diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14706 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2446,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-07,
+      "loss": 3.914,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 3.7247,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.108108108108109e-07,
+      "loss": 3.8179,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 3.9695,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3513513513513515e-06,
+      "loss": 4.035,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 3.7208,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8918918918918922e-06,
+      "loss": 3.6515,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 3.4713,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.432432432432433e-06,
+      "loss": 3.5122,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 2.5384,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9729729729729736e-06,
+      "loss": 2.6081,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 2.3291,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.513513513513514e-06,
+      "loss": 1.8066,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 1.5921,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0540540540540545e-06,
+      "loss": 1.4388,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 1.4086,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.594594594594596e-06,
+      "loss": 1.21,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 1.0597,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.135135135135135e-06,
+      "loss": 1.0664,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 1.1034,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.675675675675676e-06,
+      "loss": 1.0144,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 0.9312,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2162162162162164e-06,
+      "loss": 0.966,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 0.8463,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.7567567567567575e-06,
+      "loss": 0.9202,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 0.9266,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.297297297297298e-06,
+      "loss": 0.8965,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 0.8654,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.837837837837838e-06,
+      "loss": 0.8189,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 0.8193,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.378378378378378e-06,
+      "loss": 0.8061,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 0.8261,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.91891891891892e-06,
+      "loss": 0.8132,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 0.778,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.45945945945946e-06,
+      "loss": 0.8129,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 0.7442,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.7654,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 0.8042,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0540540540540541e-05,
+      "loss": 0.7447,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 0.8085,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1081081081081081e-05,
+      "loss": 0.6754,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 0.7993,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1621621621621622e-05,
+      "loss": 0.7228,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 0.7996,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2162162162162164e-05,
+      "loss": 0.7621,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 0.7324,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2702702702702702e-05,
+      "loss": 0.7214,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 0.7916,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3243243243243244e-05,
+      "loss": 0.7493,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 0.7094,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3783783783783784e-05,
+      "loss": 0.6915,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 0.7377,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4324324324324326e-05,
+      "loss": 0.7305,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 0.736,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4864864864864865e-05,
+      "loss": 0.7118,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 0.7813,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.540540540540541e-05,
+      "loss": 0.7843,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 0.7459,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5945945945945947e-05,
+      "loss": 0.7001,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.7714,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.648648648648649e-05,
+      "loss": 0.7398,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 0.7837,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7027027027027028e-05,
+      "loss": 0.7814,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 0.7091,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.756756756756757e-05,
+      "loss": 0.7132,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 0.7662,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8108108108108108e-05,
+      "loss": 0.7957,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 0.8157,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.864864864864865e-05,
+      "loss": 0.6874,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 0.6407,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.918918918918919e-05,
+      "loss": 0.7226,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 0.7492,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.972972972972973e-05,
+      "loss": 0.6559,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.7014,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999991229177605e-05,
+      "loss": 0.7113,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999964916725805e-05,
+      "loss": 0.7679,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999921062690757e-05,
+      "loss": 0.7724,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999859667149386e-05,
+      "loss": 0.6777,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999780730209394e-05,
+      "loss": 0.7554,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999684252009243e-05,
+      "loss": 0.6715,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999570232718174e-05,
+      "loss": 0.7803,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999438672536202e-05,
+      "loss": 0.7306,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999289571694097e-05,
+      "loss": 0.7824,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999912293045341e-05,
+      "loss": 0.7659,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998938749106455e-05,
+      "loss": 0.8113,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998737027976323e-05,
+      "loss": 0.7753,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998517767416856e-05,
+      "loss": 0.7205,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999828096781268e-05,
+      "loss": 0.7603,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998026629579178e-05,
+      "loss": 0.7418,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99977547531625e-05,
+      "loss": 0.7534,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997465339039568e-05,
+      "loss": 0.7847,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997158387718057e-05,
+      "loss": 0.7377,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996833899736406e-05,
+      "loss": 0.7227,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996491875663833e-05,
+      "loss": 0.7066,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999613231610029e-05,
+      "loss": 0.7835,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999575522167651e-05,
+      "loss": 0.7709,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995360593053983e-05,
+      "loss": 0.6646,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994948430924944e-05,
+      "loss": 0.7428,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99945187360124e-05,
+      "loss": 0.637,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994071509070104e-05,
+      "loss": 0.7413,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993606750882566e-05,
+      "loss": 0.7932,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993124462265045e-05,
+      "loss": 0.7157,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999262464406356e-05,
+      "loss": 0.7333,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992107297154872e-05,
+      "loss": 0.7356,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999157242244649e-05,
+      "loss": 0.7008,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991020020876676e-05,
+      "loss": 0.7209,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990450093414437e-05,
+      "loss": 0.736,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989862641059504e-05,
+      "loss": 0.7011,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989257664842382e-05,
+      "loss": 0.6757,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988635165824293e-05,
+      "loss": 0.6541,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99879951450972e-05,
+      "loss": 0.6673,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987337603783806e-05,
+      "loss": 0.7241,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986662543037548e-05,
+      "loss": 0.7017,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998596996404259e-05,
+      "loss": 0.6831,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985259868013836e-05,
+      "loss": 0.7272,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984532256196905e-05,
+      "loss": 0.7158,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998378712986815e-05,
+      "loss": 0.7135,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983024490334645e-05,
+      "loss": 0.697,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982244338934186e-05,
+      "loss": 0.7524,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998144667703529e-05,
+      "loss": 0.6382,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998063150603718e-05,
+      "loss": 0.7254,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99797988273698e-05,
+      "loss": 0.7586,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978948642493817e-05,
+      "loss": 0.7567,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997808095290058e-05,
+      "loss": 0.761,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997719576011217e-05,
+      "loss": 0.6922,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976293065681355e-05,
+      "loss": 0.8705,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975372871191613e-05,
+      "loss": 0.8465,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974435178257114e-05,
+      "loss": 0.6818,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973479988522727e-05,
+      "loss": 0.7871,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997250730366401e-05,
+      "loss": 0.689,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971517125387215e-05,
+      "loss": 0.7687,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997050945542928e-05,
+      "loss": 0.7739,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969484295557814e-05,
+      "loss": 0.6592,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968441647571124e-05,
+      "loss": 0.7345,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967381513298188e-05,
+      "loss": 0.699,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966303894598645e-05,
+      "loss": 0.6508,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996520879336283e-05,
+      "loss": 0.7698,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996409621151172e-05,
+      "loss": 0.6685,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996296615099697e-05,
+      "loss": 0.7438,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961818613800892e-05,
+      "loss": 0.7239,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960653601936454e-05,
+      "loss": 0.7671,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995947111744728e-05,
+      "loss": 0.6988,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995827116240764e-05,
+      "loss": 0.6864,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995705373892246e-05,
+      "loss": 0.7423,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995581884912729e-05,
+      "loss": 0.7609,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954566495188333e-05,
+      "loss": 0.7873,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995329667930243e-05,
+      "loss": 0.6349,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995200940369704e-05,
+      "loss": 0.7028,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950704670630258e-05,
+      "loss": 0.661,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949382482390803e-05,
+      "loss": 0.7158,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994804284129801e-05,
+      "loss": 0.6674,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994668574970183e-05,
+      "loss": 0.7478,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945311209982822e-05,
+      "loss": 0.7583,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943919224552154e-05,
+      "loss": 0.7372,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99425097958516e-05,
+      "loss": 0.7122,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994108292635353e-05,
+      "loss": 0.7396,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939638618560906e-05,
+      "loss": 0.6849,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938176875007284e-05,
+      "loss": 0.643,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99366976982568e-05,
+      "loss": 0.7633,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935201090904177e-05,
+      "loss": 0.6894,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933687055574705e-05,
+      "loss": 0.8241,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993215559492426e-05,
+      "loss": 0.7478,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930606711639266e-05,
+      "loss": 0.7617,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992904040843672e-05,
+      "loss": 0.6509,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992745668806418e-05,
+      "loss": 0.7458,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925855553299755e-05,
+      "loss": 0.7561,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992423700695209e-05,
+      "loss": 0.7045,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922601051860386e-05,
+      "loss": 0.7136,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920947690894376e-05,
+      "loss": 0.6977,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991927692695433e-05,
+      "loss": 0.7501,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917588762971037e-05,
+      "loss": 0.7366,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915883201905824e-05,
+      "loss": 0.6958,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914160246750517e-05,
+      "loss": 0.6859,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912419900527467e-05,
+      "loss": 0.7127,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910662166289523e-05,
+      "loss": 0.7156,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908887047120046e-05,
+      "loss": 0.7304,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990709454613288e-05,
+      "loss": 0.6756,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905284666472374e-05,
+      "loss": 0.6849,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903457411313347e-05,
+      "loss": 0.6512,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901612783861114e-05,
+      "loss": 0.669,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899750787351445e-05,
+      "loss": 0.6937,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897871425050598e-05,
+      "loss": 0.7683,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989597470025528e-05,
+      "loss": 0.7368,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989406061629265e-05,
+      "loss": 0.785,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989212917652034e-05,
+      "loss": 0.6766,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890180384326404e-05,
+      "loss": 0.642,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888214243129348e-05,
+      "loss": 0.6693,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886230756378102e-05,
+      "loss": 0.7124,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884229927552036e-05,
+      "loss": 0.7384,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882211760160924e-05,
+      "loss": 0.7166,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880176257744968e-05,
+      "loss": 0.6555,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878123423874777e-05,
+      "loss": 0.7438,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987605326215135e-05,
+      "loss": 0.666,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873965776206103e-05,
+      "loss": 0.7085,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871860969700824e-05,
+      "loss": 0.7651,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869738846327685e-05,
+      "loss": 0.7598,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867599409809248e-05,
+      "loss": 0.6693,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986544266389843e-05,
+      "loss": 0.6429,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863268612378525e-05,
+      "loss": 0.6636,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986107725906317e-05,
+      "loss": 0.6937,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985886860779636e-05,
+      "loss": 0.6555,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856642662452437e-05,
+      "loss": 0.7356,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854399426936072e-05,
+      "loss": 0.7173,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985213890518227e-05,
+      "loss": 0.6426,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984986110115636e-05,
+      "loss": 0.7005,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984756601885398e-05,
+      "loss": 0.6916,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845253662301085e-05,
+      "loss": 0.7192,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984292403555393e-05,
+      "loss": 0.7189,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984057714269906e-05,
+      "loss": 0.6656,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838212987853312e-05,
+      "loss": 0.729,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835831575163803e-05,
+      "loss": 0.7722,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833432908807927e-05,
+      "loss": 0.724,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831016992993334e-05,
+      "loss": 0.7056,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828583831957935e-05,
+      "loss": 0.6936,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98261334299699e-05,
+      "loss": 0.7692,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823665791327635e-05,
+      "loss": 0.686,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821180920359788e-05,
+      "loss": 0.7909,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818678821425227e-05,
+      "loss": 0.7024,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816159498913044e-05,
+      "loss": 0.6972,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981362295724255e-05,
+      "loss": 0.7299,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981106920086325e-05,
+      "loss": 0.7137,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980849823425486e-05,
+      "loss": 0.6248,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805910061927274e-05,
+      "loss": 0.7232,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803304688420568e-05,
+      "loss": 0.705,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800682118305006e-05,
+      "loss": 0.6893,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798042356181e-05,
+      "loss": 0.7827,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795385406679125e-05,
+      "loss": 0.6216,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792711274460116e-05,
+      "loss": 0.6915,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979001996421483e-05,
+      "loss": 0.743,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978731148066428e-05,
+      "loss": 0.7098,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978458582855958e-05,
+      "loss": 0.6684,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978184301268198e-05,
+      "loss": 0.6526,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779083037842825e-05,
+      "loss": 0.7063,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977630590888357e-05,
+      "loss": 0.6826,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977351163067575e-05,
+      "loss": 0.7599,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770700208120996e-05,
+      "loss": 0.7276,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767871646150998e-05,
+      "loss": 0.6454,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765025949727526e-05,
+      "loss": 0.7429,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976216312384239e-05,
+      "loss": 0.6716,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975928317351747e-05,
+      "loss": 0.7179,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975638610380466e-05,
+      "loss": 0.7334,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975347191978591e-05,
+      "loss": 0.7534,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750540626573164e-05,
+      "loss": 0.7452,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97475922293084e-05,
+      "loss": 0.7468,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744626733163593e-05,
+      "loss": 0.6564,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741644143340707e-05,
+      "loss": 0.662,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738644465071698e-05,
+      "loss": 0.7549,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735627703618494e-05,
+      "loss": 0.6778,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732593864272994e-05,
+      "loss": 0.7245,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729542952357045e-05,
+      "loss": 0.7622,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726474973222453e-05,
+      "loss": 0.7587,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723389932250955e-05,
+      "loss": 0.7281,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720287834854222e-05,
+      "loss": 0.6908,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717168686473845e-05,
+      "loss": 0.6357,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971403249258132e-05,
+      "loss": 0.6483,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710879258678045e-05,
+      "loss": 0.7361,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970770899029532e-05,
+      "loss": 0.7461,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704521692994305e-05,
+      "loss": 0.6225,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701317372366055e-05,
+      "loss": 0.7378,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969809603403147e-05,
+      "loss": 0.6854,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694857683641304e-05,
+      "loss": 0.7554,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969160232687616e-05,
+      "loss": 0.6839,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688329969446473e-05,
+      "loss": 0.7053,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968504061709249e-05,
+      "loss": 0.7743,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681734275584278e-05,
+      "loss": 0.7763,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96784109507217e-05,
+      "loss": 0.7409,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675070648334426e-05,
+      "loss": 0.6643,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671713374281883e-05,
+      "loss": 0.698,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966833913445329e-05,
+      "loss": 0.7453,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664947934767614e-05,
+      "loss": 0.6386,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661539781173582e-05,
+      "loss": 0.7071,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965811467964965e-05,
+      "loss": 0.6803,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654672636204014e-05,
+      "loss": 0.6056,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965121365687458e-05,
+      "loss": 0.6341,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647737747728972e-05,
+      "loss": 0.6479,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644244914864502e-05,
+      "loss": 0.7001,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640735164408176e-05,
+      "loss": 0.7345,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637208502516673e-05,
+      "loss": 0.7122,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633664935376335e-05,
+      "loss": 0.6863,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630104469203165e-05,
+      "loss": 0.6755,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626527110242808e-05,
+      "loss": 0.7279,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622932864770538e-05,
+      "loss": 0.7658,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619321739091247e-05,
+      "loss": 0.8178,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615693739539452e-05,
+      "loss": 0.7593,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961204887247926e-05,
+      "loss": 0.7031,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608387144304363e-05,
+      "loss": 0.7344,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604708561438033e-05,
+      "loss": 0.7371,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960101313033312e-05,
+      "loss": 0.6481,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597300857472e-05,
+      "loss": 0.7669,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959357174936663e-05,
+      "loss": 0.7682,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589825812558468e-05,
+      "loss": 0.6921,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95860630536185e-05,
+      "loss": 0.7272,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582283479147237e-05,
+      "loss": 0.6797,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578487095774666e-05,
+      "loss": 0.6798,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574673910160263e-05,
+      "loss": 0.6218,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957084392899299e-05,
+      "loss": 0.6786,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566997158991265e-05,
+      "loss": 0.6568,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956313360690295e-05,
+      "loss": 0.6483,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559253279505354e-05,
+      "loss": 0.6623,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955535618360521e-05,
+      "loss": 0.6965,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551442326038664e-05,
+      "loss": 0.7012,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547511713671264e-05,
+      "loss": 0.6993,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543564353397953e-05,
+      "loss": 0.6872,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953960025214305e-05,
+      "loss": 0.6712,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953561941686024e-05,
+      "loss": 0.7013,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9531621854532562e-05,
+      "loss": 0.7142,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95276075721724e-05,
+      "loss": 0.6824,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523576576821463e-05,
+      "loss": 0.7435,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519528875550783e-05,
+      "loss": 0.731,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515464475460692e-05,
+      "loss": 0.6508,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951138338368082e-05,
+      "loss": 0.77,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507285607370065e-05,
+      "loss": 0.666,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503171153716606e-05,
+      "loss": 0.6168,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949904002993787e-05,
+      "loss": 0.6891,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949489224328053e-05,
+      "loss": 0.658,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490727801020485e-05,
+      "loss": 0.698,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486546710462847e-05,
+      "loss": 0.7834,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482348978941947e-05,
+      "loss": 0.7289,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478134613821286e-05,
+      "loss": 0.6546,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473903622493554e-05,
+      "loss": 0.6534,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469656012380617e-05,
+      "loss": 0.7529,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946539179093347e-05,
+      "loss": 0.7308,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946111096563226e-05,
+      "loss": 0.7221,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945681354398627e-05,
+      "loss": 0.6508,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945249953353387e-05,
+      "loss": 0.6877,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944816894184255e-05,
+      "loss": 0.7452,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443821776508885e-05,
+      "loss": 0.6759,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943945804515851e-05,
+      "loss": 0.6791,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435077755446124e-05,
+      "loss": 0.7525,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430680915055492e-05,
+      "loss": 0.7387,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942626753169938e-05,
+      "loss": 0.6926,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421837613119597e-05,
+      "loss": 0.7244,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417391167086946e-05,
+      "loss": 0.709,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941292820140122e-05,
+      "loss": 0.674,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408448723891203e-05,
+      "loss": 0.6841,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940395274241463e-05,
+      "loss": 0.7307,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399440264858192e-05,
+      "loss": 0.7572,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394911299137522e-05,
+      "loss": 0.7091,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390365853197163e-05,
+      "loss": 0.6941,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938580393501058e-05,
+      "loss": 0.705,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938122555258013e-05,
+      "loss": 0.6568,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376630713937043e-05,
+      "loss": 0.6852,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372019427141424e-05,
+      "loss": 0.7676,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367391700282228e-05,
+      "loss": 0.6645,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362747541477255e-05,
+      "loss": 0.6581,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358086958873116e-05,
+      "loss": 0.6912,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935340996064524e-05,
+      "loss": 0.5606,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348716554997854e-05,
+      "loss": 0.6755,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344006750163962e-05,
+      "loss": 0.7071,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339280554405336e-05,
+      "loss": 0.6998,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93345379760125e-05,
+      "loss": 0.6462,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329779023304724e-05,
+      "loss": 0.6674,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325003704629982e-05,
+      "loss": 0.756,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320212028364976e-05,
+      "loss": 0.7485,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9315404002915093e-05,
+      "loss": 0.6598,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310579636714402e-05,
+      "loss": 0.7027,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930573893822563e-05,
+      "loss": 0.6232,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300881915940163e-05,
+      "loss": 0.7015,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296008578378015e-05,
+      "loss": 0.8247,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929111893408782e-05,
+      "loss": 0.7135,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286212991646823e-05,
+      "loss": 0.6709,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928129075966085e-05,
