loss": 0.5269,
+      "step": 3079
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1407188182345674e-07,
+      "loss": 0.5875,
+      "step": 3080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1166395117099955e-07,
+      "loss": 0.4316,
+      "step": 3081
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0926514052649235e-07,
+      "loss": 0.4089,
+      "step": 3082
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.068754521480244e-07,
+      "loss": 0.452,
+      "step": 3083
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.044948882851062e-07,
+      "loss": 0.4059,
+      "step": 3084
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.021234511786508e-07,
+      "loss": 0.5737,
+      "step": 3085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9976114306099034e-07,
+      "loss": 0.4393,
+      "step": 3086
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9740796615585157e-07,
+      "loss": 0.4666,
+      "step": 3087
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.950639226783747e-07,
+      "loss": 0.5527,
+      "step": 3088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.927290148350992e-07,
+      "loss": 0.4746,
+      "step": 3089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.904032448239613e-07,
+      "loss": 0.4809,
+      "step": 3090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8808661483430536e-07,
+      "loss": 0.3522,
+      "step": 3091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8577912704685793e-07,
+      "loss": 0.6208,
+      "step": 3092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8348078363375384e-07,
+      "loss": 0.4609,
+      "step": 3093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8119158675850686e-07,
+      "loss": 0.4534,
+      "step": 3094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7891153857603325e-07,
+      "loss": 0.5203,
+      "step": 3095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7664064123262613e-07,
+      "loss": 0.5037,
+      "step": 3096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.743788968659733e-07,
+      "loss": 0.5244,
+      "step": 3097
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.721263076051406e-07,
+      "loss": 0.5158,
+      "step": 3098
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6988287557057847e-07,
+      "loss": 0.4913,
+      "step": 3099
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6764860287411873e-07,
+      "loss": 0.4165,
+      "step": 3100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6542349161896463e-07,
+      "loss": 0.4873,
+      "step": 3101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.63207543899705e-07,
+      "loss": 0.5247,
+      "step": 3102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.610007618022947e-07,
+      "loss": 0.363,
+      "step": 3103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.588031474040642e-07,
+      "loss": 0.4352,
+      "step": 3104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5661470277371534e-07,
+      "loss": 0.469,
+      "step": 3105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5443542997131363e-07,
+      "loss": 0.5185,
+      "step": 3106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.522653310482959e-07,
+      "loss": 0.4807,
+      "step": 3107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5010440804745793e-07,
+      "loss": 0.4924,
+      "step": 3108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4795266300296273e-07,
+      "loss": 0.5239,
+      "step": 3109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4581009794033107e-07,
+      "loss": 0.5025,
+      "step": 3110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4367671487644406e-07,
+      "loss": 0.5258,
+      "step": 3111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.415525158195364e-07,
+      "loss": 0.5099,
+      "step": 3112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3943750276919975e-07,
+      "loss": 0.5151,
+      "step": 3113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.373316777163781e-07,
+      "loss": 0.5576,
+      "step": 3114
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3523504264336805e-07,
+      "loss": 0.5027,
+      "step": 3115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3314759952381305e-07,
+      "loss": 0.4577,
+      "step": 3116
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.310693503227035e-07,
+      "loss": 0.4946,
+      "step": 3117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2900029699637781e-07,
+      "loss": 0.5256,
+      "step": 3118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.269404414925158e-07,
+      "loss": 0.4359,
+      "step": 3119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2488978575013975e-07,
+      "loss": 0.4645,
+      "step": 3120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2284833169961107e-07,
+      "loss": 0.4089,
+      "step": 3121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2081608126263033e-07,
+      "loss": 0.5212,
+      "step": 3122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1879303635223504e-07,
+      "loss": 0.3177,
+      "step": 3123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.167791988727952e-07,
+      "loss": 0.4779,
+      "step": 3124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.147745707200144e-07,
+      "loss": 0.4749,
+      "step": 3125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.127791537809265e-07,
+      "loss": 0.5321,
+      "step": 3126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1079294993389676e-07,
+      "loss": 0.449,
+      "step": 3127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0881596104861622e-07,
+      "loss": 0.5215,
+      "step": 3128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.068481889860996e-07,
+      "loss": 0.4622,
+      "step": 3129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0488963559868848e-07,
+      "loss": 0.4004,
+      "step": 3130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0294030273004473e-07,
+      "loss": 0.5385,
+      "step": 3131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0100019221515388e-07,