+      "loss": 0.6455,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276352246764305e-05,
+      "loss": 0.6988,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271397461620154e-05,
+      "loss": 0.6794,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266426412919905e-05,
+      "loss": 0.7772,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261439109383594e-05,
+      "loss": 0.6561,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925643555975977e-05,
+      "loss": 0.6804,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925141577282549e-05,
+      "loss": 0.7263,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924637975738628e-05,
+      "loss": 0.7349,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241327522276133e-05,
+      "loss": 0.6582,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923625907635751e-05,
+      "loss": 0.625,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92311744285213e-05,
+      "loss": 0.6996,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226073587686805e-05,
+      "loss": 0.7124,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922095656280174e-05,
+      "loss": 0.7546,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921582336284221e-05,
+      "loss": 0.7235,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210673996812694e-05,
+      "loss": 0.7486,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920550847374602e-05,
+      "loss": 0.7058,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200326802703374e-05,
+      "loss": 0.7482,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919512899277425e-05,
+      "loss": 0.6735,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189915053076472e-05,
+      "loss": 0.6708,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184684992756142e-05,
+      "loss": 0.675,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179438820987645e-05,
+      "loss": 0.6281,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917417654697363e-05,
+      "loss": 0.6828,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168898179944994e-05,
+      "loss": 0.6829,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163603729160854e-05,
+      "loss": 0.7762,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158293203908552e-05,
+      "loss": 0.6461,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152966613503627e-05,
+      "loss": 0.6883,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914762396728979e-05,
+      "loss": 0.7082,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914226527463892e-05,
+      "loss": 0.7298,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136890544951046e-05,
+      "loss": 0.6838,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131499787654334e-05,
+      "loss": 0.6955,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912609301220505e-05,
+      "loss": 0.6945,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912067022808757e-05,
+      "loss": 0.6849,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115231444814356e-05,
+      "loss": 0.672,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910977667192592e-05,
+      "loss": 0.7544,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104305918990832e-05,
+      "loss": 0.6818,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098819195605697e-05,
+      "loss": 0.6401,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093316511395128e-05,
+      "loss": 0.8019,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908779787601174e-05,
+      "loss": 0.6541,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908226329913612e-05,
+      "loss": 0.7466,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907671279047683e-05,
+      "loss": 0.6918,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071146359770384e-05,
+      "loss": 0.6416,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065564016781204e-05,
+      "loss": 0.7059,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059965771301644e-05,
+      "loss": 0.6869,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054351633151945e-05,
+      "loss": 0.6925,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048721612180232e-05,
+      "loss": 0.6839,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043075718262485e-05,
+      "loss": 0.7487,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037413961302534e-05,
+      "loss": 0.6022,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031736351232025e-05,
+      "loss": 0.6931,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026042898010428e-05,
+      "loss": 0.6718,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020333611624993e-05,
+      "loss": 0.7025,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014608502090744e-05,
+      "loss": 0.6595,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008867579450472e-05,
+      "loss": 0.6941,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003110853774694e-05,
+      "loss": 0.7194,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997338335161656e-05,
+      "loss": 0.7182,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89915500337373e-05,
+      "loss": 0.6861,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985745959655268e-05,
+      "loss": 0.7109,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979926123096858e-05,
+      "loss": 0.686,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974090534271013e-05,
+      "loss": 0.6813,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896823920341432e-05,
+      "loss": 0.6497,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962372140790984e-05,
+      "loss": 0.6652,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895648935669278e-05,
+      "loss": 0.6616,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950590861439098e-05,
+      "loss": 0.6319,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944676665376858e-05,
+      "loss": 0.6901,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938746778880535e-05,
+      "loss": 0.6906,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932801212352124e-05,
+      "loss": 0.6569,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926839976221128e-05,
+      "loss": 0.6184,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920863080944534e-05,
+      "loss": 0.6378,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914870537006805e-05,
+      "loss": 0.6651,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908862354919843e-05,
+      "loss": 0.7356,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902838545222987e-05,
+      "loss": 0.6437,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896799118482995e-05,
+      "loss": 0.7163,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889074408529401e-05,
+      "loss": 0.6933,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888467345627756e-05,
+      "loss": 0.7284,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887858724208252e-05,
+      "loss": 0.6641,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872485453385124e-05,
+      "loss": 0.6717,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88663681008889e-05,
+      "loss": 0.7557,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860235195324695e-05,
+      "loss": 0.707,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854086747450636e-05,
+      "loss": 0.6398,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847922768052105e-05,
+      "loss": 0.7226,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841743267941746e-05,
+      "loss": 0.6508,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835548257959413e-05,
+      "loss": 0.7045,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882933774897217e-05,
+      "loss": 0.7008,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823111751874277e-05,
+      "loss": 0.7322,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816870277587155e-05,
+      "loss": 0.7538,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881061333705937e-05,
+      "loss": 0.6776,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804340941266638e-05,
+      "loss": 0.6401,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879805310121176e-05,
+      "loss": 0.6541,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879174982792465e-05,
+      "loss": 0.8082,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785431132462278e-05,
+      "loss": 0.5973,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8779097025908684e-05,
+      "loss": 0.6512,
+      "step": 451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772747519374927e-05,
+      "loss": 0.7673,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766382623999094e-05,
+      "loss": 0.8369,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760002350946244e-05,
+      "loss": 0.6954,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875360671140844e-05,
+      "loss": 0.7501,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747195716604675e-05,
+      "loss": 0.6257,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740769377780893e-05,
+      "loss": 0.821,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873432770620995e-05,
+      "loss": 0.6851,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727870713191593e-05,
+      "loss": 0.673,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872139841005246e-05,
+      "loss": 0.74,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714910808146024e-05,
+      "loss": 0.807,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708407918852608e-05,
+      "loss": 0.6937,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701889753579356e-05,
+      "loss": 0.6517,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695356323760197e-05,
+      "loss": 0.7829,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868880764085584e-05,
+      "loss": 0.7234,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682243716353754e-05,
+      "loss": 0.7447,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675664561768144e-05,
+      "loss": 0.6787,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669070188639924e-05,
+      "loss": 0.6403,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866246060853672e-05,
+      "loss": 0.6881,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655835833052808e-05,
+      "loss": 0.7213,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649195873809143e-05,
+      "loss": 0.7371,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642540742453302e-05,
+      "loss": 0.6393,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863587045065949e-05,
+      "loss": 0.5962,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629185010128478e-05,
+      "loss": 0.6924,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862248443258764e-05,
+      "loss": 0.618,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615768729790893e-05,
+      "loss": 0.6643,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8609037913518676e-05,
+      "loss": 0.7445,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602291995577957e-05,
+      "loss": 0.7095,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595530987802177e-05,
+      "loss": 0.6985,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8588754902051262e-05,
+      "loss": 0.7014,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581963750211577e-05,
+      "loss": 0.7189,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857515754419592e-05,
+      "loss": 0.719,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568336295943498e-05,
+      "loss": 0.6508,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561500017419902e-05,
+      "loss": 0.7039,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554648720617086e-05,
+      "loss": 0.7891,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547782417553355e-05,
+      "loss": 0.7368,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540901120273332e-05,
+      "loss": 0.6893,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534004840847943e-05,
+      "loss": 0.6616,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527093591374397e-05,
+      "loss": 0.7322,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520167383976168e-05,
+      "loss": 0.749,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513226230802958e-05,
+      "loss": 0.6647,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850627014403069e-05,
+      "loss": 0.6962,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499299135861488e-05,
+      "loss": 0.6822,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849231321852364e-05,
+      "loss": 0.6693,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485312404271604e-05,
+      "loss": 0.6608,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478296705385953e-05,
+      "loss": 0.6914,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471266134173377e-05,
+      "loss": 0.6816,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464220702966656e-05,
+      "loss": 0.7194,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457160424124637e-05,
+      "loss": 0.6852,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450085310032206e-05,
+      "loss": 0.6765,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8442995373100282e-05,
+      "loss": 0.751,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435890625765776e-05,
+      "loss": 0.7049,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428771080491582e-05,
+      "loss": 0.6654,
+      "step": 503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421636749766563e-05,
+      "loss": 0.7253,
+      "step": 504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414487646105496e-05,
+      "loss": 0.7771,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407323782049093e-05,
+      "loss": 0.7697,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840014517016395e-05,
+      "loss": 0.7785,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392951823042525e-05,
+      "loss": 0.6777,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385743753303144e-05,
+      "loss": 0.6817,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378520973589937e-05,
+      "loss": 0.6524,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837128349657285e-05,
+      "loss": 0.663,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364031334947612e-05,
+      "loss": 0.5611,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356764501435704e-05,
+      "loss": 0.6298,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349483008784346e-05,
+      "loss": 0.6959,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8342186869766475e-05,
+      "loss": 0.6743,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833487609718072e-05,
+      "loss": 0.7166,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832755070385138e-05,
+      "loss": 0.7456,
+      "step": 517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320210702628397e-05,
+      "loss": 0.6953,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312856106387343e-05,
+      "loss": 0.7597,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305486928029383e-05,
+      "loss": 0.7405,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298103180481276e-05,
+      "loss": 0.6394,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290704876695325e-05,
+      "loss": 0.6471,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828329202964937e-05,
+      "loss": 0.7252,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275864652346772e-05,
+      "loss": 0.6842,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268422757816366e-05,
+      "loss": 0.673,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.826096635911246e-05,
+      "loss": 0.6223,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253495469314803e-05,
+      "loss": 0.7223,
+      "step": 527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246010101528566e-05,
+      "loss": 0.7398,
+      "step": 528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238510268884316e-05,
+      "loss": 0.7027,
+      "step": 529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823099598453799e-05,
+      "loss": 0.7069,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223467261670885e-05,
+      "loss": 0.6365,
+      "step": 531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215924113489613e-05,
+      "loss": 0.6707,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208366553226095e-05,
+      "loss": 0.7026,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820079459413754e-05,
+      "loss": 0.6887,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193208249506408e-05,
+      "loss": 0.638,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185607532640396e-05,
+      "loss": 0.638,
+      "step": 536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817799245687241e-05,
+      "loss": 0.7042,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170363035560544e-05,
+      "loss": 0.6734,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162719282088064e-05,
+      "loss": 0.6363,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155061209863368e-05,
+      "loss": 0.6991,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814738883231997e-05,
+      "loss": 0.7348,
+      "step": 541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139702162916485e-05,
+      "loss": 0.8031,
+      "step": 542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132001215136595e-05,
+      "loss": 0.6623,
+      "step": 543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124286002489034e-05,
+      "loss": 0.6902,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116556538507547e-05,
+      "loss": 0.6894,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108812836750887e-05,
+      "loss": 0.6904,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810105491080278e-05,
+      "loss": 0.6872,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093282774271908e-05,
+      "loss": 0.6051,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085496440791874e-05,
+      "loss": 0.6974,
+      "step": 549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807769592402119e-05,
+      "loss": 0.6279,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806988123764324e-05,
+      "loss": 0.6083,
+      "step": 551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062052395366275e-05,
+      "loss": 0.7348,
+      "step": 552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805420941092337e-05,
+      "loss": 0.6704,
+      "step": 553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046352298072408e-05,
+      "loss": 0.7317,
+      "step": 554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038481070596057e-05,
+      "loss": 0.6518,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803059574230175e-05,
+      "loss": 0.6152,
+      "step": 556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022696327021645e-05,
+      "loss": 0.6646,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014782838612616e-05,
+      "loss": 0.6791,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006855290956226e-05,
+      "loss": 0.6175,
+      "step": 559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998913697958693e-05,
+      "loss": 0.6689,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990958073550882e-05,
+      "loss": 0.6883,
+      "step": 561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982988431688266e-05,
+      "loss": 0.726,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79750047863509e-05,
+      "loss": 0.7097,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967007151543425e-05,
+      "loss": 0.6863,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958995541294997e-05,
+      "loss": 0.6354,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950969969659303e-05,
+      "loss": 0.5956,
+      "step": 566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942930450714515e-05,
+      "loss": 0.6322,
+      "step": 567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934876998563263e-05,
+      "loss": 0.6341,
+      "step": 568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926809627332642e-05,
+      "loss": 0.6763,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918728351174136e-05,
+      "loss": 0.6754,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910633184263643e-05,
+      "loss": 0.6639,
+      "step": 571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790252414080141e-05,
+      "loss": 0.7094,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894401235012028e-05,
+      "loss": 0.7198,
+      "step": 573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788626448114442e-05,
+      "loss": 0.6438,
+      "step": 574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7878113893471786e-05,
+      "loss": 0.6869,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7869949486291604e-05,
+      "loss": 0.668,
+      "step": 576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7861771273925576e-05,
+      "loss": 0.7005,
+      "step": 577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853579270719635e-05,
+      "loss": 0.7021,
+      "step": 578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845373491043905e-05,
+      "loss": 0.6996,
+      "step": 579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837153949292674e-05,
+      "loss": 0.6826,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828920659884364e-05,
+      "loss": 0.7017,
+      "step": 581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782067363726153e-05,
+      "loss": 0.7321,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812412895890792e-05,
+      "loss": 0.6876,
+      "step": 583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804138450262862e-05,
+      "loss": 0.6382,
+      "step": 584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779585031489247e-05,
+      "loss": 0.6685,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787548504318372e-05,
+      "loss": 0.6965,
+      "step": 586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779233033103306e-05,
+      "loss": 0.6898,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770903915833986e-05,
+      "loss": 0.7186,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762561167121042e-05,
+      "loss": 0.5961,
+      "step": 589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775420480159903e-05,
+      "loss": 0.706,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745834833926395e-05,
+      "loss": 0.6752,
+      "step": 591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737451278785435e-05,
+      "loss": 0.5933,
+      "step": 592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772905415088228e-05,
+      "loss": 0.6891,
+      "step": 593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772064346494688e-05,
+      "loss": 0.7117,
+      "step": 594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712219235732954e-05,
+      "loss": 0.6617,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703781478017995e-05,
+      "loss": 0.6343,
+      "step": 596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769533020660321e-05,
+      "loss": 0.5853,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768686543631352e-05,
+      "loss": 0.6899,
+      "step": 598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767838718199753e-05,
+      "loss": 0.6233,
+      "step": 599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669895458527487e-05,
+      "loss": 0.7243,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766139028079927e-05,
+      "loss": 0.6416,
+      "step": 601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765287166373237e-05,
+      "loss": 0.6862,
+      "step": 602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644339622269827e-05,
+      "loss": 0.6542,
+      "step": 603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635794171378257e-05,
+      "loss": 0.7114,
+      "step": 604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762723532604778e-05,
+      "loss": 0.8002,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761866310129202e-05,
+      "loss": 0.6922,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610077512148073e-05,
+      "loss": 0.7517,
+      "step": 607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760147857367647e-05,
+      "loss": 0.6715,
+      "step": 608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7592866300961163e-05,
+      "loss": 0.6672,
+      "step": 609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584240709109498e-05,
+      "loss": 0.695,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757560181325218e-05,
+      "loss": 0.6148,
+      "step": 611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566949628543252e-05,
+      "loss": 0.6443,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558284170160073e-05,
+      "loss": 0.7007,
+      "step": 613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754960545330328e-05,
+      "loss": 0.6778,
+      "step": 614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754091349319677e-05,
+      "loss": 0.6917,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753220830508767e-05,
+      "loss": 0.656,
+      "step": 616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7523489904246312e-05,
+      "loss": 0.71,
+      "step": 617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7514758305966206e-05,
+      "loss": 0.668,
+      "step": 618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506013525564012e-05,
+      "loss": 0.6859,
+      "step": 619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497255578379514e-05,
+      "loss": 0.7139,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488484479775585e-05,
+      "loss": 0.6734,
+      "step": 621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479700245138184e-05,
+      "loss": 0.6442,
+      "step": 622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470902889876295e-05,
+      "loss": 0.6452,
+      "step": 623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746209242942193e-05,
+      "loss": 0.6672,
+      "step": 624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745326887923009e-05,
+      "loss": 0.6883,
+      "step": 625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7444432254778725e-05,
+      "loss": 0.7069,
+      "step": 626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435582571568736e-05,
+      "loss": 0.7285,
+      "step": 627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426719845123914e-05,
+      "loss": 0.6627,
+      "step": 628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417844090990947e-05,
+      "loss": 0.667,
+      "step": 629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408955324739363e-05,
+      "loss": 0.6565,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7400053561961523e-05,
+      "loss": 0.7884,
+      "step": 631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391138818272578e-05,
+      "loss": 0.6268,
+      "step": 632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738221110931046e-05,
+      "loss": 0.7447,
+      "step": 633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737327045073584e-05,
+      "loss": 0.7412,
+      "step": 634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73643168582321e-05,
+      "loss": 0.6906,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355350347505312e-05,
+      "loss": 0.6412,
+      "step": 636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346370934284214e-05,
+      "loss": 0.715,
+      "step": 637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7337378634320173e-05,
+      "loss": 0.7714,
+      "step": 638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328373463387166e-05,
+      "loss": 0.6954,
+      "step": 639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319355437281737e-05,
+      "loss": 0.6106,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731032457182299e-05,
+      "loss": 0.6845,
+      "step": 641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730128088285255e-05,
+      "loss": 0.7755,
+      "step": 642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7292224386234534e-05,
+      "loss": 0.5845,
+      "step": 643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283155097855525e-05,
+      "loss": 0.674,
+      "step": 644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727407303362455e-05,
+      "loss": 0.6503,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264978209473035e-05,
+      "loss": 0.6771,
+      "step": 646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725587064135481e-05,
+      "loss": 0.6127,
+      "step": 647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724675034524604e-05,
+      "loss": 0.7146,
+      "step": 648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7237617337145224e-05,
+      "loss": 0.6379,
+      "step": 649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228471633073164e-05,
+      "loss": 0.6424,
+      "step": 650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721931324907293e-05,
+      "loss": 0.6366,
+      "step": 651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210142201209825e-05,
+      "loss": 0.712,
+      "step": 652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200958505571386e-05,
+      "loss": 0.6458,
+      "step": 653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719176217826732e-05,
+      "loss": 0.67,
+      "step": 654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71825532354295e-05,
+      "loss": 0.6944,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173331693211922e-05,
+      "loss": 0.7089,
+      "step": 656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7164097567790693e-05,
+      "loss": 0.6424,
+      "step": 657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154850875363987e-05,
+      "loss": 0.6755,
+      "step": 658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145591632152025e-05,
+      "loss": 0.6924,
+      "step": 659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136319854397037e-05,
+      "loss": 0.6625,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712703555836325e-05,
+      "loss": 0.6687,
+      "step": 661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117738760336846e-05,
+      "loss": 0.6443,
+      "step": 662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7108429476625937e-05,
+      "loss": 0.7328,
+      "step": 663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099107723560537e-05,
+      "loss": 0.5872,
+      "step": 664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708977351749254e-05,
+      "loss": 0.6595,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080426874795666e-05,
+      "loss": 0.6828,
+      "step": 666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.6646,
+      "step": 667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061696345119304e-05,
+      "loss": 0.7373,
+      "step": 668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052312490996237e-05,
+      "loss": 0.7879,
+      "step": 669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042916265957107e-05,
+      "loss": 0.6031,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703350768648443e-05,
+      "loss": 0.6827,
+      "step": 671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702408676908241e-05,
+      "loss": 0.6262,
+      "step": 672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701465353027688e-05,
+      "loss": 0.7102,
+      "step": 673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005207986615293e-05,
+      "loss": 0.6267,
+      "step": 674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699575015466669e-05,
+      "loss": 0.7388,
+      "step": 675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698628005102166e-05,
+      "loss": 0.7303,
+      "step": 676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976797692292325e-05,
+      "loss": 0.6721,
+      "step": 677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967303095112297e-05,
+      "loss": 0.6575,
+      "step": 678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695779627613667e-05,
+      "loss": 0.7179,
+      "step": 679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948277252041957e-05,
+      "loss": 0.6756,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69387460395261e-05,
+      "loss": 0.6721,
+      "step": 681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6929202655308414e-05,
+      "loss": 0.6859,
+      "step": 682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691964711612956e-05,
+      "loss": 0.6197,
+      "step": 683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691007943875153e-05,
+      "loss": 0.6698,
+      "step": 684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900499639957596e-05,
+      "loss": 0.6647,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689090773655231e-05,
+      "loss": 0.7312,
+      "step": 686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688130374536144e-05,
+      "loss": 0.6452,
+      "step": 687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871687683231975e-05,
+      "loss": 0.6965,
+      "step": 688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686205956703206e-05,
+      "loss": 0.6312,
+      "step": 689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852419413651003e-05,
+      "loss": 0.6391,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6842767239999214e-05,
+      "loss": 0.7419,
+      "step": 691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833103063008194e-05,
+      "loss": 0.6088,
+      "step": 692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823426899630498e-05,
+      "loss": 0.625,
+      "step": 693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681373876683971e-05,
+      "loss": 0.7497,
+      "step": 694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680403868163041e-05,
+      "loss": 0.7308,
+      "step": 695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794326661018136e-05,
+      "loss": 0.698,
+      "step": 696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784602722039376e-05,
+      "loss": 0.6865,
+      "step": 697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774866881751518e-05,
+      "loss": 0.7627,
+      "step": 698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765119157232824e-05,
+      "loss": 0.6541,
+      "step": 699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755359565582408e-05,
+      "loss": 0.6987,
+      "step": 700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745588123920197e-05,
+      "loss": 0.6267,
+      "step": 701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6735804849386914e-05,
+      "loss": 0.6864,
+      "step": 702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726009759144023e-05,
+      "loss": 0.6408,
+      "step": 703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716202870373726e-05,
+      "loss": 0.7263,
+      "step": 704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670638420027892e-05,
+      "loss": 0.6588,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6696553766083167e-05,
+      "loss": 0.7192,
+      "step": 706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668671158503067e-05,
+      "loss": 0.7439,
+      "step": 707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667685767438622e-05,
+      "loss": 0.6251,
+      "step": 708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6666992051435215e-05,
+      "loss": 0.5684,
+      "step": 709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657114733483564e-05,
+      "loss": 0.7041,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647225737857716e-05,
+      "loss": 0.6618,
+      "step": 711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637325081904595e-05,
+      "loss": 0.6626,
+      "step": 712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662741278299158e-05,
+      "loss": 0.7749,
+      "step": 713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6617488858506478e-05,
+      "loss": 0.6481,
+      "step": 714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607553325857473e-05,
+      "loss": 0.6862,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659760620247313e-05,
+      "loss": 0.5548,
+      "step": 716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6587647505802342e-05,
+      "loss": 0.6451,
+      "step": 717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65776772533143e-05,
+      "loss": 0.6363,
+      "step": 718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567695462498465e-05,
+      "loss": 0.631,
+      "step": 719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557702150864538e-05,
+      "loss": 0.6345,
+      "step": 720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547697335942438e-05,
+      "loss": 0.71,
+      "step": 721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537681035282247e-05,
+      "loss": 0.6811,
+      "step": 722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652765326645421e-05,
+      "loss": 0.6662,
+      "step": 723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517614047048683e-05,
+      "loss": 0.651,
+      "step": 724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6507563394676106e-05,
+      "loss": 0.6302,
+      "step": 725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497501326966974e-05,
+      "loss": 0.7978,
+      "step": 726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487427861571815e-05,
+      "loss": 0.6405,
+      "step": 727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477343016161138e-05,
+      "loss": 0.645,
+      "step": 728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646724680842543e-05,
+      "loss": 0.6541,
+      "step": 729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457139256075084e-05,
+      "loss": 0.6866,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447020376840423e-05,
+      "loss": 0.6497,
+      "step": 731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436890188471622e-05,
+      "loss": 0.6977,
+      "step": 732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6426748708738696e-05,
+      "loss": 0.6914,
+      "step": 733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6416595955431468e-05,
+      "loss": 0.6886,
+      "step": 734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640643194635954e-05,
+      "loss": 0.671,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396256699352252e-05,
+      "loss": 0.7131,
+      "step": 736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386070232258667e-05,
+      "loss": 0.6452,
+      "step": 737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375872562947516e-05,
+      "loss": 0.6659,
+      "step": 738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6365663709307193e-05,
+      "loss": 0.6776,
+      "step": 739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635544368924571e-05,
+      "loss": 0.6921,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634521252069065e-05,
+      "loss": 0.7061,
+      "step": 741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6334970221589182e-05,
+      "loss": 0.6026,
+      "step": 742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632471680990797e-05,
+      "loss": 0.6983,
+      "step": 743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314452303633193e-05,
+      "loss": 0.6635,
+      "step": 744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304176720770482e-05,
+      "loss": 0.6981,
+      "step": 745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6293890079344892e-05,
+      "loss": 0.662,
+      "step": 746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283592397400895e-05,
+      "loss": 0.6246,
+      "step": 747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273283693002312e-05,
+      "loss": 0.6972,
+      "step": 748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262963984232307e-05,
+      "loss": 0.6705,
+      "step": 749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625263328919335e-05,
+      "loss": 0.6856,
+      "step": 750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624229162600717e-05,
+      "loss": 0.6791,
+      "step": 751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6231939012814758e-05,
+      "loss": 0.7502,
+      "step": 752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221575467776292e-05,
+      "loss": 0.7194,
+      "step": 753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6211201009071134e-05,
+      "loss": 0.7124,
+      "step": 754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200815654897798e-05,
+      "loss": 0.6352,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190419423473897e-05,
+      "loss": 0.6864,
+      "step": 756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180012333036133e-05,
+      "loss": 0.687,
+      "step": 757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169594401840255e-05,
+      "loss": 0.6668,
+      "step": 758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159165648161026e-05,
+      "loss": 0.6435,
+      "step": 759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148726090292196e-05,
+      "loss": 0.5814,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6138275746546467e-05,
+      "loss": 0.6166,
+      "step": 761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6127814635255462e-05,
+      "loss": 0.6986,
+      "step": 762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6117342774769687e-05,
+      "loss": 0.6934,
+      "step": 763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106860183458514e-05,
+      "loss": 0.7462,
+      "step": 764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096366879710127e-05,
+      "loss": 0.6056,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608586288193151e-05,
+      "loss": 0.7394,
+      "step": 766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075348208548395e-05,
+      "loss": 0.683,
+      "step": 767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064822878005262e-05,
+      "loss": 0.7596,
+      "step": 768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605428690876526e-05,
+      "loss": 0.6305,
+      "step": 769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6043740319310218e-05,
+      "loss": 0.6892,
+      "step": 770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6033183128140585e-05,
+      "loss": 0.6026,
+      "step": 771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602261535377542e-05,
+      "loss": 0.7555,
+      "step": 772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6012037014752322e-05,
+      "loss": 0.683,
+      "step": 773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600144812962745e-05,
+      "loss": 0.6646,
+      "step": 774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990848716975447e-05,
+      "loss": 0.6627,
+      "step": 775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980238795389424e-05,
+      "loss": 0.6583,
+      "step": 776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969618383480926e-05,
+      "loss": 0.6748,
+      "step": 777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595898749987991e-05,
+      "loss": 0.6598,
+      "step": 778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948346163234694e-05,
+      "loss": 0.7278,
+      "step": 779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5937694392211923e-05,
+      "loss": 0.6913,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927032205496565e-05,
+      "loss": 0.6754,
+      "step": 781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916359621791847e-05,
+      "loss": 0.6236,
+      "step": 782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905676659819232e-05,
+      "loss": 0.6992,
+      "step": 783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894983338318396e-05,
+      "loss": 0.6583,
+      "step": 784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884279676047186e-05,
+      "loss": 0.6383,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587356569178158e-05,
+      "loss": 0.7256,
+      "step": 786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862841404315675e-05,
+      "loss": 0.6713,
+      "step": 787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585210683246163e-05,
+      "loss": 0.7189,
+      "step": 788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5841361995049655e-05,
+      "loss": 0.6596,
+      "step": 789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830606910927956e-05,
+      "loss": 0.7346,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819841598962722e-05,
+      "loss": 0.8004,
+      "step": 791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5809066078038082e-05,
+      "loss": 0.6389,
+      "step": 792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798280367056072e-05,
+      "loss": 0.6678,
+      "step": 793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57874844849366e-05,
+      "loss": 0.6333,
+      "step": 794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5776678450617426e-05,
+      "loss": 0.6448,
+      "step": 795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5765862283054105e-05,
+      "loss": 0.6582,
+      "step": 796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755036001219974e-05,
+      "loss": 0.777,
+      "step": 797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5744199624106115e-05,
+      "loss": 0.7066,
+      "step": 798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5733353170721316e-05,
+      "loss": 0.7088,
+      "step": 799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572249666009204e-05,
+      "loss": 0.7351,
+      "step": 800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571163011126239e-05,
+      "loss": 0.7026,
+      "step": 801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570075354329408e-05,
+      "loss": 0.6308,
+      "step": 802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689866975266404e-05,
+      "loss": 0.6274,
+      "step": 803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5678970426276186e-05,
+      "loss": 0.623,
+      "step": 804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668063915437768e-05,
+      "loss": 0.6752,
+      "step": 805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657147461882965e-05,
+      "loss": 0.7502,
+      "step": 806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5646221084761032e-05,
+      "loss": 0.6367,
+      "step": 807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5635284803238632e-05,
+      "loss": 0.5682,
+      "step": 808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56243386364998e-05,
+      "loss": 0.6257,
+      "step": 809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5613382603745918e-05,
+      "loss": 0.6056,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560241672419566e-05,
+      "loss": 0.6451,
+      "step": 811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559144101708499e-05,
+      "loss": 0.6364,
+      "step": 812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55804555016671e-05,
+      "loss": 0.6185,
+      "step": 813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556946019721239e-05,
+      "loss": 0.6231,
+      "step": 814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555845512300844e-05,
+      "loss": 0.6298,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5547440298359948e-05,
+      "loss": 0.7487,
+      "step": 816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536415742588737e-05,
+      "loss": 0.6154,
+      "step": 817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5525381475033692e-05,
+      "loss": 0.62,
+      "step": 818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551433751505073e-05,
+      "loss": 0.7088,
+      "step": 819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550328388201277e-05,
+      "loss": 0.7701,
+      "step": 820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549222059530971e-05,
+      "loss": 0.7539,
+      "step": 821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5481147674348366e-05,
+      "loss": 0.6378,
+      "step": 822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547006513855247e-05,
+      "loss": 0.7523,
+      "step": 823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545897300736261e-05,
+      "loss": 0.6113,
+      "step": 824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5447871300236207e-05,
+      "loss": 0.6059,
+      "step": 825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5436760036647485e-05,
+      "loss": 0.6869,
+      "step": 826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425639236087425e-05,
+      "loss": 0.6053,
+      "step": 827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541450891806374e-05,
+      "loss": 0.6317,
+      "step": 828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403369102100837e-05,
+      "loss": 0.6418,
+      "step": 829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539221980773979e-05,
+      "loss": 0.6198,
+      "step": 830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5381061054538294e-05,
+      "loss": 0.7284,
+      "step": 831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369892862070636e-05,
+      "loss": 0.7141,
+      "step": 832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5358715249927663e-05,
+      "loss": 0.6593,
+      "step": 833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347528237716742e-05,
+      "loss": 0.6567,
+      "step": 834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533633184506174e-05,
+      "loss": 0.7172,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325126091602965e-05,
+      "loss": 0.6276,
+      "step": 836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313910996997156e-05,
+      "loss": 0.6423,
+      "step": 837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5302686580917428e-05,
+      "loss": 0.7415,
+      "step": 838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291452863053257e-05,
+      "loss": 0.6518,
+      "step": 839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528020986311043e-05,
+      "loss": 0.7427,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526895760081102e-05,
+      "loss": 0.6607,
+      "step": 841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525769609589335e-05,
+      "loss": 0.6617,
+      "step": 842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5246425368111944e-05,
+      "loss": 0.7278,
+      "step": 843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523514543723751e-05,
+      "loss": 0.6573,
+      "step": 844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223856323056909e-05,
+      "loss": 0.6599,
+      "step": 845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212558045373106e-05,
+      "loss": 0.6769,
+      "step": 846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201250624005133e-05,
+      "loss": 0.7313,
+      "step": 847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189934078788069e-05,
+      "loss": 0.5937,
+      "step": 848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178608429572996e-05,
+      "loss": 0.7483,
+      "step": 849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5167273696226965e-05,
+      "loss": 0.6856,
+      "step": 850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155929898632959e-05,
+      "loss": 0.5891,
+      "step": 851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144577056689872e-05,
+      "loss": 0.659,
+      "step": 852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513321519031245e-05,
+      "loss": 0.7016,
+      "step": 853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.512184431943128e-05,
+      "loss": 0.6616,
+      "step": 854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110464463992736e-05,
+      "loss": 0.71,
+      "step": 855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099075643958959e-05,
+      "loss": 0.705,
+      "step": 856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5087677879307811e-05,
+      "loss": 0.6308,
+      "step": 857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076271190032845e-05,
+      "loss": 0.6952,
+      "step": 858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064855596143271e-05,
+      "loss": 0.6808,
+      "step": 859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5053431117663922e-05,
+      "loss": 0.6553,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5041997774635206e-05,
+      "loss": 0.6436,
+      "step": 861
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030555587113091e-05,
+      "loss": 0.6478,
+      "step": 862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019104575169055e-05,
+      "loss": 0.6779,
+      "step": 863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007644758890059e-05,
+      "loss": 0.672,
+      "step": 864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49961761583785e-05,
+      "loss": 0.6943,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4984698793752193e-05,
+      "loss": 0.6371,
+      "step": 866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973212685144324e-05,
+      "loss": 0.6548,
+      "step": 867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961717852703417e-05,
+      "loss": 0.7141,
+      "step": 868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4950214316593295e-05,
+      "loss": 0.6136,
+      "step": 869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938702096993057e-05,
+      "loss": 0.7227,
+      "step": 870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927181214097028e-05,
+      "loss": 0.6827,
+      "step": 871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915651688114733e-05,
+      "loss": 0.7099,
+      "step": 872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904113539270852e-05,
+      "loss": 0.6496,
+      "step": 873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48925667878052e-05,
+      "loss": 0.6346,
+      "step": 874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4881011453972675e-05,
+      "loss": 0.6683,
+      "step": 875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486944755804324e-05,
+      "loss": 0.6045,
+      "step": 876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857875120301866e-05,
+      "loss": 0.6831,
+      "step": 877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484629416104851e-05,
+      "loss": 0.561,
+      "step": 878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834704700598084e-05,
+      "loss": 0.6503,
+      "step": 879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823106759280404e-05,
+      "loss": 0.6729,
+      "step": 880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811500357440166e-05,
+      "loss": 0.6476,
+      "step": 881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799885515436912e-05,
+      "loss": 0.6282,
+      "step": 882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4788262253644983e-05,
+      "loss": 0.6954,
+      "step": 883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776630592453492e-05,
+      "loss": 0.6332,
+      "step": 884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764990552266287e-05,
+      "loss": 0.6766,
+      "step": 885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753342153501913e-05,
+      "loss": 0.692,
+      "step": 886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741685416593574e-05,
+      "loss": 0.6567,
+      "step": 887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730020361989108e-05,
+      "loss": 0.631,
+      "step": 888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718347010150936e-05,
+      "loss": 0.6719,
+      "step": 889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470666538155604e-05,
+      "loss": 0.6715,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694975496695918e-05,
+      "loss": 0.6915,
+      "step": 891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683277376076548e-05,
+      "loss": 0.7192,
+      "step": 892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.467157104021836e-05,
+      "loss": 0.6995,
+      "step": 893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659856509656194e-05,
+      "loss": 0.6105,
+      "step": 894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648133804939257e-05,
+      "loss": 0.6545,
+      "step": 895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636402946631108e-05,
+      "loss": 0.6753,
+      "step": 896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624663955309597e-05,
+      "loss": 0.6135,
+      "step": 897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612916851566851e-05,
+      "loss": 0.6606,
+      "step": 898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601161656009218e-05,
+      "loss": 0.569,
+      "step": 899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589398389257246e-05,
+      "loss": 0.7102,
+      "step": 900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4577627071945642e-05,
+      "loss": 0.6408,
+      "step": 901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565847724723225e-05,
+      "loss": 0.6975,
+      "step": 902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554060368252916e-05,
+      "loss": 0.6743,
+      "step": 903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454226502321167e-05,
+      "loss": 0.6644,
+      "step": 904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530461710290467e-05,
+      "loss": 0.6601,
+      "step": 905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518650450194261e-05,
+      "loss": 0.6432,
+      "step": 906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4506831263641939e-05,
+      "loss": 0.6417,
+      "step": 907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495004171366302e-05,
+      "loss": 0.6346,
+      "step": 908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4483169194114014e-05,
+      "loss": 0.6178,
+      "step": 909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471326352645573e-05,
+      "loss": 0.7311,
+      "step": 910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445947566773527e-05,
+      "loss": 0.7122,
+      "step": 911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447617160171154e-05,
+      "loss": 0.6745,
+      "step": 912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4435750850755001e-05,
+      "loss": 0.7504,
+      "step": 913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423876760302266e-05,
+      "loss": 0.7315,
+      "step": 914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411994909642059e-05,
+      "loss": 0.7143,
+      "step": 915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400105319617102e-05,
+      "loss": 0.7352,
+      "step": 916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4388208011083688e-05,
+      "loss": 0.6552,
+      "step": 917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376303004911654e-05,
+      "loss": 0.6561,
+      "step": 918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364390321984334e-05,
+      "loss": 0.6391,
+      "step": 919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352469983198542e-05,
+      "loss": 0.5917,
+      "step": 920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340542009464513e-05,
+      "loss": 0.7887,
+      "step": 921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328606421705868e-05,
+      "loss": 0.7109,
+      "step": 922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316663240859595e-05,
+      "loss": 0.6414,
+      "step": 923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304712487875999e-05,
+      "loss": 0.6761,
+      "step": 924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429275418371866e-05,
+      "loss": 0.6634,
+      "step": 925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280788349364414e-05,
+      "loss": 0.6568,
+      "step": 926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4268815005803305e-05,
+      "loss": 0.7302,
+      "step": 927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4256834174038545e-05,
+      "loss": 0.6097,
+      "step": 928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244845875086486e-05,
+      "loss": 0.7543,
+      "step": 929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4232850129976573e-05,
+      "loss": 0.7901,
+      "step": 930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220846959751315e-05,
+      "loss": 0.6433,
+      "step": 931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420883638546625e-05,
+      "loss": 0.6497,
+      "step": 932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41968184281899e-05,
+      "loss": 0.7194,
+      "step": 933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184793109003734e-05,
+      "loss": 0.5924,
+      "step": 934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172760449002148e-05,
+      "loss": 0.6818,
+      "step": 935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160720469292402e-05,
+      "loss": 0.6699,
+      "step": 936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41486731909946e-05,
+      "loss": 0.6364,
+      "step": 937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4136618635241655e-05,
+      "loss": 0.6482,
+      "step": 938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124556823179234e-05,
+      "loss": 0.6275,
+      "step": 939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4112487775965741e-05,
+      "loss": 0.7372,
+      "step": 940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.410041151477227e-05,
+      "loss": 0.6012,
+      "step": 941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4088328060782573e-05,
+      "loss": 0.7349,
+      "step": 942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4076237435193011e-05,
+      "loss": 0.6688,
+      "step": 943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4064139659212534e-05,
+      "loss": 0.7441,
+      "step": 944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4052034754062626e-05,
+      "loss": 0.7159,
+      "step": 945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403992274097729e-05,
+      "loss": 0.6244,
+      "step": 946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402780364120298e-05,
+      "loss": 0.6445,
+      "step": 947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4015677475998595e-05,
+      "loss": 0.6686,
+      "step": 948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003544266635419e-05,
+      "loss": 0.6924,
+      "step": 949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991404034397102e-05,
+      "loss": 0.6661,
+      "step": 950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3979256800579605e-05,
+      "loss": 0.6329,
+      "step": 951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967102586491179e-05,
+      "loss": 0.716,
+      "step": 952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954941413452309e-05,
+      "loss": 0.7043,
+      "step": 953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3942773302795697e-05,
+      "loss": 0.6532,
+      "step": 954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930598275866205e-05,
+      "loss": 0.6309,
+      "step": 955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3918416354020836e-05,
+      "loss": 0.5948,
+      "step": 956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3906227558628684e-05,
+      "loss": 0.6993,
+      "step": 957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894031911070904e-05,
+      "loss": 0.6112,
+      "step": 958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881829432740665e-05,
+      "loss": 0.6158,
+      "step": 959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869620145043123e-05,
+      "loss": 0.6842,
+      "step": 960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3857404069395373e-05,
+      "loss": 0.6148,
+      "step": 961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845181227226423e-05,
+      "loss": 0.6307,
+      "step": 962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383295163997715e-05,
+      "loss": 0.729,
+      "step": 963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820715329100256e-05,
+      "loss": 0.6533,
+      "step": 964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808472316060251e-05,
+      "loss": 0.6902,
+      "step": 965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3796222622333389e-05,
+      "loss": 0.6106,
+      "step": 966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783966269407647e-05,
+      "loss": 0.6843,
+      "step": 967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771703278782689e-05,
+      "loss": 0.6475,
+      "step": 968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375943367196981e-05,
+      "loss": 0.6113,
+      "step": 969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747157470491923e-05,
+      "loss": 0.617,
+      "step": 970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734874695883504e-05,
+      "loss": 0.6739,
+      "step": 971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372258536969056e-05,
+      "loss": 0.5792,
+      "step": 972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371028951347059e-05,
+      "loss": 0.6993,
+      "step": 973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3697987148792546e-05,
+      "loss": 0.689,
+      "step": 974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3685678297236805e-05,
+      "loss": 0.6088,
+      "step": 975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673362980395115e-05,
+      "loss": 0.6996,
+      "step": 976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661041219870563e-05,
+      "loss": 0.6402,
+      "step": 977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364871303727755e-05,
+      "loss": 0.6131,
+      "step": 978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3636378454241728e-05,
+      "loss": 0.6024,
+      "step": 979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624037492399991e-05,
+      "loss": 0.6856,
+      "step": 980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611690173400414e-05,
+      "loss": 0.6063,
+      "step": 981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599336518902228e-05,
+      "loss": 0.6971,
+      "step": 982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358697655057577e-05,
+      "loss": 0.6286,
+      "step": 983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574610290102462e-05,