+      "loss": 0.4811,
+      "step": 3132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9906930588031392e-07,
+      "loss": 0.5847,
+      "step": 3133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9714764554314537e-07,
+      "loss": 0.5218,
+      "step": 3134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9523521301258118e-07,
+      "loss": 0.4697,
+      "step": 3135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9333201008887026e-07,
+      "loss": 0.6428,
+      "step": 3136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9143803856356946e-07,
+      "loss": 0.4971,
+      "step": 3137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8955330021954933e-07,
+      "loss": 0.4909,
+      "step": 3138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.876777968309862e-07,
+      "loss": 0.447,
+      "step": 3139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.858115301633634e-07,
+      "loss": 0.4366,
+      "step": 3140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8395450197347453e-07,
+      "loss": 0.4627,
+      "step": 3141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8210671400940905e-07,
+      "loss": 0.4425,
+      "step": 3142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8026816801056336e-07,
+      "loss": 0.5113,
+      "step": 3143
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.784388657076308e-07,
+      "loss": 0.51,
+      "step": 3144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.766188088226073e-07,
+      "loss": 0.4437,
+      "step": 3145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.748079990687801e-07,
+      "loss": 0.5336,
+      "step": 3146
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.730064381507379e-07,
+      "loss": 0.4765,
+      "step": 3147
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7121412776435975e-07,
+      "loss": 0.4779,
+      "step": 3148
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6943106959681377e-07,
+      "loss": 0.515,
+      "step": 3149
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6765726532656734e-07,
+      "loss": 0.3702,
+      "step": 3150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6589271662336703e-07,
+      "loss": 0.5189,
+      "step": 3151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6413742514825415e-07,
+      "loss": 0.4941,
+      "step": 3152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.623913925535492e-07,
+      "loss": 0.445,
+      "step": 3153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6065462048286516e-07,
+      "loss": 0.5378,
+      "step": 3154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.58927110571091e-07,
+      "loss": 0.5672,
+      "step": 3155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.572088644443992e-07,
+      "loss": 0.4664,
+      "step": 3156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.554998837202426e-07,
+      "loss": 0.4325,
+      "step": 3157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5380017000734883e-07,
+      "loss": 0.5905,
+      "step": 3158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5210972490572794e-07,
+      "loss": 0.4902,
+      "step": 3159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5042855000665936e-07,
+      "loss": 0.5519,
+      "step": 3160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.487566468926993e-07,
+      "loss": 0.4791,
+      "step": 3161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4709401713767445e-07,
+      "loss": 0.4798,
+      "step": 3162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4544066230668398e-07,
+      "loss": 0.4983,
+      "step": 3163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4379658395609508e-07,
+      "loss": 0.4928,
+      "step": 3164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4216178363354206e-07,
+      "loss": 0.5715,
+      "step": 3165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.405362628779272e-07,
+      "loss": 0.4167,
+      "step": 3166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3892002321941434e-07,
+      "loss": 0.4376,
+      "step": 3167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3731306617943528e-07,
+      "loss": 0.5212,
+      "step": 3168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3571539327068006e-07,
+      "loss": 0.5412,
+      "step": 3169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3412700599710115e-07,
+      "loss": 0.4631,
+      "step": 3170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3254790585390808e-07,
+      "loss": 0.4092,
+      "step": 3171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3097809432756958e-07,
+      "loss": 0.4478,
+      "step": 3172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2941757289581136e-07,
+      "loss": 0.5738,
+      "step": 3173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.278663430276117e-07,
+      "loss": 0.449,
+      "step": 3174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2632440618320473e-07,
+      "loss": 0.5916,
+      "step": 3175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.247917638140761e-07,
+      "loss": 0.637,
+      "step": 3176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2326841736295946e-07,
+      "loss": 0.4558,
+      "step": 3177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2175436826384224e-07,
+      "loss": 0.4301,
+      "step": 3178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2024961794195767e-07,
+      "loss": 0.4551,
+      "step": 3179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.187541678137838e-07,
+      "loss": 0.4659,
+      "step": 3180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.17268019287049e-07,
+      "loss": 0.5281,
+      "step": 3181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1579117376072091e-07,
+      "loss": 0.4202,
+      "step": 3182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1432363262501079e-07,
+      "loss": 0.5512,
+      "step": 3183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.128653972613758e-07,
+      "loss": 0.6474,