+      "loss": 0.6562,
+      "step": 984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3562237759174755e-05,
+      "loss": 0.6645,
+      "step": 985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3549858979496104e-05,
+      "loss": 0.656,
+      "step": 986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537473972780924e-05,
+      "loss": 0.6358,
+      "step": 987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525082760754557e-05,
+      "loss": 0.612,
+      "step": 988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512685365153226e-05,
+      "loss": 0.6456,
+      "step": 989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3500281807724003e-05,
+      "loss": 0.6466,
+      "step": 990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3487872110224765e-05,
+      "loss": 0.6977,
+      "step": 991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475456294424163e-05,
+      "loss": 0.6613,
+      "step": 992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463034382101581e-05,
+      "loss": 0.6815,
+      "step": 993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450606395047094e-05,
+      "loss": 0.6862,
+      "step": 994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343817235506144e-05,
+      "loss": 0.5749,
+      "step": 995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3425732283955968e-05,
+      "loss": 0.6317,
+      "step": 996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341328620355261e-05,
+      "loss": 0.6255,
+      "step": 997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400834135683836e-05,
+      "loss": 0.6544,
+      "step": 998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388376102192622e-05,
+      "loss": 0.6195,
+      "step": 999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3375912124932406e-05,
+      "loss": 0.7076,
+      "step": 1000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363442225767055e-05,
+      "loss": 0.6262,
+      "step": 1001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350966426570825e-05,
+      "loss": 0.5793,
+      "step": 1002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333848474922832e-05,
+      "loss": 0.7112,
+      "step": 1003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3325997215634457e-05,
+      "loss": 0.6114,
+      "step": 1004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313503847694418e-05,
+      "loss": 0.633,
+      "step": 1005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330100466732363e-05,
+      "loss": 0.7571,
+      "step": 1006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3288499696447708e-05,
+      "loss": 0.7234,
+      "step": 1007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3275988957002429e-05,
+      "loss": 0.648,
+      "step": 1008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3263472470933682e-05,
+      "loss": 0.7128,
+      "step": 1009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3250950260197455e-05,
+      "loss": 0.6688,
+      "step": 1010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3238422346759758e-05,
+      "loss": 0.6612,
+      "step": 1011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225888752596614e-05,
+      "loss": 0.654,
+      "step": 1012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213349499694004e-05,
+      "loss": 0.6246,
+      "step": 1013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200804610047842e-05,
+      "loss": 0.6233,
+      "step": 1014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318825410566393e-05,
+      "loss": 0.6381,
+      "step": 1015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175698008557914e-05,
+      "loss": 0.626,
+      "step": 1016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3163136340755254e-05,
+      "loss": 0.6816,
+      "step": 1017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150569124291186e-05,
+      "loss": 0.5663,
+      "step": 1018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3137996381210672e-05,
+      "loss": 0.7454,
+      "step": 1019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125418133568369e-05,
+      "loss": 0.6833,
+      "step": 1020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3112834403428593e-05,
+      "loss": 0.6682,
+      "step": 1021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100245212865279e-05,
+      "loss": 0.5654,
+      "step": 1022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087650583961936e-05,
+      "loss": 0.6166,
+      "step": 1023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075050538811611e-05,
+      "loss": 0.7112,
+      "step": 1024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3062445099516863e-05,
+      "loss": 0.7039,
+      "step": 1025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3049834288189702e-05,
+      "loss": 0.7618,
+      "step": 1026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037218126951567e-05,
+      "loss": 0.6539,
+      "step": 1027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3024596637933277e-05,
+      "loss": 0.6936,
+      "step": 1028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011969843275003e-05,
+      "loss": 0.6558,
+      "step": 1029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999337765126217e-05,
+      "loss": 0.6517,
+      "step": 1030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2986700425645663e-05,
+      "loss": 0.6512,
+      "step": 1031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974057847001306e-05,
+      "loss": 0.6663,
+      "step": 1032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2961410051370322e-05,
+      "loss": 0.7137,
+      "step": 1033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948757060939019e-05,
+      "loss": 0.5942,
+      "step": 1034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293609889790282e-05,
+      "loss": 0.6702,
+      "step": 1035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923435584466228e-05,
+      "loss": 0.7645,
+      "step": 1036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2910767142842777e-05,
+      "loss": 0.6491,
+      "step": 1037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2898093595254998e-05,
+      "loss": 0.5949,
+      "step": 1038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885414963934374e-05,
+      "loss": 0.5756,
+      "step": 1039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872731271121315e-05,
+      "loss": 0.6602,
+      "step": 1040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28600425390651e-05,
+      "loss": 0.6815,
+      "step": 1041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2847348790023858e-05,
+      "loss": 0.6687,
+      "step": 1042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283465004626451e-05,
+      "loss": 0.5653,
+      "step": 1043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2821946330062738e-05,
+      "loss": 0.7003,
+      "step": 1044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2809237663702951e-05,
+      "loss": 0.5824,
+      "step": 1045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796524069478243e-05,
+      "loss": 0.7416,
+      "step": 1046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783805569690343e-05,
+      "loss": 0.6764,
+      "step": 1047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27710821866496e-05,
+      "loss": 0.6823,
+      "step": 1048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275835394267492e-05,
+      "loss": 0.6791,
+      "step": 1049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2745620860093726e-05,
+      "loss": 0.6758,
+      "step": 1050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273288296124195e-05,
+      "loss": 0.7278,
+      "step": 1051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720140268463958e-05,
+      "loss": 0.568,
+      "step": 1052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2707392804112529e-05,
+      "loss": 0.6155,
+      "step": 1053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694640590548808e-05,
+      "loss": 0.6709,
+      "step": 1054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681883650142283e-05,
+      "loss": 0.6508,
+      "step": 1055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669122005270724e-05,
+      "loss": 0.757,
+      "step": 1056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265635567832015e-05,
+      "loss": 0.6936,
+      "step": 1057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643584691684802e-05,
+      "loss": 0.6403,
+      "step": 1058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2630809067767094e-05,
+      "loss": 0.64,
+      "step": 1059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618028828977563e-05,
+      "loss": 0.6297,
+      "step": 1060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605243997734857e-05,
+      "loss": 0.7047,
+      "step": 1061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259245459646567e-05,
+      "loss": 0.6494,
+      "step": 1062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2579660647604715e-05,
+      "loss": 0.6567,
+      "step": 1063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2566862173594689e-05,
+      "loss": 0.6357,
+      "step": 1064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554059196886212e-05,
+      "loss": 0.6891,
+      "step": 1065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541251739937814e-05,
+      "loss": 0.6514,
+      "step": 1066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528439825215884e-05,
+      "loss": 0.7016,
+      "step": 1067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515623475194623e-05,
+      "loss": 0.5765,
+      "step": 1068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502802712356017e-05,
+      "loss": 0.6344,
+      "step": 1069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489977559189796e-05,
+      "loss": 0.6569,
+      "step": 1070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2477148038193392e-05,
+      "loss": 0.7095,
+      "step": 1071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2464314171871888e-05,
+      "loss": 0.6326,
+      "step": 1072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451475982737996e-05,
+      "loss": 0.6426,
+      "step": 1073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438633493312016e-05,
+      "loss": 0.666,
+      "step": 1074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2425786726121783e-05,
+      "loss": 0.7223,
+      "step": 1075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241293570370264e-05,
+      "loss": 0.7513,
+      "step": 1076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2400080448597396e-05,
+      "loss": 0.7125,
+      "step": 1077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2387220983356283e-05,
+      "loss": 0.6276,
+      "step": 1078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374357330536919e-05,
+      "loss": 0.5754,
+      "step": 1079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2361489512704264e-05,
+      "loss": 0.6219,
+      "step": 1080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234861755243059e-05,
+      "loss": 0.7364,
+      "step": 1081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2335741472295426e-05,
+      "loss": 0.6935,
+      "step": 1082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2322861294885542e-05,
+      "loss": 0.6673,
+      "step": 1083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230997704279488e-05,
+      "loss": 0.667,
+      "step": 1084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2297088738624548e-05,
+      "loss": 0.6083,
+      "step": 1085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284196404982746e-05,
+      "loss": 0.6952,
+      "step": 1086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2271300064484746e-05,
+      "loss": 0.5991,
+      "step": 1087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2258399739752848e-05,
+      "loss": 0.6693,
+      "step": 1088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245495453416352e-05,
+      "loss": 0.6277,
+      "step": 1089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223258722811149e-05,
+      "loss": 0.6063,
+      "step": 1090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2219675086481418e-05,
+      "loss": 0.6917,
+      "step": 1091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2206759051176151e-05,
+      "loss": 0.6114,
+      "step": 1092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2193839144852546e-05,
+      "loss": 0.6676,
+      "step": 1093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218091539017424e-05,
+      "loss": 0.6606,
+      "step": 1094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167987809811625e-05,
+      "loss": 0.7116,
+      "step": 1095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155056426441803e-05,
+      "loss": 0.624,
+      "step": 1096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2142121262748545e-05,
+      "loss": 0.6998,
+      "step": 1097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212918234142226e-05,
+      "loss": 0.6688,
+      "step": 1098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2116239685159941e-05,
+      "loss": 0.6496,
+      "step": 1099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2103293316665137e-05,
+      "loss": 0.6768,
+      "step": 1100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2090343258647912e-05,
+      "loss": 0.6663,
+      "step": 1101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077389533824789e-05,
+      "loss": 0.642,
+      "step": 1102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064432164918738e-05,
+      "loss": 0.6956,
+      "step": 1103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051471174659116e-05,
+      "loss": 0.627,
+      "step": 1104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038506585781626e-05,
+      "loss": 0.7477,
+      "step": 1105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025538421028293e-05,
+      "loss": 0.6547,
+      "step": 1106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201256670314741e-05,
+      "loss": 0.6947,
+      "step": 1107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999591454893511e-05,
+      "loss": 0.7223,
+      "step": 1108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986612699027305e-05,
+      "loss": 0.5807,
+      "step": 1109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973630458315667e-05,
+      "loss": 0.6823,
+      "step": 1110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960644755531587e-05,
+      "loss": 0.6151,
+      "step": 1111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194765561345412e-05,
+      "loss": 0.6339,
+      "step": 1112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1934663054868358e-05,
+      "loss": 0.6318,
+      "step": 1113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921667102565384e-05,
+      "loss": 0.6354,
+      "step": 1114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908667779342243e-05,
+      "loss": 0.6517,
+      "step": 1115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1895665108001879e-05,
+      "loss": 0.6742,
+      "step": 1116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1882659111353118e-05,
+      "loss": 0.6612,
+      "step": 1117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1869649812210618e-05,
+      "loss": 0.7326,
+      "step": 1118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185663723339483e-05,
+      "loss": 0.6571,
+      "step": 1119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843621397731954e-05,
+      "loss": 0.6196,
+      "step": 1120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830602328053911e-05,
+      "loss": 0.7858,
+      "step": 1121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817580047198287e-05,
+      "loss": 0.7536,
+      "step": 1122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180455457800831e-05,
+      "loss": 0.7371,
+      "step": 1123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179152594333279e-05,
+      "loss": 0.6093,
+      "step": 1124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778494166026096e-05,
+      "loss": 0.6953,
+      "step": 1125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765459268948111e-05,
+      "loss": 0.5785,
+      "step": 1126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752421274964188e-05,
+      "loss": 0.6453,
+      "step": 1127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1739380206945108e-05,
+      "loss": 0.5911,
+      "step": 1128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1726336087767054e-05,
+      "loss": 0.6226,
+      "step": 1129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1713288940311562e-05,
+      "loss": 0.6772,
+      "step": 1130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700238787465463e-05,
+      "loss": 0.6431,
+      "step": 1131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168718565212088e-05,
+      "loss": 0.6819,
+      "step": 1132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674129557175156e-05,
+      "loss": 0.6575,
+      "step": 1133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1661070525530827e-05,
+      "loss": 0.6557,
+      "step": 1134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648008580095587e-05,
+      "loss": 0.671,
+      "step": 1135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634943743782235e-05,
+      "loss": 0.6451,
+      "step": 1136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1621876039508638e-05,
+      "loss": 0.6994,
+      "step": 1137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160880549019771e-05,
+      "loss": 0.6015,
+      "step": 1138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1595732118777332e-05,
+      "loss": 0.5878,
+      "step": 1139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582655948180357e-05,
+      "loss": 0.6532,
+      "step": 1140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569577001344532e-05,
+      "loss": 0.7016,
+      "step": 1141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556495301212485e-05,
+      "loss": 0.6226,
+      "step": 1142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154341087073167e-05,
+      "loss": 0.7173,
+      "step": 1143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1530323732854326e-05,
+      "loss": 0.6313,
+      "step": 1144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1517233910537453e-05,
+      "loss": 0.6334,
+      "step": 1145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1504141426742744e-05,
+      "loss": 0.6283,
+      "step": 1146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491046304436575e-05,
+      "loss": 0.7231,
+      "step": 1147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147794856658994e-05,
+      "loss": 0.5591,
+      "step": 1148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146484823617843e-05,
+      "loss": 0.597,
+      "step": 1149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1451745336182173e-05,
+      "loss": 0.6484,
+      "step": 1150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438639889585818e-05,
+      "loss": 0.6414,
+      "step": 1151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425531919378469e-05,
+      "loss": 0.6337,
+      "step": 1152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412421448553665e-05,
+      "loss": 0.5903,
+      "step": 1153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399308500109326e-05,
+      "loss": 0.6858,
+      "step": 1154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138619309704772e-05,
+      "loss": 0.6119,
+      "step": 1155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1373075262375421e-05,
+      "loss": 0.657,
+      "step": 1156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359955019103273e-05,
+      "loss": 0.6326,
+      "step": 1157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346832390246334e-05,
+      "loss": 0.6183,
+      "step": 1158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333707398823856e-05,
+      "loss": 0.68,
+      "step": 1159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132058006785924e-05,
+      "loss": 0.6826,
+      "step": 1160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1307450420379971e-05,
+      "loss": 0.6321,
+      "step": 1161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1294318479417618e-05,
+      "loss": 0.7396,
+      "step": 1162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281184268007766e-05,
+      "loss": 0.6435,
+      "step": 1163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268047809189976e-05,
+      "loss": 0.7927,
+      "step": 1164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254909126007765e-05,
+      "loss": 0.7204,
+      "step": 1165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241768241508537e-05,
+      "loss": 0.7077,
+      "step": 1166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228625178743572e-05,
+      "loss": 0.6306,
+      "step": 1167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215479960767958e-05,
+      "loss": 0.767,
+      "step": 1168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1202332610640574e-05,
+      "loss": 0.5404,
+      "step": 1169
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118918315142403e-05,
+      "loss": 0.6272,
+      "step": 1170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1176031606184645e-05,
+      "loss": 0.6591,
+      "step": 1171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162877997992389e-05,
+      "loss": 0.6481,
+      "step": 1172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149722349920855e-05,
+      "loss": 0.6648,
+      "step": 1173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136564685047213e-05,
+      "loss": 0.6006,
+      "step": 1174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123405026452176e-05,
+      "loss": 0.6927,
+      "step": 1175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1110243397219945e-05,
+      "loss": 0.6562,
+      "step": 1176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097079820438181e-05,
+      "loss": 0.6103,
+      "step": 1177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083914319197967e-05,
+      "loss": 0.6468,
+      "step": 1178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1070746916593756e-05,
+      "loss": 0.6356,
+      "step": 1179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1057577635723337e-05,
+      "loss": 0.5703,
+      "step": 1180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1044406499687797e-05,
+      "loss": 0.6586,
+      "step": 1181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031233531591471e-05,
+      "loss": 0.7869,
+      "step": 1182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018058754541915e-05,
+      "loss": 0.6808,
+      "step": 1183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1004882191649857e-05,
+      "loss": 0.6611,
+      "step": 1184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099170386602915e-05,
+      "loss": 0.7283,
+      "step": 1185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0978523800796747e-05,
+      "loss": 0.6204,
+      "step": 1186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0965342019072654e-05,
+      "loss": 0.664,
+      "step": 1187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0952158543979878e-05,
+      "loss": 0.7295,
+      "step": 1188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938973398644407e-05,
+      "loss": 0.7179,
+      "step": 1189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925786606195153e-05,
+      "loss": 0.546,
+      "step": 1190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912598189763919e-05,
+      "loss": 0.6948,
+      "step": 1191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0899408172485357e-05,
+      "loss": 0.6008,
+      "step": 1192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886216577496926e-05,
+      "loss": 0.6768,
+      "step": 1193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873023427938855e-05,
+      "loss": 0.6157,
+      "step": 1194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0859828746954098e-05,
+      "loss": 0.6702,
+      "step": 1195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846632557688295e-05,
+      "loss": 0.6785,
+      "step": 1196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083343488328973e-05,
+      "loss": 0.6163,
+      "step": 1197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08202357469093e-05,
+      "loss": 0.6327,
+      "step": 1198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0807035171700455e-05,
+      "loss": 0.6351,
+      "step": 1199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793833180819183e-05,
+      "loss": 0.5806,
+      "step": 1200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780629797423938e-05,
+      "loss": 0.714,
+      "step": 1201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0767425044675634e-05,
+      "loss": 0.7203,
+      "step": 1202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0754218945737575e-05,
+      "loss": 0.701,
+      "step": 1203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0741011523775433e-05,
+      "loss": 0.6727,
+      "step": 1204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0727802801957198e-05,
+      "loss": 0.5973,
+      "step": 1205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714592803453138e-05,
+      "loss": 0.7249,
+      "step": 1206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701381551435767e-05,
+      "loss": 0.5716,
+      "step": 1207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688169069079793e-05,
+      "loss": 0.6326,
+      "step": 1208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.067495537956208e-05,
+      "loss": 0.6047,
+      "step": 1209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0661740506061616e-05,
+      "loss": 0.617,
+      "step": 1210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0648524471759463e-05,
+      "loss": 0.6481,
+      "step": 1211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635307299838715e-05,
+      "loss": 0.5981,
+      "step": 1212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062208901348447e-05,
+      "loss": 0.6404,
+      "step": 1213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0608869635883776e-05,
+      "loss": 0.7058,
+      "step": 1214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595649190225593e-05,
+      "loss": 0.6684,
+      "step": 1215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582427699700759e-05,
+      "loss": 0.6638,
+      "step": 1216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569205187501943e-05,
+      "loss": 0.7179,
+      "step": 1217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555981676823606e-05,
+      "loss": 0.6787,
+      "step": 1218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542757190861959e-05,
+      "loss": 0.6338,
+      "step": 1219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529531752814928e-05,
+      "loss": 0.7203,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0516305385882103e-05,
+      "loss": 0.6431,
+      "step": 1221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0503078113264715e-05,
+      "loss": 0.6501,
+      "step": 1222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0489849958165567e-05,
+      "loss": 0.678,
+      "step": 1223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476620943789021e-05,
+      "loss": 0.6206,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046339109334095e-05,
+      "loss": 0.6784,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450160430028679e-05,
+      "loss": 0.7266,
+      "step": 1226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0436928977060973e-05,
+      "loss": 0.6878,
+      "step": 1227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423696757647977e-05,
+      "loss": 0.6823,
+      "step": 1228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0410463795001177e-05,
+      "loss": 0.6077,
+      "step": 1229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039723011233337e-05,
+      "loss": 0.5919,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383995732858609e-05,
+      "loss": 0.6481,
+      "step": 1231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0370760679792173e-05,
+      "loss": 0.653,
+      "step": 1232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0357524976350522e-05,
+      "loss": 0.6937,
+      "step": 1233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0344288645751257e-05,
+      "loss": 0.6202,
+      "step": 1234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0331051711213079e-05,
+      "loss": 0.6429,
+      "step": 1235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031781419595575e-05,
+      "loss": 0.5797,
+      "step": 1236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0304576123200045e-05,
+      "loss": 0.6533,
+      "step": 1237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291337516167725e-05,
+      "loss": 0.6404,
+      "step": 1238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278098398081482e-05,
+      "loss": 0.6956,
+      "step": 1239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264858792164908e-05,
+      "loss": 0.6668,
+      "step": 1240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025161872164245e-05,
+      "loss": 0.6413,
+      "step": 1241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0238378209739366e-05,
+      "loss": 0.6431,
+      "step": 1242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225137279681696e-05,
+      "loss": 0.6379,
+      "step": 1243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0211895954696204e-05,
+      "loss": 0.652,
+      "step": 1244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019865425801036e-05,
+      "loss": 0.6102,
+      "step": 1245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0185412212852268e-05,
+      "loss": 0.5827,
+      "step": 1246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017216984245066e-05,
+      "loss": 0.6555,
+      "step": 1247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0158927170034831e-05,
+      "loss": 0.6318,
+      "step": 1248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145684218834604e-05,
+      "loss": 0.7373,
+      "step": 1249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0132441012080296e-05,
+      "loss": 0.6725,
+      "step": 1250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011919757300267e-05,
+      "loss": 0.6071,
+      "step": 1251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105953924832894e-05,
+      "loss": 0.65,
+      "step": 1252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0092710090802506e-05,
+      "loss": 0.6226,
+      "step": 1253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079466094143373e-05,
+      "loss": 0.5319,
+      "step": 1254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0066221958087636e-05,
+      "loss": 0.6565,
+      "step": 1255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052977705867697e-05,
+      "loss": 0.6192,
+      "step": 1256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039733360716145e-05,
+      "loss": 0.6654,
+      "step": 1257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026488945865744e-05,
+      "loss": 0.6046,
+      "step": 1258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013244484549376e-05,
+      "loss": 0.6024,
+      "step": 1259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.6731,
+      "step": 1260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986755515450625e-06,
+      "loss": 0.7332,
+      "step": 1261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973511054134259e-06,
+      "loss": 0.6355,
+      "step": 1262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960266639283857e-06,
+      "loss": 0.6642,
+      "step": 1263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.947022294132306e-06,
+      "loss": 0.6271,
+      "step": 1264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.933778041912365e-06,
+      "loss": 0.6326,
+      "step": 1265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920533905856634e-06,
+      "loss": 0.6505,
+      "step": 1266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.907289909197496e-06,
+      "loss": 0.7731,
+      "step": 1267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894046075167106e-06,
+      "loss": 0.6125,
+      "step": 1268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880802426997334e-06,
+      "loss": 0.7185,
+      "step": 1269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.867558987919704e-06,
+      "loss": 0.7106,
+      "step": 1270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854315781165398e-06,
+      "loss": 0.7425,
+      "step": 1271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.841072829965172e-06,
+      "loss": 0.6497,
+      "step": 1272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827830157549344e-06,
+      "loss": 0.5556,