+      "step": 3184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1141646904250569e-07,
+      "loss": 0.505,
+      "step": 3185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.099768493323361e-07,
+      "loss": 0.4826,
+      "step": 3186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0854653948603522e-07,
+      "loss": 0.4462,
+      "step": 3187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0712554085000937e-07,
+      "loss": 0.4545,
+      "step": 3188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0571385476190077e-07,
+      "loss": 0.4322,
+      "step": 3189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.043114825505842e-07,
+      "loss": 0.5203,
+      "step": 3190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0291842553616815e-07,
+      "loss": 0.454,
+      "step": 3191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.015346850299892e-07,
+      "loss": 0.4683,
+      "step": 3192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0016026233461873e-07,
+      "loss": 0.5496,
+      "step": 3193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.879515874385071e-08,
+      "loss": 0.3597,
+      "step": 3194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.743937554271499e-08,
+      "loss": 0.3873,
+      "step": 3195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.609291400746068e-08,
+      "loss": 0.4275,
+      "step": 3196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.47557754055628e-08,
+      "loss": 0.4243,
+      "step": 3197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.342796099572559e-08,
+      "loss": 0.3422,
+      "step": 3198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.210947202786924e-08,
+      "loss": 0.5875,
+      "step": 3199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.080030974314203e-08,
+      "loss": 0.4255,
+      "step": 3200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.950047537390483e-08,
+      "loss": 0.5442,
+      "step": 3201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.820997014374666e-08,
+      "loss": 0.5061,
+      "step": 3202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.6928795267468e-08,
+      "loss": 0.4013,
+      "step": 3203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.565695195108969e-08,
+      "loss": 0.5549,
+      "step": 3204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.439444139184628e-08,
+      "loss": 0.4091,
+      "step": 3205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.31412647781893e-08,
+      "loss": 0.4768,
+      "step": 3206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.18974232897829e-08,
+      "loss": 0.5104,
+      "step": 3207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.06629180975016e-08,
+      "loss": 0.5086,
+      "step": 3208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.943775036343581e-08,
+      "loss": 0.6623,
+      "step": 3209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.822192124088079e-08,
+      "loss": 0.4697,
+      "step": 3210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.701543187434656e-08,
+      "loss": 0.5119,
+      "step": 3211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.581828339954578e-08,
+      "loss": 0.5267,
+      "step": 3212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.463047694340253e-08,
+      "loss": 0.5163,
+      "step": 3213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.345201362404575e-08,
+      "loss": 0.5003,
+      "step": 3214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.228289455080805e-08,
+      "loss": 0.4505,
+      "step": 3215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.11231208242269e-08,
+      "loss": 0.4663,
+      "step": 3216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.99726935360423e-08,
+      "loss": 0.4771,
+      "step": 3217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.88316137691969e-08,
+      "loss": 0.4938,
+      "step": 3218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.769988259783256e-08,
+      "loss": 0.4406,
+      "step": 3219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.657750108729267e-08,
+      "loss": 0.421,
+      "step": 3220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.546447029411763e-08,
+      "loss": 0.4785,
+      "step": 3221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.436079126604822e-08,
+      "loss": 0.4863,
+      "step": 3222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.326646504202006e-08,
+      "loss": 0.4902,
+      "step": 3223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.218149265216356e-08,
+      "loss": 0.5207,
+      "step": 3224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.110587511780619e-08,
+      "loss": 0.3697,
+      "step": 3225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.003961345147024e-08,
+      "loss": 0.5182,
+      "step": 3226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.898270865686728e-08,
+      "loss": 0.41,
+      "step": 3227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.793516172890368e-08,
+      "loss": 0.5704,
+      "step": 3228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.689697365367619e-08,
+      "loss": 0.4489,
+      "step": 3229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.586814540847196e-08,
+      "loss": 0.3555,
+      "step": 3230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4848677961766296e-08,
+      "loss": 0.4868,
+      "step": 3231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.383857227322486e-08,
+      "loss": 0.4089,
+      "step": 3232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.283782929369819e-08,
+      "loss": 0.5331,
+      "step": 3233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.184644996522492e-08,
+      "loss": 0.5133,
+      "step": 3234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.086443522102968e-08,
+      "loss": 0.4901,
+      "step": 3235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.989178598551969e-08,
+      "loss": 0.6207,
+      "step": 3236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.892850317428921e-08,
+      "loss": 0.4437,
+      "step": 3237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.797458769411401e-08,
+      "loss": 0.4355,
+      "step": 3238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.703004044295023e-08,
+      "loss": 0.5805,
+      "step": 3239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.609486230993665e-08,
+      "loss": 0.4591,
+      "step": 3240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5169054175395744e-08,
+      "loss": 0.4479,
+      "step": 3241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.425261691082483e-08,
+      "loss": 0.5373,
+      "step": 3242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3345551378901616e-08,
+      "loss": 0.4228,
+      "step": 3243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.244785843348309e-08,
+      "loss": 0.4966,
+      "step": 3244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1559538919602185e-08,
+      "loss": 0.4458,
+      "step": 3245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.068059367346888e-08,
+      "loss": 0.4222,
+      "step": 3246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9811023522468016e-08,
+      "loss": 0.3943,
+      "step": 3247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.895082928516036e-08,
+      "loss": 0.4706,
+      "step": 3248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.810001177128042e-08,
+      "loss": 0.4496,
+      "step": 3249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.725857178173531e-08,
+      "loss": 0.4081,
+      "step": 3250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.642651010860587e-08,
+      "loss": 0.5353,
+      "step": 3251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.560382753514335e-08,
+      "loss": 0.4985,
+      "step": 3252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4790524835770503e-08,
+      "loss": 0.4346,
+      "step": 3253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.398660277608379e-08,
+      "loss": 0.4308,
+      "step": 3254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3192062112842315e-08,
+      "loss": 0.4261,
+      "step": 3255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.240690359398002e-08,
+      "loss": 0.4541,
+      "step": 3256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1631127958597905e-08,
+      "loss": 0.525,
+      "step": 3257
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.086473593696293e-08,
+      "loss": 0.4854,
+      "step": 3258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0107728250510226e-08,
+      "loss": 0.5419,
+      "step": 3259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9360105611838662e-08,
+      "loss": 0.4837,
+      "step": 3260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8621868724717506e-08,
+      "loss": 0.5404,
+      "step": 3261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7893018284076424e-08,
+      "loss": 0.5052,
+      "step": 3262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7173554976009932e-08,
+      "loss": 0.5672,
+      "step": 3263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6463479477778497e-08,
+      "loss": 0.4336,
+      "step": 3264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5762792457802998e-08,
+      "loss": 0.4543,
+      "step": 3265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5071494575666933e-08,
+      "loss": 0.5338,
+      "step": 3266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4389586482117534e-08,
+      "loss": 0.5429,
+      "step": 3267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3717068819059107e-08,
+      "loss": 0.5226,
+      "step": 3268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.305394221956081e-08,
+      "loss": 0.5936,
+      "step": 3269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2400207307848866e-08,
+      "loss": 0.5263,
+      "step": 3270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1755864699307684e-08,
+      "loss": 0.4336,
+      "step": 3271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1120915000484298e-08,
+      "loss": 0.5001,
+      "step": 3272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.049535880908171e-08,
+      "loss": 0.5334,
+      "step": 3273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.987919671395888e-08,
+      "loss": 0.5156,
+      "step": 3274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9272429295132954e-08,
+      "loss": 0.3712,
+      "step": 3275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8675057123779262e-08,
+      "loss": 0.4689,
+      "step": 3276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.808708076222687e-08,
+      "loss": 0.4711,
+      "step": 3277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7508500763960824e-08,
+      "loss": 0.4051,
+      "step": 3278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6939317673621003e-08,
+      "loss": 0.4247,
+      "step": 3279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6379532027001043e-08,
+      "loss": 0.4779,
+      "step": 3280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.582914435104943e-08,
+      "loss": 0.3901,
+      "step": 3281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.528815516386728e-08,
+      "loss": 0.5258,
+      "step": 3282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4756564974709454e-08,
+      "loss": 0.5735,
+      "step": 3283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4234374283982333e-08,
+      "loss": 0.6357,
+      "step": 3284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3721583583242715e-08,
+      "loss": 0.3964,
+      "step": 3285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3218193355203357e-08,
+      "loss": 0.4509,
+      "step": 3286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.272420407372299e-08,
+      "loss": 0.3395,
+      "step": 3287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2239616203812976e-08,
+      "loss": 0.4767,
+      "step": 3288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1764430201636201e-08,
+      "loss": 0.5923,
+      "step": 3289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.129864651450152e-08,