+      "step": 1273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.814587787147735e-06,
+      "loss": 0.6899,
+      "step": 1274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.801345741989646e-06,
+      "loss": 0.5726,
+      "step": 1275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788104045303797e-06,
+      "loss": 0.6331,
+      "step": 1276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774862720318305e-06,
+      "loss": 0.7118,
+      "step": 1277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761621790260636e-06,
+      "loss": 0.6878,
+      "step": 1278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748381278357554e-06,
+      "loss": 0.6541,
+      "step": 1279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735141207835095e-06,
+      "loss": 0.6494,
+      "step": 1280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72190160191852e-06,
+      "loss": 0.7121,
+      "step": 1281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708662483832279e-06,
+      "loss": 0.6267,
+      "step": 1282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695423876799957e-06,
+      "loss": 0.5758,
+      "step": 1283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682185804044252e-06,
+      "loss": 0.6203,
+      "step": 1284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668948288786923e-06,
+      "loss": 0.6874,
+      "step": 1285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.655711354248747e-06,
+      "loss": 0.6808,
+      "step": 1286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.642475023649483e-06,
+      "loss": 0.6903,
+      "step": 1287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62923932020783e-06,
+      "loss": 0.6611,
+      "step": 1288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616004267141396e-06,
+      "loss": 0.6374,
+      "step": 1289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602769887666633e-06,
+      "loss": 0.6496,
+      "step": 1290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.589536204998823e-06,
+      "loss": 0.7096,
+      "step": 1291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576303242352025e-06,
+      "loss": 0.6436,
+      "step": 1292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563071022939028e-06,
+      "loss": 0.6056,
+      "step": 1293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.549839569971323e-06,
+      "loss": 0.5985,
+      "step": 1294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536608906659052e-06,
+      "loss": 0.6914,
+      "step": 1295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523379056210982e-06,
+      "loss": 0.6994,
+      "step": 1296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.510150041834436e-06,
+      "loss": 0.6008,
+      "step": 1297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496921886735287e-06,
+      "loss": 0.6901,
+      "step": 1298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.483694614117898e-06,
+      "loss": 0.6402,
+      "step": 1299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470468247185076e-06,
+      "loss": 0.6977,
+      "step": 1300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457242809138045e-06,
+      "loss": 0.6134,
+      "step": 1301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444018323176399e-06,
+      "loss": 0.7395,
+      "step": 1302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430794812498062e-06,
+      "loss": 0.6213,
+      "step": 1303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417572300299244e-06,
+      "loss": 0.6868,
+      "step": 1304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404350809774412e-06,
+      "loss": 0.6937,
+      "step": 1305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391130364116226e-06,
+      "loss": 0.6271,
+      "step": 1306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377910986515529e-06,
+      "loss": 0.7128,
+      "step": 1307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.364692700161287e-06,
+      "loss": 0.6465,
+      "step": 1308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.35147552824054e-06,
+      "loss": 0.635,
+      "step": 1309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33825949393839e-06,
+      "loss": 0.699,
+      "step": 1310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325044620437924e-06,
+      "loss": 0.7016,
+      "step": 1311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311830930920214e-06,
+      "loss": 0.6355,
+      "step": 1312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298618448564236e-06,
+      "loss": 0.6209,
+      "step": 1313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285407196546862e-06,
+      "loss": 0.6426,
+      "step": 1314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272197198042804e-06,
+      "loss": 0.6308,
+      "step": 1315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25898847622457e-06,
+      "loss": 0.723,
+      "step": 1316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24578105426243e-06,
+      "loss": 0.5931,
+      "step": 1317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232574955324369e-06,
+      "loss": 0.7124,
+      "step": 1318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219370202576067e-06,
+      "loss": 0.7423,
+      "step": 1319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206166819180822e-06,
+      "loss": 0.6254,
+      "step": 1320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192964828299545e-06,
+      "loss": 0.5903,
+      "step": 1321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179764253090703e-06,
+      "loss": 0.6314,
+      "step": 1322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166565116710273e-06,
+      "loss": 0.6325,
+      "step": 1323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153367442311712e-06,
+      "loss": 0.6226,
+      "step": 1324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140171253045906e-06,
+      "loss": 0.6444,
+      "step": 1325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12697657206115e-06,
+      "loss": 0.5342,
+      "step": 1326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113783422503076e-06,
+      "loss": 0.6056,
+      "step": 1327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.100591827514643e-06,
+      "loss": 0.6532,
+      "step": 1328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087401810236084e-06,
+      "loss": 0.6605,
+      "step": 1329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07421339380485e-06,
+      "loss": 0.5691,
+      "step": 1330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061026601355596e-06,
+      "loss": 0.6582,
+      "step": 1331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.047841456020125e-06,
+      "loss": 0.6211,
+      "step": 1332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034657980927351e-06,
+      "loss": 0.5902,
+      "step": 1333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021476199203255e-06,
+      "loss": 0.757,
+      "step": 1334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.008296133970855e-06,
+      "loss": 0.7054,
+      "step": 1335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995117808350146e-06,
+      "loss": 0.6219,
+      "step": 1336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981941245458086e-06,
+      "loss": 0.6548,
+      "step": 1337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.968766468408532e-06,
+      "loss": 0.6361,
+      "step": 1338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.955593500312208e-06,
+      "loss": 0.6198,
+      "step": 1339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.942422364276668e-06,
+      "loss": 0.682,
+      "step": 1340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929253083406248e-06,
+      "loss": 0.6167,
+      "step": 1341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916085680802038e-06,
+      "loss": 0.6933,
+      "step": 1342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902920179561822e-06,
+      "loss": 0.6746,
+      "step": 1343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889756602780059e-06,
+      "loss": 0.6791,
+      "step": 1344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876594973547825e-06,
+      "loss": 0.6868,
+      "step": 1345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.863435314952787e-06,
+      "loss": 0.592,
+      "step": 1346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85027765007915e-06,
+      "loss": 0.65,
+      "step": 1347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837122002007614e-06,
+      "loss": 0.6359,
+      "step": 1348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82396839381536e-06,
+      "loss": 0.6293,
+      "step": 1349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810816848575971e-06,
+      "loss": 0.6673,
+      "step": 1350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797667389359426e-06,
+      "loss": 0.6763,
+      "step": 1351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784520039232044e-06,
+      "loss": 0.5985,
+      "step": 1352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771374821256431e-06,
+      "loss": 0.629,
+      "step": 1353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.758231758491467e-06,
+      "loss": 0.6728,
+      "step": 1354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745090873992239e-06,
+      "loss": 0.7148,
+      "step": 1355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731952190810029e-06,
+      "loss": 0.6358,
+      "step": 1356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718815731992239e-06,
+      "loss": 0.7123,
+      "step": 1357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705681520582382e-06,
+      "loss": 0.7215,
+      "step": 1358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69254957962003e-06,
+      "loss": 0.5794,
+      "step": 1359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679419932140765e-06,
+      "loss": 0.6664,
+      "step": 1360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666292601176146e-06,
+      "loss": 0.6564,
+      "step": 1361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.653167609753667e-06,
+      "loss": 0.7458,
+      "step": 1362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640044980896734e-06,
+      "loss": 0.6763,
+      "step": 1363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62692473762458e-06,
+      "loss": 0.6906,
+      "step": 1364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613806902952286e-06,
+      "loss": 0.6191,
+      "step": 1365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600691499890677e-06,
+      "loss": 0.6593,
+      "step": 1366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587578551446338e-06,
+      "loss": 0.6647,
+      "step": 1367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574468080621533e-06,
+      "loss": 0.6891,
+      "step": 1368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.561360110414185e-06,
+      "loss": 0.7016,
+      "step": 1369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54825466381783e-06,
+      "loss": 0.623,
+      "step": 1370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.535151763821574e-06,
+      "loss": 0.5997,
+      "step": 1371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522051433410064e-06,
+      "loss": 0.6429,
+      "step": 1372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508953695563428e-06,
+      "loss": 0.6178,
+      "step": 1373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.495858573257258e-06,
+      "loss": 0.5654,
+      "step": 1374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48276608946255e-06,
+      "loss": 0.6463,
+      "step": 1375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.469676267145674e-06,
+      "loss": 0.6155,
+      "step": 1376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456589129268335e-06,
+      "loss": 0.6585,
+      "step": 1377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443504698787517e-06,
+      "loss": 0.6637,
+      "step": 1378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430422998655473e-06,
+      "loss": 0.6316,
+      "step": 1379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.417344051819646e-06,
+      "loss": 0.5966,
+      "step": 1380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40426788122267e-06,
+      "loss": 0.6883,
+      "step": 1381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.391194509802294e-06,
+      "loss": 0.5809,
+      "step": 1382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.378123960491362e-06,
+      "loss": 0.6205,
+      "step": 1383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.365056256217772e-06,
+      "loss": 0.6574,
+      "step": 1384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351991419904416e-06,
+      "loss": 0.6135,
+      "step": 1385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338929474469177e-06,
+      "loss": 0.6732,
+      "step": 1386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325870442824848e-06,
+      "loss": 0.571,
+      "step": 1387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312814347879121e-06,
+      "loss": 0.6187,
+      "step": 1388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299761212534539e-06,
+      "loss": 0.7331,
+      "step": 1389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286711059688441e-06,
+      "loss": 0.5907,
+      "step": 1390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.273663912232948e-06,
+      "loss": 0.6352,
+      "step": 1391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260619793054894e-06,
+      "loss": 0.6186,
+      "step": 1392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247578725035819e-06,
+      "loss": 0.6895,
+      "step": 1393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234540731051892e-06,
+      "loss": 0.5519,
+      "step": 1394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221505833973908e-06,
+      "loss": 0.6142,
+      "step": 1395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208474056667212e-06,
+      "loss": 0.591,
+      "step": 1396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195445421991692e-06,
+      "loss": 0.634,
+      "step": 1397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.182419952801716e-06,
+      "loss": 0.7016,
+      "step": 1398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169397671946092e-06,
+      "loss": 0.6793,
+      "step": 1399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15637860226805e-06,
+      "loss": 0.6162,
+      "step": 1400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143362766605173e-06,
+      "loss": 0.5626,
+      "step": 1401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.130350187789387e-06,
+      "loss": 0.6736,
+      "step": 1402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117340888646885e-06,
+      "loss": 0.6167,
+      "step": 1403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104334891998124e-06,
+      "loss": 0.6183,
+      "step": 1404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091332220657759e-06,
+      "loss": 0.6364,
+      "step": 1405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078332897434617e-06,
+      "loss": 0.7039,
+      "step": 1406
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.065336945131647e-06,
+      "loss": 0.6088,
+      "step": 1407
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.052344386545882e-06,
+      "loss": 0.5764,
+      "step": 1408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039355244468418e-06,
+      "loss": 0.6885,
+      "step": 1409
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026369541684334e-06,
+      "loss": 0.6421,
+      "step": 1410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.013387300972698e-06,
+      "loss": 0.7,
+      "step": 1411
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000408545106492e-06,
+      "loss": 0.6638,
+      "step": 1412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987433296852589e-06,
+      "loss": 0.6438,
+      "step": 1413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97446157897171e-06,
+      "loss": 0.6526,
+      "step": 1414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.961493414218377e-06,
+      "loss": 0.6316,
+      "step": 1415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948528825340891e-06,
+      "loss": 0.6365,
+      "step": 1416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935567835081265e-06,
+      "loss": 0.5808,
+      "step": 1417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922610466175213e-06,
+      "loss": 0.6729,
+      "step": 1418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.909656741352092e-06,
+      "loss": 0.6027,
+      "step": 1419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.896706683334863e-06,
+      "loss": 0.5762,
+      "step": 1420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.883760314840064e-06,
+      "loss": 0.6939,
+      "step": 1421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870817658577743e-06,
+      "loss": 0.6654,
+      "step": 1422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85787873725146e-06,
+      "loss": 0.6737,
+      "step": 1423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844943573558202e-06,
+      "loss": 0.5907,
+      "step": 1424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83201219018838e-06,
+      "loss": 0.6303,
+      "step": 1425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.819084609825762e-06,
+      "loss": 0.6821,
+      "step": 1426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806160855147456e-06,
+      "loss": 0.6315,
+      "step": 1427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793240948823852e-06,
+      "loss": 0.6252,
+      "step": 1428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780324913518586e-06,
+      "loss": 0.6126,
+      "step": 1429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767412771888515e-06,
+      "loss": 0.5741,
+      "step": 1430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.754504546583652e-06,
+      "loss": 0.6822,
+      "step": 1431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.741600260247155e-06,
+      "loss": 0.6312,
+      "step": 1432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728699935515257e-06,
+      "loss": 0.6389,
+      "step": 1433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715803595017257e-06,
+      "loss": 0.6923,
+      "step": 1434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702911261375454e-06,
+      "loss": 0.622,
+      "step": 1435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69002295720512e-06,
+      "loss": 0.6477,
+      "step": 1436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677138705114463e-06,
+      "loss": 0.6774,
+      "step": 1437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664258527704576e-06,
+      "loss": 0.655,
+      "step": 1438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651382447569418e-06,
+      "loss": 0.6349,
+      "step": 1439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.638510487295738e-06,
+      "loss": 0.5454,
+      "step": 1440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625642669463084e-06,
+      "loss": 0.6604,
+      "step": 1441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61277901664372e-06,
+      "loss": 0.5886,
+      "step": 1442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.599919551402606e-06,
+      "loss": 0.6668,
+      "step": 1443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587064296297364e-06,
+      "loss": 0.6225,
+      "step": 1444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574213273878221e-06,
+      "loss": 0.6207,
+      "step": 1445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56136650668799e-06,
+      "loss": 0.6111,
+      "step": 1446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548524017262007e-06,
+      "loss": 0.7042,
+      "step": 1447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.535685828128117e-06,
+      "loss": 0.685,
+      "step": 1448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52285196180661e-06,
+      "loss": 0.697,
+      "step": 1449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510022440810203e-06,
+      "loss": 0.6194,
+      "step": 1450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497197287643986e-06,
+      "loss": 0.5635,
+      "step": 1451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48437652480538e-06,
+      "loss": 0.6443,
+      "step": 1452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471560174784122e-06,
+      "loss": 0.6482,
+      "step": 1453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458748260062187e-06,
+      "loss": 0.6442,
+      "step": 1454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.445940803113792e-06,
+      "loss": 0.6554,
+      "step": 1455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.433137826405314e-06,
+      "loss": 0.6285,
+      "step": 1456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.420339352395285e-06,
+      "loss": 0.5656,
+      "step": 1457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407545403534334e-06,
+      "loss": 0.7026,
+      "step": 1458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394756002265147e-06,
+      "loss": 0.6109,
+      "step": 1459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.381971171022442e-06,
+      "loss": 0.631,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.369190932232911e-06,
+      "loss": 0.6284,
+      "step": 1461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356415308315201e-06,
+      "loss": 0.6287,
+      "step": 1462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.343644321679851e-06,
+      "loss": 0.6145,
+      "step": 1463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3308779947292776e-06,
+      "loss": 0.5997,
+      "step": 1464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318116349857719e-06,
+      "loss": 0.7314,
+      "step": 1465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305359409451192e-06,
+      "loss": 0.663,
+      "step": 1466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2926071958874765e-06,
+      "loss": 0.6189,
+      "step": 1467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.279859731536045e-06,
+      "loss": 0.7363,
+      "step": 1468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2671170387580534e-06,
+      "loss": 0.6659,
+      "step": 1469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2543791399062755e-06,
+      "loss": 0.552,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241646057325084e-06,
+      "loss": 0.6937,
+      "step": 1471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228917813350404e-06,
+      "loss": 0.6792,
+      "step": 1472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216194430309657e-06,
+      "loss": 0.6138,
+      "step": 1473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203475930521764e-06,
+      "loss": 0.662,
+      "step": 1474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.190762336297052e-06,
+      "loss": 0.6375,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1780536699372685e-06,
+      "loss": 0.6624,
+      "step": 1476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165349953735494e-06,
+      "loss": 0.6535,
+      "step": 1477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1526512099761424e-06,
+      "loss": 0.6433,
+      "step": 1478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139957460934902e-06,
+      "loss": 0.6151,
+      "step": 1479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.127268728878687e-06,
+      "loss": 0.6065,
+      "step": 1480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.11458503606563e-06,
+      "loss": 0.6915,
+      "step": 1481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.101906404745006e-06,
+      "loss": 0.6012,
+      "step": 1482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089232857157228e-06,
+      "loss": 0.6355,
+      "step": 1483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076564415533774e-06,
+      "loss": 0.5928,
+      "step": 1484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.063901102097184e-06,
+      "loss": 0.6511,
+      "step": 1485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0512429390609825e-06,
+      "loss": 0.6225,
+      "step": 1486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038589948629677e-06,
+      "loss": 0.6154,
+      "step": 1487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0259421529986946e-06,
+      "loss": 0.5748,
+      "step": 1488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013299574354342e-06,
+      "loss": 0.6038,
+      "step": 1489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00066223487379e-06,
+      "loss": 0.6378,
+      "step": 1490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9880301567250005e-06,
+      "loss": 0.6833,
+      "step": 1491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975403362066727e-06,
+      "loss": 0.6165,
+      "step": 1492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962781873048435e-06,
+      "loss": 0.6691,
+      "step": 1493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9501657118102994e-06,
+      "loss": 0.5977,
+      "step": 1494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93755490048314e-06,
+      "loss": 0.6229,
+      "step": 1495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92494946118839e-06,
+      "loss": 0.6642,
+      "step": 1496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91234941603807e-06,
+      "loss": 0.6479,
+      "step": 1497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899754787134725e-06,
+      "loss": 0.6376,
+      "step": 1498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.887165596571411e-06,
+      "loss": 0.6789,
+      "step": 1499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874581866431633e-06,
+      "loss": 0.7342,
+      "step": 1500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86200361878933e-06,
+      "loss": 0.6027,
+      "step": 1501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849430875708818e-06,
+      "loss": 0.6349,
+      "step": 1502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.836863659244746e-06,
+      "loss": 0.6059,
+      "step": 1503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82430199144209e-06,
+      "loss": 0.6761,
+      "step": 1504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811745894336074e-06,
+      "loss": 0.646,
+      "step": 1505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799195389952163e-06,
+      "loss": 0.6818,
+      "step": 1506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786650500306e-06,
+      "loss": 0.6167,
+      "step": 1507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77411124740339e-06,
+      "loss": 0.7354,
+      "step": 1508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7615776532402456e-06,
+      "loss": 0.5956,
+      "step": 1509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7490497398025444e-06,
+      "loss": 0.6544,
+      "step": 1510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.736527529066319e-06,
+      "loss": 0.6957,
+      "step": 1511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.724011042997576e-06,
+      "loss": 0.6441,
+      "step": 1512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7115003035522985e-06,
+      "loss": 0.6593,
+      "step": 1513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.698995332676375e-06,
+      "loss": 0.642,
+      "step": 1514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686496152305586e-06,
+      "loss": 0.6411,
+      "step": 1515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.674002784365547e-06,
+      "loss": 0.6947,
+      "step": 1516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66151525077168e-06,
+      "loss": 0.6317,
+      "step": 1517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.649033573429178e-06,
+      "loss": 0.5698,
+      "step": 1518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6365577742329455e-06,
+      "loss": 0.622,
+      "step": 1519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6240878750676e-06,
+      "loss": 0.6135,
+      "step": 1520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611623897807382e-06,
+      "loss": 0.7081,
+      "step": 1521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5991658643161696e-06,
+      "loss": 0.6611,
+      "step": 1522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586713796447392e-06,
+      "loss": 0.5996,
+      "step": 1523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574267716044033e-06,
+      "loss": 0.6354,
+      "step": 1524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561827644938563e-06,
+      "loss": 0.5704,
+      "step": 1525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549393604952906e-06,
+      "loss": 0.6245,
+      "step": 1526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.536965617898423e-06,
+      "loss": 0.69,
+      "step": 1527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.524543705575839e-06,
+      "loss": 0.649,
+      "step": 1528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.512127889775239e-06,
+      "loss": 0.6557,
+      "step": 1529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499718192275999e-06,
+      "loss": 0.6709,
+      "step": 1530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487314634846774e-06,
+      "loss": 0.6524,
+      "step": 1531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474917239245445e-06,
+      "loss": 0.5817,
+      "step": 1532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4625260272190775e-06,
+      "loss": 0.7199,
+      "step": 1533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.450141020503902e-06,
+      "loss": 0.6143,
+      "step": 1534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43776224082525e-06,
+      "loss": 0.5836,
+      "step": 1535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.425389709897543e-06,
+      "loss": 0.5933,
+      "step": 1536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4130234494242315e-06,
+      "loss": 0.7129,
+      "step": 1537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400663481097774e-06,
+      "loss": 0.6216,
+      "step": 1538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388309826599588e-06,
+      "loss": 0.7252,
+      "step": 1539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375962507600009e-06,
+      "loss": 0.6145,
+      "step": 1540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.363621545758276e-06,
+      "loss": 0.6862,
+      "step": 1541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3512869627224535e-06,
+      "loss": 0.6897,
+      "step": 1542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338958780129441e-06,
+      "loss": 0.6848,
+      "step": 1543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326637019604888e-06,
+      "loss": 0.659,
+      "step": 1544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314321702763198e-06,
+      "loss": 0.5883,
+      "step": 1545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302012851207455e-06,
+      "loss": 0.6645,
+      "step": 1546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.289710486529412e-06,
+      "loss": 0.6635,
+      "step": 1547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277414630309444e-06,
+      "loss": 0.5986,
+      "step": 1548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265125304116498e-06,
+      "loss": 0.7029,
+      "step": 1549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252842529508081e-06,
+      "loss": 0.6566,
+      "step": 1550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240566328030193e-06,
+      "loss": 0.6431,
+      "step": 1551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228296721217317e-06,
+      "loss": 0.6174,
+      "step": 1552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216033730592357e-06,
+      "loss": 0.6436,
+      "step": 1553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2037773776666134e-06,
+      "loss": 0.6042,
+      "step": 1554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191527683939753e-06,
+      "loss": 0.648,
+      "step": 1555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179284670899745e-06,
+      "loss": 0.6231,
+      "step": 1556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167048360022856e-06,
+      "loss": 0.6698,
+      "step": 1557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154818772773579e-06,
+      "loss": 0.6573,
+      "step": 1558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142595930604631e-06,
+      "loss": 0.7049,
+      "step": 1559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.130379854956879e-06,
+      "loss": 0.6447,
+      "step": 1560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118170567259336e-06,
+      "loss": 0.7026,
+      "step": 1561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105968088929098e-06,
+      "loss": 0.6007,
+      "step": 1562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.093772441371315e-06,
+      "loss": 0.6078,
+      "step": 1563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.081583645979168e-06,
+      "loss": 0.6222,
+      "step": 1564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069401724133796e-06,
+      "loss": 0.6396,
+      "step": 1565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.057226697204308e-06,
+      "loss": 0.6941,
+      "step": 1566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045058586547692e-06,
+      "loss": 0.6709,
+      "step": 1567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032897413508822e-06,
+      "loss": 0.6249,
+      "step": 1568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0207431994203955e-06,
+      "loss": 0.6003,
+      "step": 1569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0085959656028994e-06,
+      "loss": 0.661,
+      "step": 1570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9964557333645845e-06,
+      "loss": 0.6845,
+      "step": 1571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.984322524001409e-06,
+      "loss": 0.6322,
+      "step": 1572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972196358797024e-06,
+      "loss": 0.6658,
+      "step": 1573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960077259022713e-06,
+      "loss": 0.632,
+      "step": 1574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947965245937375e-06,
+      "loss": 0.6772,
+      "step": 1575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9358603407874695e-06,
+      "loss": 0.6451,
+      "step": 1576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9237625648069895e-06,
+      "loss": 0.6318,
+      "step": 1577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9116719392174304e-06,
+      "loss": 0.6824,
+      "step": 1578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89958848522773e-06,
+      "loss": 0.6331,
+      "step": 1579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.887512224034263e-06,
+      "loss": 0.6536,
+      "step": 1580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8754431768207694e-06,
+      "loss": 0.6422,
+      "step": 1581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8633813647583505e-06,
+      "loss": 0.6291,
+      "step": 1582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851326809005402e-06,
+      "loss": 0.6412,
+      "step": 1583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8392795307076e-06,
+      "loss": 0.6001,
+      "step": 1584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827239550997856e-06,
+      "loss": 0.6229,
+      "step": 1585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815206890996267e-06,
+      "loss": 0.6071,
+      "step": 1586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803181571810106e-06,
+      "loss": 0.691,
+      "step": 1587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791163614533753e-06,
+      "loss": 0.5727,
+      "step": 1588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7791530402486884e-06,
+      "loss": 0.6147,
+      "step": 1589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76714987002343e-06,
+      "loss": 0.6413,
+      "step": 1590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755154124913514e-06,
+      "loss": 0.7356,
+      "step": 1591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.743165825961454e-06,
+      "loss": 0.6126,
+      "step": 1592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731184994196697e-06,
+      "loss": 0.7526,
+      "step": 1593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719211650635586e-06,
+      "loss": 0.6431,
+      "step": 1594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.707245816281345e-06,
+      "loss": 0.5429,
+      "step": 1595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695287512124011e-06,
+      "loss": 0.6359,
+      "step": 1596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683336759140409e-06,
+      "loss": 0.5654,
+      "step": 1597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671393578294133e-06,
+      "loss": 0.6572,
+      "step": 1598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659457990535491e-06,
+      "loss": 0.6754,
+      "step": 1599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647530016801457e-06,
+      "loss": 0.6868,
+      "step": 1600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.635609678015668e-06,
+      "loss": 0.647,
+      "step": 1601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62369699508835e-06,
+      "loss": 0.5354,
+      "step": 1602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611791988916317e-06,
+      "loss": 0.6123,
+      "step": 1603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5998946803829e-06,
+      "loss": 0.578,
+      "step": 1604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588005090357943e-06,
+      "loss": 0.652,
+      "step": 1605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576123239697735e-06,
+      "loss": 0.6212,
+      "step": 1606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564249149244998e-06,
+      "loss": 0.7071,
+      "step": 1607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552382839828847e-06,
+      "loss": 0.6945,
+      "step": 1608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.540524332264734e-06,
+      "loss": 0.6102,
+      "step": 1609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528673647354432e-06,
+      "loss": 0.6677,
+      "step": 1610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.516830805885989e-06,
+      "loss": 0.5845,
+      "step": 1611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504995828633704e-06,
+      "loss": 0.6303,
+      "step": 1612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493168736358063e-06,
+      "loss": 0.6699,
+      "step": 1613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481349549805741e-06,
+      "loss": 0.6219,
+      "step": 1614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469538289709535e-06,
+      "loss": 0.5898,
+      "step": 1615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457734976788331e-06,
+      "loss": 0.6263,
+      "step": 1616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445939631747089e-06,
+      "loss": 0.6121,
+      "step": 1617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434152275276776e-06,
+      "loss": 0.6207,
+      "step": 1618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4223729280543634e-06,
+      "loss": 0.607,
+      "step": 1619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410601610742754e-06,
+      "loss": 0.6156,
+      "step": 1620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39883834399078e-06,
+      "loss": 0.696,
+      "step": 1621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38708314843315e-06,
+      "loss": 0.6141,
+      "step": 1622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375336044690405e-06,
+      "loss": 0.6294,
+      "step": 1623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363597053368897e-06,
+      "loss": 0.6457,
+      "step": 1624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3518661950607465e-06,
+      "loss": 0.5826,
+      "step": 1625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340143490343813e-06,
+      "loss": 0.5124,
+      "step": 1626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328428959781643e-06,
+      "loss": 0.6587,
+      "step": 1627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316722623923454e-06,
+      "loss": 0.5949,
+      "step": 1628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305024503304086e-06,
+      "loss": 0.6249,
+      "step": 1629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293334618443962e-06,
+      "loss": 0.5911,
+      "step": 1630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281652989849067e-06,
+      "loss": 0.6717,
+      "step": 1631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269979638010893e-06,
+      "loss": 0.6239,
+      "step": 1632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2583145834064295e-06,
+      "loss": 0.669,
+      "step": 1633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.24665784649809e-06,
+      "loss": 0.6046,
+      "step": 1634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235009447733717e-06,
+      "loss": 0.637,
+      "step": 1635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.223369407546509e-06,
+      "loss": 0.6707,
+      "step": 1636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211737746355021e-06,
+      "loss": 0.6737,
+      "step": 1637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2001144845630906e-06,
+      "loss": 0.6084,
+      "step": 1638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188499642559838e-06,
+      "loss": 0.6288,
+      "step": 1639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176893240719602e-06,
+      "loss": 0.6209,
+      "step": 1640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165295299401921e-06,
+      "loss": 0.7151,
+      "step": 1641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.153705838951495e-06,
+      "loss": 0.6241,
+      "step": 1642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1421248796981385e-06,
+      "loss": 0.6221,
+      "step": 1643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1305524419567595e-06,
+      "loss": 0.6406,
+      "step": 1644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1189885460273255e-06,
+      "loss": 0.6736,
+      "step": 1645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107433212194801e-06,
+      "loss": 0.6042,
+      "step": 1646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.095886460729152e-06,
+      "loss": 0.6682,
+      "step": 1647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08434831188527e-06,
+      "loss": 0.7017,
+      "step": 1648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072818785902975e-06,
+      "loss": 0.6445,
+      "step": 1649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061297903006943e-06,
+      "loss": 0.5683,
+      "step": 1650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049785683406704e-06,
+      "loss": 0.5783,
+      "step": 1651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038282147296585e-06,
+      "loss": 0.613,
+      "step": 1652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026787314855679e-06,
+      "loss": 0.539,
+      "step": 1653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015301206247813e-06,
+      "loss": 0.6494,
+      "step": 1654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003823841621504e-06,
+      "loss": 0.6016,
+      "step": 1655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.992355241109949e-06,
+      "loss": 0.604,
+      "step": 1656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980895424830948e-06,
+      "loss": 0.5705,
+      "step": 1657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96944441288691e-06,
+      "loss": 0.6403,
+      "step": 1658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958002225364797e-06,
+      "loss": 0.578,
+      "step": 1659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94656888233608e-06,
+      "loss": 0.6708,
+      "step": 1660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935144403856731e-06,
+      "loss": 0.6255,
+      "step": 1661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923728809967156e-06,
+      "loss": 0.629,
+      "step": 1662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912322120692194e-06,
+      "loss": 0.6734,
+      "step": 1663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900924356041044e-06,
+      "loss": 0.6246,
+      "step": 1664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889535536007267e-06,
+      "loss": 0.5585,
+      "step": 1665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878155680568721e-06,
+      "loss": 0.5961,
+      "step": 1666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866784809687553e-06,
+      "loss": 0.5819,
+      "step": 1667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855422943310129e-06,
+      "loss": 0.5051,
+      "step": 1668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844070101367043e-06,
+      "loss": 0.6588,
+      "step": 1669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832726303773042e-06,
+      "loss": 0.6368,
+      "step": 1670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821391570427008e-06,
+      "loss": 0.5931,
+      "step": 1671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810065921211936e-06,
+      "loss": 0.7271,
+      "step": 1672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79874937599487e-06,
+      "loss": 0.6914,
+      "step": 1673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787441954626895e-06,
+      "loss": 0.5705,
+      "step": 1674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776143676943093e-06,
+      "loss": 0.6522,
+      "step": 1675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764854562762491e-06,
+      "loss": 0.647,
+      "step": 1676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.753574631888063e-06,
+      "loss": 0.6377,
+      "step": 1677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742303904106653e-06,
+      "loss": 0.6902,
+      "step": 1678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731042399188981e-06,
+      "loss": 0.7178,
+      "step": 1679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719790136889569e-06,
+      "loss": 0.6393,
+      "step": 1680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.708547136946742e-06,
+      "loss": 0.5724,
+      "step": 1681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697313419082573e-06,
+      "loss": 0.6538,
+      "step": 1682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6860890030028485e-06,
+      "loss": 0.7321,
+      "step": 1683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674873908397039e-06,
+      "loss": 0.6671,
+      "step": 1684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663668154938262e-06,
+      "loss": 0.6276,
+      "step": 1685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65247176228326e-06,
+      "loss": 0.6324,
+      "step": 1686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64128475007234e-06,
+      "loss": 0.6553,
+      "step": 1687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630107137929365e-06,
+      "loss": 0.5936,
+      "step": 1688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.618938945461708e-06,
+      "loss": 0.6397,
+      "step": 1689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6077801922602105e-06,
+      "loss": 0.5604,
+      "step": 1690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596630897899164e-06,
+      "loss": 0.7081,
+      "step": 1691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585491081936263e-06,
+      "loss": 0.6613,
+      "step": 1692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57436076391258e-06,
+      "loss": 0.687,
+      "step": 1693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563239963352517e-06,
+      "loss": 0.5905,
+      "step": 1694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.552128699763795e-06,
+      "loss": 0.6323,
+      "step": 1695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5410269926373905e-06,
+      "loss": 0.6664,
+      "step": 1696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529934861447532e-06,
+      "loss": 0.6958,
+      "step": 1697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518852325651638e-06,
+      "loss": 0.6338,
+      "step": 1698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507779404690294e-06,
+      "loss": 0.6107,
+      "step": 1699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.496716117987234e-06,
+      "loss": 0.633,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485662484949275e-06,
+      "loss": 0.6524,
+      "step": 1701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474618524966313e-06,
+      "loss": 0.6004,
+      "step": 1702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463584257411264e-06,
+      "loss": 0.6358,
+      "step": 1703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452559701640053e-06,
+      "loss": 0.652,
+      "step": 1704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441544876991566e-06,
+      "loss": 0.533,
+      "step": 1705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43053980278761e-06,
+      "loss": 0.6784,
+      "step": 1706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4195444983329035e-06,
+      "loss": 0.6295,
+      "step": 1707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4085589829150125e-06,
+      "loss": 0.5547,
+      "step": 1708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397583275804344e-06,
+      "loss": 0.5615,
+      "step": 1709
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386617396254085e-06,
+      "loss": 0.6918,
+      "step": 1710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375661363500201e-06,
+      "loss": 0.5957,
+      "step": 1711
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364715196761368e-06,
+      "loss": 0.6164,
+      "step": 1712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353778915238969e-06,
+      "loss": 0.6334,
+      "step": 1713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342852538117039e-06,
+      "loss": 0.6347,
+      "step": 1714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331936084562235e-06,
+      "loss": 0.6551,
+      "step": 1715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32102957372382e-06,
+      "loss": 0.6247,
+      "step": 1716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310133024733602e-06,
+      "loss": 0.6626,
+      "step": 1717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299246456705921e-06,
+      "loss": 0.6122,
+      "step": 1718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288369888737614e-06,
+      "loss": 0.6568,
+      "step": 1719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277503339907961e-06,
+      "loss": 0.7027,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266646829278685e-06,
+      "loss": 0.5958,
+      "step": 1721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255800375893885e-06,
+      "loss": 0.6179,
+      "step": 1722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244963998780028e-06,
+      "loss": 0.6805,
+      "step": 1723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.234137716945897e-06,
+      "loss": 0.6868,
+      "step": 1724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.223321549382578e-06,
+      "loss": 0.5573,
+      "step": 1725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212515515063399e-06,
+      "loss": 0.597,
+      "step": 1726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201719632943931e-06,
+      "loss": 0.5847,
+      "step": 1727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1909339219619225e-06,
+      "loss": 0.647,
+      "step": 1728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.180158401037282e-06,
+      "loss": 0.6489,
+      "step": 1729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16939308907205e-06,
+      "loss": 0.651,
+      "step": 1730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15863800495035e-06,
+      "loss": 0.6043,
+      "step": 1731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147893167538375e-06,
+      "loss": 0.5854,
+      "step": 1732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137158595684329e-06,
+      "loss": 0.6114,
+      "step": 1733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126434308218421e-06,
+      "loss": 0.6202,
+      "step": 1734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.115720323952818e-06,
+      "loss": 0.6272,
+      "step": 1735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105016661681605e-06,
+      "loss": 0.6163,
+      "step": 1736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0943233401807715e-06,
+      "loss": 0.6743,
+      "step": 1737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083640378208156e-06,
+      "loss": 0.7581,
+      "step": 1738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072967794503437e-06,
+      "loss": 0.6449,
+      "step": 1739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0623056077880775e-06,
+      "loss": 0.6043,
+      "step": 1740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05165383676531e-06,
+      "loss": 0.6249,
+      "step": 1741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.04101250012009e-06,
+      "loss": 0.6138,
+      "step": 1742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030381616519074e-06,
+      "loss": 0.6771,
+      "step": 1743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0197612046105815e-06,
+      "loss": 0.6382,
+      "step": 1744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009151283024557e-06,
+      "loss": 0.658,
+      "step": 1745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998551870372554e-06,
+      "loss": 0.6129,
+      "step": 1746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98796298524768e-06,
+      "loss": 0.6675,
+      "step": 1747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977384646224584e-06,
+      "loss": 0.5419,
+      "step": 1748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9668168718594155e-06,
+      "loss": 0.6792,
+      "step": 1749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956259680689784e-06,
+      "loss": 0.5761,
+      "step": 1750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945713091234743e-06,
+      "loss": 0.5577,
+      "step": 1751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935177121994741e-06,
+      "loss": 0.5979,
+      "step": 1752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924651791451606e-06,
+      "loss": 0.649,
+      "step": 1753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9141371180684925e-06,
+      "loss": 0.6042,
+      "step": 1754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903633120289876e-06,
+      "loss": 0.5785,
+      "step": 1755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893139816541487e-06,
+      "loss": 0.5849,
+      "step": 1756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8826572252303145e-06,
+      "loss": 0.6188,
+      "step": 1757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872185364744543e-06,
+      "loss": 0.5875,
+      "step": 1758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861724253453535e-06,
+      "loss": 0.5596,
+      "step": 1759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851273909707809e-06,
+      "loss": 0.6497,
+      "step": 1760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840834351838977e-06,
+      "loss": 0.5923,
+      "step": 1761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8304055981597495e-06,
+      "loss": 0.668,
+      "step": 1762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819987666963869e-06,
+      "loss": 0.5548,
+      "step": 1763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809580576526104e-06,
+      "loss": 0.5588,
+      "step": 1764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799184345102205e-06,
+      "loss": 0.6508,
+      "step": 1765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7887989909288648e-06,
+      "loss": 0.5973,
+      "step": 1766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7784245322237113e-06,
+      "loss": 0.5946,
+      "step": 1767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7680609871852436e-06,
+      "loss": 0.5996,
+      "step": 1768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7577083739928313e-06,
+      "loss": 0.601,
+      "step": 1769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7473667108066524e-06,
+      "loss": 0.6741,
+      "step": 1770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7370360157676955e-06,
+      "loss": 0.6801,
+      "step": 1771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726716306997692e-06,
+      "loss": 0.6208,
+      "step": 1772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7164076025991068e-06,
+      "loss": 0.6188,
+      "step": 1773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.70610992065511e-06,
+      "loss": 0.573,
+      "step": 1774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.695823279229521e-06,
+      "loss": 0.6968,
+      "step": 1775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.68554769636681e-06,
+      "loss": 0.6047,
+      "step": 1776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6752831900920306e-06,
+      "loss": 0.5545,
+      "step": 1777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.665029778410819e-06,
+      "loss": 0.6083,
+      "step": 1778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6547874793093497e-06,
+      "loss": 0.6485,
+      "step": 1779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6445563107542925e-06,
+      "loss": 0.6835,
+      "step": 1780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.634336290692808e-06,
+      "loss": 0.6244,
+      "step": 1781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.624127437052484e-06,
+      "loss": 0.6111,
+      "step": 1782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6139297677413367e-06,
+      "loss": 0.5409,
+      "step": 1783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6037433006477475e-06,
+      "loss": 0.5767,
+      "step": 1784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5935680536404626e-06,
+      "loss": 0.6752,
+      "step": 1785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5834040445685325e-06,
+      "loss": 0.5394,
+      "step": 1786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5732512912613073e-06,
+      "loss": 0.571,
+      "step": 1787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5631098115283833e-06,
+      "loss": 0.6408,
+      "step": 1788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5529796231595793e-06,
+      "loss": 0.6153,
+      "step": 1789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5428607439249197e-06,
+      "loss": 0.6044,
+      "step": 1790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.532753191574576e-06,
+      "loss": 0.6072,
+      "step": 1791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5226569838388647e-06,
+      "loss": 0.6745,
+      "step": 1792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5125721384281874e-06,
+      "loss": 0.5946,
+      "step": 1793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502498673033026e-06,
+      "loss": 0.5848,
+      "step": 1794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4924366053238977e-06,
+      "loss": 0.6502,
+      "step": 1795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482385952951318e-06,
+      "loss": 0.6443,
+      "step": 1796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472346733545792e-06,
+      "loss": 0.6579,
+      "step": 1797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4623189647177533e-06,
+      "loss": 0.5671,
+      "step": 1798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4523026640575664e-06,
+      "loss": 0.6066,
+      "step": 1799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.442297849135462e-06,
+      "loss": 0.6075,
+      "step": 1800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4323045375015384e-06,
+      "loss": 0.6898,
+      "step": 1801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4223227466857045e-06,
+      "loss": 0.597,
+      "step": 1802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4123524941976593e-06,
+      "loss": 0.6686,
+      "step": 1803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4023937975268728e-06,
+      "loss": 0.6153,
+      "step": 1804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.39244667414253e-06,
+      "loss": 0.6611,
+      "step": 1805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3825111414935287e-06,
+      "loss": 0.5729,
+      "step": 1806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3725872170084193e-06,
+      "loss": 0.6562,
+      "step": 1807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3626749180954033e-06,
+      "loss": 0.6539,
+      "step": 1808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352774262142284e-06,
+      "loss": 0.6655,
+      "step": 1809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.342885266516436e-06,
+      "loss": 0.6419,
+      "step": 1810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3330079485647894e-06,
+      "loss": 0.6072,
+      "step": 1811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3231423256137784e-06,
+      "loss": 0.5955,
+      "step": 1812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3132884149693346e-06,
+      "loss": 0.6535,
+      "step": 1813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3034462339168317e-06,
+      "loss": 0.6097,
+      "step": 1814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2936157997210816e-06,
+      "loss": 0.6541,
+      "step": 1815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283797129626274e-06,
+      "loss": 0.5382,
+      "step": 1816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.27399024085598e-06,
+      "loss": 0.6066,
+      "step": 1817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.264195150613091e-06,
+      "loss": 0.5851,
+      "step": 1818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.254411876079803e-06,
+      "loss": 0.6842,
+      "step": 1819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244640434417595e-06,
+      "loss": 0.5912,
+      "step": 1820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2348808427671784e-06,
+      "loss": 0.6499,
+      "step": 1821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2251331182484868e-06,
+      "loss": 0.6182,
+      "step": 1822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.215397277960626e-06,
+      "loss": 0.6126,
+      "step": 1823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.205673338981865e-06,
+      "loss": 0.6458,
+      "step": 1824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.195961318369595e-06,
+      "loss": 0.6305,
+      "step": 1825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1862612331602906e-06,
+      "loss": 0.5899,
+      "step": 1826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.176573100369504e-06,
+      "loss": 0.6442,
+      "step": 1827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.166896936991808e-06,
+      "loss": 0.7171,
+      "step": 1828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.157232760000789e-06,
+      "loss": 0.6163,
+      "step": 1829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.147580586348998e-06,
+      "loss": 0.585,
+      "step": 1830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137940432967942e-06,
+      "loss": 0.5771,
+      "step": 1831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1283123167680306e-06,
+      "loss": 0.6859,
+      "step": 1832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1186962546385613e-06,
+      "loss": 0.5714,
+      "step": 1833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1090922634476963e-06,
+      "loss": 0.6438,
+      "step": 1834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.099500360042407e-06,
+      "loss": 0.6202,
+      "step": 1835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089920561248476e-06,
+      "loss": 0.6056,
+      "step": 1836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080352883870442e-06,
+      "loss": 0.6483,
+      "step": 1837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0707973446915863e-06,
+      "loss": 0.5747,
+      "step": 1838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0612539604739e-06,
+      "loss": 0.6485,
+      "step": 1839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0517227479580425e-06,
+      "loss": 0.6238,
+      "step": 1840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.042203723863334e-06,
+      "loss": 0.6091,
+      "step": 1841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0326969048877032e-06,
+      "loss": 0.5766,
+      "step": 1842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.023202307707679e-06,
+      "loss": 0.5843,
+      "step": 1843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013719948978342e-06,
+      "loss": 0.6129,
+      "step": 1844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0042498453333137e-06,
+      "loss": 0.6563,
+      "step": 1845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9947920133847108e-06,
+      "loss": 0.6044,
+      "step": 1846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.985346469723124e-06,
+      "loss": 0.6976,
+      "step": 1847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.975913230917595e-06,
+      "loss": 0.6492,
+      "step": 1848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9664923135155723e-06,
+      "loss": 0.6004,
+      "step": 1849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9570837340428994e-06,
+      "loss": 0.6734,
+      "step": 1850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.947687509003766e-06,
+      "loss": 0.6337,