+      "loss": 0.4564,
+      "step": 3290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0842265580871537e-08,
+      "loss": 0.5283,
+      "step": 3291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0395287830354816e-08,
+      "loss": 0.4629,
+      "step": 3292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.957713683707015e-09,
+      "loss": 0.5394,
+      "step": 3293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.529543552836418e-09,
+      "loss": 0.5474,
+      "step": 3294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.11077784079506e-09,
+      "loss": 0.4197,
+      "step": 3295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.70141694178206e-09,
+      "loss": 0.4354,
+      "step": 3296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.301461241148057e-09,
+      "loss": 0.5306,
+      "step": 3297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.910911115383002e-09,
+      "loss": 0.4883,
+      "step": 3298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.529766932129479e-09,
+      "loss": 0.4833,
+      "step": 3299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.15802905017382e-09,
+      "loss": 0.4284,
+      "step": 3300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.795697819445002e-09,
+      "loss": 0.5004,
+      "step": 3301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.4427735810212995e-09,
+      "loss": 0.4568,
+      "step": 3302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.099256667123632e-09,
+      "loss": 0.4454,
+      "step": 3303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.765147401117776e-09,
+      "loss": 0.4949,
+      "step": 3304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.440446097514374e-09,
+      "loss": 0.4706,
+      "step": 3305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.125153061965593e-09,
+      "loss": 0.5555,
+      "step": 3306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.819268591270687e-09,
+      "loss": 0.4943,
+      "step": 3307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.522792973370438e-09,
+      "loss": 0.5605,
+      "step": 3308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.235726487349379e-09,
+      "loss": 0.5554,
+      "step": 3309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.958069403432463e-09,
+      "loss": 0.5643,
+      "step": 3310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6898219829906158e-09,
+      "loss": 0.5013,
+      "step": 3311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.430984478535182e-09,
+      "loss": 0.6608,
+      "step": 3312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1815571337201478e-09,
+      "loss": 0.4866,
+      "step": 3313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9415401833410294e-09,
+      "loss": 0.511,
+      "step": 3314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.710933853335984e-09,
+      "loss": 0.4907,
+      "step": 3315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.489738360782479e-09,
+      "loss": 0.5407,
+      "step": 3316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2779539139006213e-09,
+      "loss": 0.4535,
+      "step": 3317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.07558071205205e-09,
+      "loss": 0.4121,
+      "step": 3318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.882618945737713e-09,
+      "loss": 0.5068,
+      "step": 3319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6990687966000896e-09,
+      "loss": 0.5282,
+      "step": 3320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5249304374231888e-09,
+      "loss": 0.4394,
+      "step": 3321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3602040321292198e-09,
+      "loss": 0.4424,
+      "step": 3322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2048897357819222e-09,
+      "loss": 0.5798,
+      "step": 3323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0589876945843457e-09,
+      "loss": 0.4072,
+      "step": 3324
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.224980458799604e-10,
+      "loss": 0.482,
+      "step": 3325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.95420918151546e-10,
+      "loss": 0.5168,
+      "step": 3326
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.77756431023413e-10,
+      "loss": 0.6163,
+      "step": 3327
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.695046952547412e-10,
+      "loss": 0.4039,
+      "step": 3328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.706658127495711e-10,
+      "loss": 0.5695,
+      "step": 3329
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8123987654792307e-10,
+      "loss": 0.5159,
+      "step": 3330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0122697083023735e-10,
+      "loss": 0.4705,
+      "step": 3331
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3062717091404395e-10,
+      "loss": 0.4933,
+      "step": 3332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6944054326062388e-10,
+      "loss": 0.4576,
+      "step": 3333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.176671454650169e-10,
+      "loss": 0.5231,
+      "step": 3334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.530702626490361e-11,
+      "loss": 0.425,
+      "step": 3335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2360225535054144e-11,
+      "loss": 0.491,
+      "step": 3336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8826774288438487e-11,
+      "loss": 0.4364,
+      "step": 3337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.706694679557089e-12,
+      "loss": 0.4563,
+      "step": 3338
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.4427,
+      "step": 3339
+    },
+    {
+      "epoch": 1.0,
+      "step": 3339,
+      "total_flos": 357485874503680.0,
+      "train_loss": 0.5490463662108249,
+      "train_runtime": 47833.3877,
+      "train_samples_per_second": 6.702,
+      "train_steps_per_second": 0.07
+    }
+  ],
+  "max_steps": 3339,
+  "num_train_epochs": 1,
+  "total_flos": 357485874503680.0,
+  "trial_name": null,
+  "trial_params": null
+}