+      "step": 1851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.938303654880702e-06,
+      "loss": 0.5893,
+      "step": 1852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.6212,
+      "step": 1853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9195731252043333e-06,
+      "loss": 0.6412,
+      "step": 1854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9102264825074657e-06,
+      "loss": 0.6447,
+      "step": 1855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.900892276439463e-06,
+      "loss": 0.6326,
+      "step": 1856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8915705233740653e-06,
+      "loss": 0.639,
+      "step": 1857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8822612396631557e-06,
+      "loss": 0.6734,
+      "step": 1858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.872964441636752e-06,
+      "loss": 0.6496,
+      "step": 1859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.863680145602963e-06,
+      "loss": 0.6014,
+      "step": 1860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.854408367847977e-06,
+      "loss": 0.5486,
+      "step": 1861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.845149124636014e-06,
+      "loss": 0.6532,
+      "step": 1862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8359024322093067e-06,
+      "loss": 0.6207,
+      "step": 1863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8266683067880807e-06,
+      "loss": 0.5728,
+      "step": 1864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.817446764570504e-06,
+      "loss": 0.5612,
+      "step": 1865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8082378217326843e-06,
+      "loss": 0.6476,
+      "step": 1866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.799041494428617e-06,
+      "loss": 0.5984,
+      "step": 1867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7898577987901786e-06,
+      "loss": 0.7022,
+      "step": 1868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7806867509270754e-06,
+      "loss": 0.5719,
+      "step": 1869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.771528366926837e-06,
+      "loss": 0.6535,
+      "step": 1870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.762382662854778e-06,
+      "loss": 0.6677,
+      "step": 1871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7532496547539623e-06,
+      "loss": 0.6694,
+      "step": 1872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7441293586451936e-06,
+      "loss": 0.6706,
+      "step": 1873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7350217905269647e-06,
+      "loss": 0.5942,
+      "step": 1874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.725926966375456e-06,
+      "loss": 0.6493,
+      "step": 1875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716844902144481e-06,
+      "loss": 0.6309,
+      "step": 1876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.707775613765471e-06,
+      "loss": 0.6233,
+      "step": 1877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6987191171474548e-06,
+      "loss": 0.6256,
+      "step": 1878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.689675428177013e-06,
+      "loss": 0.6972,
+      "step": 1879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6806445627182686e-06,
+      "loss": 0.6433,
+      "step": 1880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.671626536612838e-06,
+      "loss": 0.5779,
+      "step": 1881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6626213656798295e-06,
+      "loss": 0.6134,
+      "step": 1882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6536290657157883e-06,
+      "loss": 0.6828,
+      "step": 1883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6446496524946894e-06,
+      "loss": 0.674,
+      "step": 1884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.635683141767904e-06,
+      "loss": 0.7182,
+      "step": 1885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.626729549264161e-06,
+      "loss": 0.637,
+      "step": 1886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6177888906895398e-06,
+      "loss": 0.6608,
+      "step": 1887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.608861181727421e-06,
+      "loss": 0.5861,
+      "step": 1888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.599946438038481e-06,
+      "loss": 0.6304,
+      "step": 1889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.591044675260641e-06,
+      "loss": 0.5967,
+      "step": 1890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5821559090090565e-06,
+      "loss": 0.6824,
+      "step": 1891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5732801548760898e-06,
+      "loss": 0.6641,
+      "step": 1892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5644174284312686e-06,
+      "loss": 0.5963,
+      "step": 1893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5555677452212792e-06,
+      "loss": 0.6095,
+      "step": 1894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5467311207699143e-06,
+      "loss": 0.5949,
+      "step": 1895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5379075705780733e-06,
+      "loss": 0.6551,
+      "step": 1896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5290971101237083e-06,
+      "loss": 0.6472,
+      "step": 1897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5202997548618226e-06,
+      "loss": 0.7092,
+      "step": 1898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511515520224418e-06,
+      "loss": 0.6603,
+      "step": 1899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5027444216204888e-06,
+      "loss": 0.6436,
+      "step": 1900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.49398647443599e-06,
+      "loss": 0.6595,
+      "step": 1901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485241694033793e-06,
+      "loss": 0.6608,
+      "step": 1902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.476510095753688e-06,
+      "loss": 0.6017,
+      "step": 1903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.467791694912329e-06,
+      "loss": 0.6715,
+      "step": 1904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.459086506803231e-06,
+      "loss": 0.614,
+      "step": 1905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.450394546696723e-06,
+      "loss": 0.6029,
+      "step": 1906
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441715829839928e-06,
+      "loss": 0.5832,
+      "step": 1907
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.43305037145675e-06,
+      "loss": 0.6692,
+      "step": 1908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.424398186747823e-06,
+      "loss": 0.6473,
+      "step": 1909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.415759290890506e-06,
+      "loss": 0.6604,
+      "step": 1910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4071336990388396e-06,
+      "loss": 0.6651,
+      "step": 1911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3985214263235344e-06,
+      "loss": 0.6874,
+      "step": 1912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.38992248785193e-06,
+      "loss": 0.567,
+      "step": 1913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.38133689870798e-06,
+      "loss": 0.6496,
+      "step": 1914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3727646739522226e-06,
+      "loss": 0.6692,
+      "step": 1915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364205828621745e-06,
+      "loss": 0.5966,
+      "step": 1916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3556603777301745e-06,
+      "loss": 0.625,
+      "step": 1917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3471283362676334e-06,
+      "loss": 0.6688,
+      "step": 1918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3386097192007296e-06,
+      "loss": 0.5881,
+      "step": 1919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3301045414725167e-06,
+      "loss": 0.6437,
+      "step": 1920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321612818002472e-06,
+      "loss": 0.6213,
+      "step": 1921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.313134563686482e-06,
+      "loss": 0.6376,
+      "step": 1922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.304669793396793e-06,
+      "loss": 0.6311,
+      "step": 1923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.29621852198201e-06,
+      "loss": 0.7239,
+      "step": 1924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287780764267047e-06,
+      "loss": 0.577,
+      "step": 1925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2793565350531243e-06,
+      "loss": 0.6204,
+      "step": 1926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.270945849117722e-06,
+      "loss": 0.5755,
+      "step": 1927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262548721214569e-06,
+      "loss": 0.6284,
+      "step": 1928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.254165166073605e-06,
+      "loss": 0.6777,
+      "step": 1929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2457951984009684e-06,
+      "loss": 0.5707,
+      "step": 1930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.237438832878961e-06,
+      "loss": 0.5719,
+      "step": 1931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2290960841660157e-06,
+      "loss": 0.675,
+      "step": 1932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2207669668966934e-06,
+      "loss": 0.6788,
+      "step": 1933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212451495681629e-06,
+      "loss": 0.6523,
+      "step": 1934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2041496851075316e-06,
+      "loss": 0.545,
+      "step": 1935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1958615497371416e-06,
+      "loss": 0.6307,
+      "step": 1936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.187587104109208e-06,
+      "loss": 0.608,
+      "step": 1937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1793263627384753e-06,
+      "loss": 0.6708,
+      "step": 1938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171079340115636e-06,
+      "loss": 0.5998,
+      "step": 1939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16284605070733e-06,
+      "loss": 0.673,
+      "step": 1940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.154626508956097e-06,
+      "loss": 0.57,
+      "step": 1941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1464207292803696e-06,
+      "loss": 0.6615,
+      "step": 1942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1382287260744283e-06,
+      "loss": 0.6227,
+      "step": 1943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.130050513708399e-06,
+      "loss": 0.6064,
+      "step": 1944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1218861065282137e-06,
+      "loss": 0.672,
+      "step": 1945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1137355188555796e-06,
+      "loss": 0.6932,
+      "step": 1946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.105598764987973e-06,
+      "loss": 0.6351,
+      "step": 1947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0974758591985945e-06,
+      "loss": 0.6846,
+      "step": 1948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08936681573636e-06,
+      "loss": 0.7049,
+      "step": 1949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0812716488258655e-06,
+      "loss": 0.6386,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0731903726673596e-06,
+      "loss": 0.6514,
+      "step": 1951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0651230014367385e-06,
+      "loss": 0.7024,
+      "step": 1952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.057069549285491e-06,
+      "loss": 0.6957,
+      "step": 1953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0490300303407017e-06,
+      "loss": 0.5995,
+      "step": 1954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041004458705006e-06,
+      "loss": 0.6713,
+      "step": 1955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0329928484565784e-06,
+      "loss": 0.5666,
+      "step": 1956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.024995213649099e-06,
+      "loss": 0.5551,
+      "step": 1957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.01701156831174e-06,
+      "loss": 0.5885,
+      "step": 1958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.00904192644912e-06,
+      "loss": 0.6711,
+      "step": 1959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0010863020413075e-06,
+      "loss": 0.5906,
+      "step": 1960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.993144709043777e-06,
+      "loss": 0.638,
+      "step": 1961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9852171613873837e-06,
+      "loss": 0.5898,
+      "step": 1962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977303672978357e-06,
+      "loss": 0.6459,
+      "step": 1963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969404257698253e-06,
+      "loss": 0.6811,
+      "step": 1964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961518929403944e-06,
+      "loss": 0.6946,
+      "step": 1965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9536477019275955e-06,
+      "loss": 0.651,
+      "step": 1966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9457905890766325e-06,
+      "loss": 0.5738,
+      "step": 1967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9379476046337285e-06,
+      "loss": 0.5757,
+      "step": 1968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9301187623567606e-06,
+      "loss": 0.6743,
+      "step": 1969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9223040759788138e-06,
+      "loss": 0.6349,
+      "step": 1970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9145035592081274e-06,
+      "loss": 0.6481,
+      "step": 1971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.906717225728094e-06,
+      "loss": 0.5878,
+      "step": 1972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8989450891972205e-06,
+      "loss": 0.634,
+      "step": 1973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8911871632491153e-06,
+      "loss": 0.6869,
+      "step": 1974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8834434614924567e-06,
+      "loss": 0.6659,
+      "step": 1975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8757139975109683e-06,
+      "loss": 0.5889,
+      "step": 1976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8679987848634063e-06,
+      "loss": 0.5565,
+      "step": 1977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8602978370835156e-06,
+      "loss": 0.628,
+      "step": 1978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.852611167680033e-06,
+      "loss": 0.5637,
+      "step": 1979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8449387901366366e-06,
+      "loss": 0.6013,
+      "step": 1980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8372807179119366e-06,
+      "loss": 0.6237,
+      "step": 1981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8296369644394562e-06,
+      "loss": 0.5399,
+      "step": 1982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8220075431275918e-06,
+      "loss": 0.5737,
+      "step": 1983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.814392467359607e-06,
+      "loss": 0.6142,
+      "step": 1984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.806791750493594e-06,
+      "loss": 0.6594,
+      "step": 1985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799205405862463e-06,
+      "loss": 0.6424,
+      "step": 1986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7916334467739083e-06,
+      "loss": 0.6634,
+      "step": 1987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7840758865103934e-06,
+      "loss": 0.6226,
+      "step": 1988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7765327383291187e-06,
+      "loss": 0.6087,
+      "step": 1989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7690040154620092e-06,
+      "loss": 0.5728,
+      "step": 1990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7614897311156864e-06,
+      "loss": 0.6584,
+      "step": 1991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7539898984714342e-06,
+      "loss": 0.5712,
+      "step": 1992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.746504530685199e-06,
+      "loss": 0.67,
+      "step": 1993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.739033640887544e-06,
+      "loss": 0.6612,
+      "step": 1994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7315772421836364e-06,
+      "loss": 0.5817,
+      "step": 1995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7241353476532307e-06,
+      "loss": 0.641,
+      "step": 1996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7167079703506296e-06,
+      "loss": 0.6266,
+      "step": 1997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7092951233046795e-06,
+      "loss": 0.6762,
+      "step": 1998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701896819518727e-06,
+      "loss": 0.5922,
+      "step": 1999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6945130719706205e-06,
+      "loss": 0.6613,
+      "step": 2000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6871438936126604e-06,
+      "loss": 0.6853,
+      "step": 2001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6797892973716057e-06,
+      "loss": 0.6097,
+      "step": 2002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6724492961486206e-06,
+      "loss": 0.6052,
+      "step": 2003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.665123902819279e-06,
+      "loss": 0.6344,
+      "step": 2004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6578131302335255e-06,
+      "loss": 0.6501,
+      "step": 2005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6505169912156548e-06,
+      "loss": 0.635,
+      "step": 2006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6432354985642984e-06,
+      "loss": 0.6982,
+      "step": 2007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6359686650523888e-06,
+      "loss": 0.6389,
+      "step": 2008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6287165034271503e-06,
+      "loss": 0.6041,
+      "step": 2009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6214790264100666e-06,
+      "loss": 0.6301,
+      "step": 2010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.614256246696858e-06,
+      "loss": 0.6172,
+      "step": 2011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6070481769574753e-06,
+      "loss": 0.5996,
+      "step": 2012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5998548298360527e-06,
+      "loss": 0.6409,
+      "step": 2013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5926762179509093e-06,
+      "loss": 0.6465,
+      "step": 2014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.585512353894505e-06,
+      "loss": 0.546,
+      "step": 2015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5783632502334411e-06,
+      "loss": 0.577,
+      "step": 2016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5712289195084185e-06,
+      "loss": 0.6104,
+      "step": 2017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5641093742342284e-06,
+      "loss": 0.5518,
+      "step": 2018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5570046268997209e-06,
+      "loss": 0.6396,
+      "step": 2019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5499146899677942e-06,
+      "loss": 0.5909,
+      "step": 2020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5428395758753655e-06,
+      "loss": 0.7107,
+      "step": 2021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.535779297033344e-06,
+      "loss": 0.5169,
+      "step": 2022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528733865826625e-06,
+      "loss": 0.5648,
+      "step": 2023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.521703294614052e-06,
+      "loss": 0.6138,
+      "step": 2024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5146875957284012e-06,
+      "loss": 0.5918,
+      "step": 2025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5076867814763629e-06,
+      "loss": 0.5838,
+      "step": 2026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5007008641385168e-06,
+      "loss": 0.6687,
+      "step": 2027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4937298559693136e-06,
+      "loss": 0.6345,
+      "step": 2028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4867737691970441e-06,
+      "loss": 0.5792,
+      "step": 2029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4798326160238342e-06,
+      "loss": 0.6309,
+      "step": 2030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4729064086256017e-06,
+      "loss": 0.5896,
+      "step": 2031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4659951591520593e-06,
+      "loss": 0.6591,
+      "step": 2032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4590988797266704e-06,
+      "loss": 0.5904,
+      "step": 2033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4522175824466456e-06,
+      "loss": 0.6664,
+      "step": 2034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.445351279382915e-06,
+      "loss": 0.614,
+      "step": 2035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4384999825800984e-06,
+      "loss": 0.6194,
+      "step": 2036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4316637040565029e-06,
+      "loss": 0.6533,
+      "step": 2037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4248424558040819e-06,
+      "loss": 0.7002,
+      "step": 2038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4180362497884247e-06,
+      "loss": 0.5797,
+      "step": 2039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4112450979487412e-06,
+      "loss": 0.653,
+      "step": 2040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4044690121978244e-06,
+      "loss": 0.6399,
+      "step": 2041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.397708004422047e-06,
+      "loss": 0.6636,
+      "step": 2042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3909620864813246e-06,
+      "loss": 0.7017,
+      "step": 2043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.384231270209111e-06,
+      "loss": 0.5476,
+      "step": 2044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3775155674123598e-06,
+      "loss": 0.617,
+      "step": 2045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.370814989871525e-06,
+      "loss": 0.6658,
+      "step": 2046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.364129549340516e-06,
+      "loss": 0.6601,
+      "step": 2047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3574592575466995e-06,
+      "loss": 0.586,
+      "step": 2048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.350804126190859e-06,
+      "loss": 0.5854,
+      "step": 2049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.344164166947194e-06,
+      "loss": 0.6498,
+      "step": 2050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.337539391463285e-06,
+      "loss": 0.6421,
+      "step": 2051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3309298113600755e-06,
+      "loss": 0.6372,
+      "step": 2052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3243354382318585e-06,
+      "loss": 0.6794,
+      "step": 2053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3177562836462487e-06,
+      "loss": 0.5993,
+      "step": 2054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3111923591441643e-06,
+      "loss": 0.5806,
+      "step": 2055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3046436762398073e-06,
+      "loss": 0.6225,
+      "step": 2056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2981102464206463e-06,
+      "loss": 0.6666,
+      "step": 2057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2915920811473937e-06,
+      "loss": 0.684,
+      "step": 2058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2850891918539787e-06,
+      "loss": 0.6658,
+      "step": 2059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2786015899475445e-06,
+      "loss": 0.6397,
+      "step": 2060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2721292868084068e-06,
+      "loss": 0.6532,
+      "step": 2061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2656722937900534e-06,
+      "loss": 0.5878,
+      "step": 2062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2592306222191086e-06,
+      "loss": 0.6735,
+      "step": 2063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2528042833953269e-06,
+      "loss": 0.6792,
+      "step": 2064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2463932885915643e-06,
+      "loss": 0.6204,
+      "step": 2065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2399976490537557e-06,
+      "loss": 0.611,
+      "step": 2066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2336173760009096e-06,
+      "loss": 0.6381,
+      "step": 2067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.227252480625074e-06,
+      "loss": 0.6973,
+      "step": 2068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.220902974091317e-06,
+      "loss": 0.5756,
+      "step": 2069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2145688675377243e-06,
+      "loss": 0.6774,
+      "step": 2070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2082501720753538e-06,
+      "loss": 0.6484,
+      "step": 2071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2019468987882433e-06,
+      "loss": 0.6147,
+      "step": 2072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.195659058733366e-06,
+      "loss": 0.6161,
+      "step": 2073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1893866629406315e-06,
+      "loss": 0.6277,
+      "step": 2074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1831297224128491e-06,
+      "loss": 0.593,
+      "step": 2075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.176888248125726e-06,
+      "loss": 0.5932,
+      "step": 2076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.170662251027831e-06,
+      "loss": 0.5988,
+      "step": 2077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.164451742040591e-06,
+      "loss": 0.5786,
+      "step": 2078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1582567320582561e-06,
+      "loss": 0.6803,
+      "step": 2079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1520772319478945e-06,
+      "loss": 0.6807,
+      "step": 2080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1459132525493677e-06,
+      "loss": 0.5992,
+      "step": 2081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1397648046753062e-06,
+      "loss": 0.6366,
+      "step": 2082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1336318991111028e-06,
+      "loss": 0.6556,
+      "step": 2083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1275145466148807e-06,
+      "loss": 0.6221,
+      "step": 2084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1214127579174804e-06,
+      "loss": 0.5919,
+      "step": 2085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1153265437224437e-06,
+      "loss": 0.6734,
+      "step": 2086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1092559147059912e-06,
+      "loss": 0.5559,
+      "step": 2087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1032008815170082e-06,
+      "loss": 0.6266,
+      "step": 2088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0971614547770138e-06,
+      "loss": 0.6294,
+      "step": 2089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0911376450801603e-06,
+      "loss": 0.578,
+      "step": 2090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0851294629931962e-06,
+      "loss": 0.565,
+      "step": 2091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0791369190554658e-06,
+      "loss": 0.6215,
+      "step": 2092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0731600237788731e-06,
+      "loss": 0.5673,
+      "step": 2093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0671987876478763e-06,
+      "loss": 0.6209,
+      "step": 2094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061253221119467e-06,
+      "loss": 0.6993,
+      "step": 2095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.055323334623143e-06,
+      "loss": 0.6755,
+      "step": 2096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0494091385609029e-06,
+      "loss": 0.6167,
+      "step": 2097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0435106433072195e-06,
+      "loss": 0.6113,
+      "step": 2098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0376278592090217e-06,
+      "loss": 0.624,
+      "step": 2099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0317607965856802e-06,
+      "loss": 0.6265,
+      "step": 2100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0259094657289893e-06,
+      "loss": 0.6008,
+      "step": 2101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.020073876903147e-06,
+      "loss": 0.6696,
+      "step": 2102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0142540403447321e-06,
+      "loss": 0.6099,
+      "step": 2103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0084499662627e-06,
+      "loss": 0.6025,
+      "step": 2104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0026616648383468e-06,
+      "loss": 0.6235,
+      "step": 2105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.968891462253084e-07,
+      "loss": 0.6619,
+      "step": 2106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.911324205495298e-07,
+      "loss": 0.5741,
+      "step": 2107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.853914979092571e-07,
+      "loss": 0.6666,
+      "step": 2108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.79666388375009e-07,
+      "loss": 0.6399,
+      "step": 2109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.739571019895721e-07,
+      "loss": 0.6142,
+      "step": 2110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.682636487679753e-07,
+      "loss": 0.5909,
+      "step": 2111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.625860386974705e-07,
+      "loss": 0.6401,
+      "step": 2112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.569242817375169e-07,
+      "loss": 0.6571,
+      "step": 2113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.512783878197706e-07,
+      "loss": 0.6035,
+      "step": 2114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.456483668480587e-07,
+      "loss": 0.6043,
+      "step": 2115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.400342286983599e-07,
+      "loss": 0.6343,
+      "step": 2116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.344359832187999e-07,
+      "loss": 0.5704,
+      "step": 2117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.288536402296211e-07,
+      "loss": 0.669,
+      "step": 2118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.232872095231693e-07,
+      "loss": 0.6577,
+      "step": 2119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.177367008638838e-07,
+      "loss": 0.6304,
+      "step": 2120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.12202123988265e-07,
+      "loss": 0.5941,
+      "step": 2121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.066834886048748e-07,
+      "loss": 0.5584,
+      "step": 2122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.011808043943038e-07,
+      "loss": 0.5959,
+      "step": 2123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.956940810091674e-07,
+      "loss": 0.6213,
+      "step": 2124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.902233280740824e-07,
+      "loss": 0.606,
+      "step": 2125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.847685551856455e-07,
+      "loss": 0.6536,
+      "step": 2126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.793297719124294e-07,
+      "loss": 0.5667,
+      "step": 2127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.739069877949524e-07,
+      "loss": 0.6364,
+      "step": 2128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.685002123456699e-07,
+      "loss": 0.5998,
+      "step": 2129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.631094550489538e-07,
+      "loss": 0.6742,
+      "step": 2130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.577347253610813e-07,
+      "loss": 0.5554,
+      "step": 2131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.523760327102126e-07,
+      "loss": 0.5789,
+      "step": 2132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.470333864963742e-07,
+      "loss": 0.6365,
+      "step": 2133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.417067960914482e-07,
+      "loss": 0.6097,
+      "step": 2134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.363962708391482e-07,
+      "loss": 0.6142,
+      "step": 2135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.311018200550114e-07,
+      "loss": 0.6367,
+      "step": 2136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.25823453026372e-07,
+      "loss": 0.6714,
+      "step": 2137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.20561179012358e-07,
+      "loss": 0.5737,
+      "step": 2138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.153150072438598e-07,
+      "loss": 0.6305,
+      "step": 2139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.100849469235272e-07,
+      "loss": 0.6203,
+      "step": 2140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.048710072257481e-07,
+      "loss": 0.68,
+      "step": 2141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.996731972966287e-07,
+      "loss": 0.6501,
+      "step": 2142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.944915262539798e-07,
+      "loss": 0.671,
+      "step": 2143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.893260031873096e-07,
+      "loss": 0.6335,
+      "step": 2144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.841766371577919e-07,
+      "loss": 0.5948,
+      "step": 2145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.790434371982625e-07,
+      "loss": 0.6684,
+      "step": 2146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.739264123131973e-07,
+      "loss": 0.5923,
+      "step": 2147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.68825571478703e-07,
+      "loss": 0.6126,
+      "step": 2148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.637409236424887e-07,
+      "loss": 0.602,
+      "step": 2149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.586724777238686e-07,
+      "loss": 0.5567,
+      "step": 2150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.53620242613724e-07,
+      "loss": 0.6304,
+      "step": 2151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.485842271745125e-07,
+      "loss": 0.6843,
+      "step": 2152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.435644402402298e-07,
+      "loss": 0.6658,
+      "step": 2153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.385608906164077e-07,
+      "loss": 0.6084,
+      "step": 2154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.335735870800975e-07,
+      "loss": 0.5483,
+      "step": 2155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.28602538379849e-07,
+      "loss": 0.6574,
+      "step": 2156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.236477532356978e-07,
+      "loss": 0.6799,
+      "step": 2157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.187092403391549e-07,
+      "loss": 0.5712,
+      "step": 2158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.137870083531818e-07,
+      "loss": 0.6346,
+      "step": 2159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.088810659121815e-07,
+      "loss": 0.5653,
+      "step": 2160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.039914216219867e-07,
+      "loss": 0.6856,
+      "step": 2161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.991180840598388e-07,
+      "loss": 0.6299,
+      "step": 2162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.942610617743706e-07,
+      "loss": 0.7086,
+      "step": 2163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.894203632856e-07,
+      "loss": 0.6392,
+      "step": 2164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.845959970849059e-07,
+      "loss": 0.6112,
+      "step": 2165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.797879716350242e-07,
+      "loss": 0.6383,
+      "step": 2166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.749962953700184e-07,
+      "loss": 0.5758,
+      "step": 2167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.702209766952794e-07,
+      "loss": 0.5865,
+      "step": 2168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.654620239874987e-07,
+      "loss": 0.5631,
+      "step": 2169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.607194455946641e-07,
+      "loss": 0.6528,
+      "step": 2170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5599324983604e-07,
+      "loss": 0.6998,
+      "step": 2171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.512834450021488e-07,
+      "loss": 0.5728,
+      "step": 2172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.465900393547641e-07,
+      "loss": 0.612,
+      "step": 2173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.419130411268881e-07,
+      "loss": 0.6587,
+      "step": 2174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.37252458522748e-07,
+      "loss": 0.6191,
+      "step": 2175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.326082997177718e-07,
+      "loss": 0.6054,
+      "step": 2176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.279805728585764e-07,
+      "loss": 0.6761,
+      "step": 2177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.233692860629592e-07,
+      "loss": 0.6088,
+      "step": 2178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.187744474198721e-07,
+      "loss": 0.628,
+      "step": 2179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.141960649894208e-07,
+      "loss": 0.5644,
+      "step": 2180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.096341468028377e-07,
+      "loss": 0.5896,
+      "step": 2181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.050887008624817e-07,
+      "loss": 0.6135,
+      "step": 2182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.005597351418091e-07,
+      "loss": 0.5854,
+      "step": 2183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.960472575853715e-07,
+      "loss": 0.5639,
+      "step": 2184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.915512761087993e-07,
+      "loss": 0.6133,
+      "step": 2185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.870717985987817e-07,
+      "loss": 0.6259,
+      "step": 2186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.826088329130575e-07,
+      "loss": 0.6688,
+      "step": 2187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.781623868804054e-07,
+      "loss": 0.5743,
+      "step": 2188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.737324683006218e-07,
+      "loss": 0.6643,
+      "step": 2189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.693190849445107e-07,
+      "loss": 0.6536,
+      "step": 2190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.649222445538749e-07,
+      "loss": 0.6991,
+      "step": 2191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.605419548414947e-07,
+      "loss": 0.5986,
+      "step": 2192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.561782234911195e-07,
+      "loss": 0.6193,
+      "step": 2193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.518310581574515e-07,
+      "loss": 0.6578,
+      "step": 2194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.475004664661321e-07,
+      "loss": 0.6087,
+      "step": 2195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.431864560137334e-07,
+      "loss": 0.6861,
+      "step": 2196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.388890343677389e-07,
+      "loss": 0.5871,
+      "step": 2197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.346082090665328e-07,
+      "loss": 0.5883,
+      "step": 2198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.303439876193861e-07,
+      "loss": 0.6302,
+      "step": 2199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.260963775064443e-07,
+      "loss": 0.5928,
+      "step": 2200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.218653861787171e-07,
+      "loss": 0.6021,
+      "step": 2201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.176510210580577e-07,
+      "loss": 0.6564,
+      "step": 2202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.134532895371546e-07,
+      "loss": 0.6093,
+      "step": 2203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.09272198979519e-07,
+      "loss": 0.5523,
+      "step": 2204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.051077567194729e-07,
+      "loss": 0.6599,
+      "step": 2205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.009599700621315e-07,
+      "loss": 0.6436,
+      "step": 2206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.968288462833959e-07,
+      "loss": 0.6187,
+      "step": 2207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.927143926299383e-07,
+      "loss": 0.6571,
+      "step": 2208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.88616616319183e-07,
+      "loss": 0.5974,
+      "step": 2209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.84535524539309e-07,
+      "loss": 0.6324,
+      "step": 2210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.804711244492166e-07,
+      "loss": 0.5792,
+      "step": 2211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.764234231785381e-07,
+      "loss": 0.5577,
+      "step": 2212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.723924278276015e-07,
+      "loss": 0.6164,
+      "step": 2213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6837814546743853e-07,
+      "loss": 0.6528,
+      "step": 2214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6438058313976295e-07,
+      "loss": 0.699,
+      "step": 2215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.603997478569533e-07,
+      "loss": 0.6786,
+      "step": 2216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5643564660205033e-07,
+      "loss": 0.6093,
+      "step": 2217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5248828632873944e-07,
+      "loss": 0.7176,
+      "step": 2218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4855767396134064e-07,
+      "loss": 0.5165,
+      "step": 2219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.44643816394793e-07,
+      "loss": 0.5428,
+      "step": 2220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4074672049464695e-07,
+      "loss": 0.5496,
+      "step": 2221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3686639309705183e-07,
+      "loss": 0.6485,
+      "step": 2222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3300284100873615e-07,
+      "loss": 0.5744,
+      "step": 2223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2915607100701084e-07,
+      "loss": 0.7151,
+      "step": 2224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.253260898397382e-07,
+      "loss": 0.5757,
+      "step": 2225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.215129042253385e-07,
+      "loss": 0.6434,
+      "step": 2226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.177165208527645e-07,
+      "loss": 0.5767,
+      "step": 2227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.139369463814991e-07,
+      "loss": 0.622,
+      "step": 2228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1017418744153437e-07,
+      "loss": 0.7016,
+      "step": 2229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.064282506333728e-07,
+      "loss": 0.6253,
+      "step": 2230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0269914252799914e-07,
+      "loss": 0.6492,
+      "step": 2231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.989868696668864e-07,
+      "loss": 0.5724,
+      "step": 2232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9529143856196995e-07,
+      "loss": 0.6454,
+      "step": 2233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9161285569564114e-07,
+      "loss": 0.7064,
+      "step": 2234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8795112752074483e-07,
+      "loss": 0.5768,
+      "step": 2235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.843062604605496e-07,
+      "loss": 0.5388,
+      "step": 2236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8067826090875336e-07,
+      "loss": 0.5691,
+      "step": 2237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7706713522946637e-07,
+      "loss": 0.6014,
+      "step": 2238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7347288975719374e-07,
+      "loss": 0.6078,
+      "step": 2239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.698955307968355e-07,
+      "loss": 0.5974,
+      "step": 2240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.663350646236663e-07,
+      "loss": 0.5968,
+      "step": 2241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.627914974833302e-07,
+      "loss": 0.5903,
+      "step": 2242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.59264835591826e-07,
+      "loss": 0.6221,
+      "step": 2243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.557550851354985e-07,
+      "loss": 0.6264,
+      "step": 2244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.522622522710306e-07,
+      "loss": 0.6753,
+      "step": 2245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4878634312542124e-07,
+      "loss": 0.6668,
+      "step": 2246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4532736379598974e-07,
+      "loss": 0.625,
+      "step": 2247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.418853203503525e-07,
+      "loss": 0.5118,
+      "step": 2248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.384602188264219e-07,
+      "loss": 0.5872,
+      "step": 2249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.350520652323863e-07,
+      "loss": 0.617,
+      "step": 2250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.316608655467113e-07,
+      "loss": 0.62,
+      "step": 2251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.282866257181183e-07,
+      "loss": 0.7399,
+      "step": 2252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2492935166557606e-07,
+      "loss": 0.6344,
+      "step": 2253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2158904927829913e-07,
+      "loss": 0.6004,
+      "step": 2254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.18265724415725e-07,
+      "loss": 0.6416,
+      "step": 2255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.149593829075137e-07,
+      "loss": 0.5987,
+      "step": 2256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.116700305535303e-07,
+      "loss": 0.6007,
+      "step": 2257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.083976731238414e-07,
+      "loss": 0.6365,
+      "step": 2258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0514231635869863e-07,
+      "loss": 0.6236,
+      "step": 2259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.019039659685341e-07,
+      "loss": 0.5542,
+      "step": 2260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9868262763394717e-07,
+      "loss": 0.5988,
+      "step": 2261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9547830700569545e-07,
+      "loss": 0.5921,
+      "step": 2262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.922910097046838e-07,
+      "loss": 0.628,
+      "step": 2263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8912074132195545e-07,
+      "loss": 0.6705,
+      "step": 2264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8596750741868295e-07,
+      "loss": 0.6937,
+      "step": 2265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8283131352615733e-07,
+      "loss": 0.6698,
+      "step": 2266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7971216514577906e-07,
+      "loss": 0.599,
+      "step": 2267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7661006774904687e-07,
+      "loss": 0.5504,
+      "step": 2268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.735250267775491e-07,
+      "loss": 0.6754,
+      "step": 2269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7045704764295686e-07,
+      "loss": 0.6503,
+      "step": 2270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.674061357270086e-07,
+      "loss": 0.6587,
+      "step": 2271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6437229638150655e-07,
+      "loss": 0.6321,
+      "step": 2272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.613555349283026e-07,
+      "loss": 0.6236,
+      "step": 2273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5835585665929365e-07,
+      "loss": 0.6365,
+      "step": 2274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.553732668364095e-07,
+      "loss": 0.6323,
+      "step": 2275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.524077706916028e-07,
+      "loss": 0.6625,
+      "step": 2276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4945937342684024e-07,
+      "loss": 0.5559,
+      "step": 2277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4652808021409567e-07,
+      "loss": 0.704,
+      "step": 2278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4361389619534157e-07,
+      "loss": 0.6453,
+      "step": 2279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4071682648253437e-07,
+      "loss": 0.6299,
+      "step": 2280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3783687615760998e-07,
+      "loss": 0.604,
+      "step": 2281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3497405027247845e-07,
+      "loss": 0.5761,
+      "step": 2282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3212835384900268e-07,
+      "loss": 0.6042,
+      "step": 2283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.292997918790063e-07,
+      "loss": 0.5963,
+      "step": 2284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2648836932425032e-07,
+      "loss": 0.6466,
+      "step": 2285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2369409111643204e-07,
+      "loss": 0.6451,
+      "step": 2286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2091696215717606e-07,
+      "loss": 0.6075,
+      "step": 2287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1815698731802227e-07,
+      "loss": 0.6647,
+      "step": 2288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.154141714404212e-07,
+      "loss": 0.6471,
+      "step": 2289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.126885193357231e-07,
+      "loss": 0.5999,
+      "step": 2290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0998003578517e-07,
+      "loss": 0.7041,
+      "step": 2291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.07288725539887e-07,
+      "loss": 0.5889,
+      "step": 2292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0461459332087653e-07,
+      "loss": 0.6521,
+      "step": 2293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0195764381900295e-07,
+      "loss": 0.6011,
+      "step": 2294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.99317881694997e-07,
+      "loss": 0.7214,
+      "step": 2295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9669531157943233e-07,
+      "loss": 0.6214,
+      "step": 2296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9408993807272903e-07,
+      "loss": 0.7617,
+      "step": 2297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9150176574514233e-07,
+      "loss": 0.6108,
+      "step": 2298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.889307991367506e-07,
+      "loss": 0.6425,
+      "step": 2299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8637704275745294e-07,
+      "loss": 0.657,
+      "step": 2300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8384050108695707e-07,
+      "loss": 0.5848,
+      "step": 2301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8132117857477593e-07,
+      "loss": 0.5859,
+      "step": 2302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.788190796402134e-07,
+      "loss": 0.6068,
+      "step": 2303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7633420867236518e-07,
+      "loss": 0.5742,
+      "step": 2304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.738665700301012e-07,
+      "loss": 0.5427,
+      "step": 2305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7141616804206784e-07,
+      "loss": 0.6511,
+      "step": 2306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6898300700667115e-07,
+      "loss": 0.5694,
+      "step": 2307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.665670911920758e-07,
+      "loss": 0.6221,
+      "step": 2308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6416842483619743e-07,
+      "loss": 0.6609,
+      "step": 2309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6178701214669023e-07,
+      "loss": 0.5805,
+      "step": 2310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5942285730094265e-07,
+      "loss": 0.6571,
+      "step": 2311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5707596444607288e-07,
+      "loss": 0.5605,
+      "step": 2312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5474633769891667e-07,
+      "loss": 0.6185,
+      "step": 2313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5243398114602182e-07,
+      "loss": 0.6843,
+      "step": 2314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5013889884364253e-07,
+      "loss": 0.7244,
+      "step": 2315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4786109481773058e-07,
+      "loss": 0.5748,
+      "step": 2316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4560057306392872e-07,
+      "loss": 0.5641,
+      "step": 2317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4335733754756388e-07,
+      "loss": 0.6369,
+      "step": 2318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4113139220364058e-07,
+      "loss": 0.644,
+      "step": 2319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3892274093683323e-07,
+      "loss": 0.6212,
+      "step": 2320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.367313876214782e-07,
+      "loss": 0.6529,
+      "step": 2321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.345573361015706e-07,
+      "loss": 0.6405,
+      "step": 2322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3240059019075546e-07,
+      "loss": 0.5578,
+      "step": 2323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.302611536723164e-07,
+      "loss": 0.5952,
+      "step": 2324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2813903029918028e-07,
+      "loss": 0.6696,
+      "step": 2325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.260342237938994e-07,
+      "loss": 0.6686,
+      "step": 2326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.239467378486503e-07,
+      "loss": 0.6727,
+      "step": 2327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2187657612522717e-07,
+      "loss": 0.7231,
+      "step": 2328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1982374225503412e-07,
+      "loss": 0.6203,
+      "step": 2329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.177882398390795e-07,
+      "loss": 0.6314,
+      "step": 2330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1577007244796823e-07,
+      "loss": 0.6109,
+      "step": 2331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1376924362189956e-07,
+      "loss": 0.6494,
+      "step": 2332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1178575687065485e-07,
+      "loss": 0.7083,
+      "step": 2333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0981961567359756e-07,
+      "loss": 0.5516,
+      "step": 2334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0787082347966104e-07,
+      "loss": 0.5751,
+      "step": 2335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0593938370734969e-07,
+      "loss": 0.5631,
+      "step": 2336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0402529974472442e-07,
+      "loss": 0.5252,
+      "step": 2337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0212857494940276e-07,
+      "loss": 0.5919,
+      "step": 2338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0024921264855436e-07,
+      "loss": 0.59,
+      "step": 2339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.83872161388888e-08,
+      "loss": 0.6296,
+      "step": 2340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.654258868665334e-08,
+      "loss": 0.5512,
+      "step": 2341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.471533352762962e-08,
+      "loss": 0.5536,
+      "step": 2342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.290545386712147e-08,
+      "loss": 0.6088,
+      "step": 2343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.111295287995814e-08,
+      "loss": 0.5847,
+      "step": 2344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.933783371047888e-08,
+      "loss": 0.5719,
+      "step": 2345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.758009947253621e-08,
+      "loss": 0.5565,
+      "step": 2346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.583975324948479e-08,
+      "loss": 0.5576,
+      "step": 2347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.411679809417706e-08,
+      "loss": 0.644,
+      "step": 2348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.241123702896093e-08,
+      "loss": 0.5563,
+      "step": 2349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.072307304567206e-08,
+      "loss": 0.6191,
+      "step": 2350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.9052309105625e-08,
+      "loss": 0.6436,
+      "step": 2351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.739894813961645e-08,
+      "loss": 0.6221,
+      "step": 2352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.576299304791312e-08,
+      "loss": 0.6291,
+      "step": 2353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.414444670024834e-08,
+      "loss": 0.672,
+      "step": 2354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.254331193581987e-08,
+      "loss": 0.617,
+      "step": 2355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.095959156328103e-08,
+      "loss": 0.6161,
+      "step": 2356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.939328836073733e-08,
+      "loss": 0.6588,
+      "step": 2357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.784440507574319e-08,
+      "loss": 0.6391,
+      "step": 2358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.631294442529412e-08,
+      "loss": 0.6367,
+      "step": 2359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.479890909582453e-08,
+      "loss": 0.6447,
+      "step": 2360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.330230174319996e-08,
+      "loss": 0.5896,
+      "step": 2361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.182312499271703e-08,
+      "loss": 0.6164,
+      "step": 2362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.036138143909464e-08,
+      "loss": 0.5986,
+      "step": 2363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.891707364647281e-08,
+      "loss": 0.6594,
+      "step": 2364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.749020414840267e-08,
+      "loss": 0.6121,
+      "step": 2365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6080775447849846e-08,
+      "loss": 0.6253,
+      "step": 2366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.46887900171833e-08,
+      "loss": 0.6113,
+      "step": 2367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3314250298173164e-08,
+      "loss": 0.5733,
+      "step": 2368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.195715870199181e-08,
+      "loss": 0.617,
+      "step": 2369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.06175176091972e-08,
+      "loss": 0.6222,
+      "step": 2370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9295329369741794e-08,
+      "loss": 0.6593,
+      "step": 2371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.799059630296144e-08,
+      "loss": 0.6411,
+      "step": 2372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6703320697572e-08,
+      "loss": 0.5944,
+      "step": 2373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5433504811667193e-08,
+      "loss": 0.5662,
+      "step": 2374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.418115087271302e-08,
+      "loss": 0.5919,
+      "step": 2375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2946261077544404e-08,
+      "loss": 0.6008,
+      "step": 2376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.172883759235968e-08,
+      "loss": 0.6319,
+      "step": 2377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0528882552720584e-08,
+      "loss": 0.632,
+      "step": 2378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9346398063546676e-08,
+      "loss": 0.665,
+      "step": 2379
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8181386199109825e-08,
+      "loss": 0.6213,
+      "step": 2380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7033849003031975e-08,
+      "loss": 0.6881,
+      "step": 2381
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.59037884882818e-08,
+      "loss": 0.6056,
+      "step": 2382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4791206637172504e-08,
+      "loss": 0.5901,
+      "step": 2383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.369610540135404e-08,
+      "loss": 0.6251,
+      "step": 2384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.261848670181533e-08,
+      "loss": 0.6456,
+      "step": 2385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.15583524288765e-08,
+      "loss": 0.6233,
+      "step": 2386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.051570444218665e-08,
+      "loss": 0.6181,
+      "step": 2387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9490544570723867e-08,
+      "loss": 0.7135,
+      "step": 2388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8482874612785205e-08,
+      "loss": 0.6401,
+      "step": 2389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7492696335991165e-08,
+      "loss": 0.5945,
+      "step": 2390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6520011477274567e-08,
+      "loss": 0.585,
+      "step": 2391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5564821742888324e-08,
+      "loss": 0.6019,
+      "step": 2392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.46271288083888e-08,
+      "loss": 0.5982,
+      "step": 2393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.37069343186469e-08,
+      "loss": 0.6452,
+      "step": 2394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2804239887832536e-08,
+      "loss": 0.5608,
+      "step": 2395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1919047099421274e-08,
+      "loss": 0.6123,
+      "step": 2396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.10513575061877e-08,
+      "loss": 0.5814,
+      "step": 2397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0201172630199827e-08,
+      "loss": 0.6203,
+      "step": 2398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9368493962823585e-08,
+      "loss": 0.6164,
+      "step": 2399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8553322964713906e-08,
+      "loss": 0.5999,
+      "step": 2400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7755661065814723e-08,
+      "loss": 0.6118,
+      "step": 2401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.697550966535566e-08,
+      "loss": 0.5894,
+      "step": 2402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.621287013185202e-08,
+      "loss": 0.5942,
+      "step": 2403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.546774380309701e-08,
+      "loss": 0.5385,
+      "step": 2404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4740131986166196e-08,
+      "loss": 0.6258,
+      "step": 2405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4030035957410815e-08,
+      "loss": 0.576,
+      "step": 2406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3337456962454476e-08,
+      "loss": 0.5796,
+      "step": 2407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2662396216196471e-08,
+      "loss": 0.5841,
+      "step": 2408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2004854902801789e-08,
+      "loss": 0.6282,
+      "step": 2409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1364834175708883e-08,
+      "loss": 0.6947,
+      "step": 2410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0742335157618578e-08,
+      "loss": 0.5592,
+      "step": 2411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0137358940495168e-08,
+      "loss": 0.6063,
+      "step": 2412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.549906585567536e-09,
+      "loss": 0.623,
+      "step": 2413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.979979123323601e-09,
+      "loss": 0.5986,
+      "step": 2414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.427577553509203e-09,
+      "loss": 0.6024,
+      "step": 2415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.892702845129218e-09,
+      "loss": 0.6173,
+      "step": 2416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.375355936440898e-09,
+      "loss": 0.7068,
+      "step": 2417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.875537734954973e-09,
+      "loss": 0.6872,
+      "step": 2418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3932491174356626e-09,
+      "loss": 0.6079,
+      "step": 2419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.928490929897335e-09,
+      "loss": 0.5665,
+      "step": 2420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4812639876000715e-09,
+      "loss": 0.6371,
+      "step": 2421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.051569075056328e-09,
+      "loss": 0.5822,
+      "step": 2422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.639406946018721e-09,
+      "loss": 0.6264,
+      "step": 2423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2447783234900175e-09,
+      "loss": 0.625,
+      "step": 2424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.867683899712038e-09,
+      "loss": 0.6325,
+      "step": 2425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5081243361700934e-09,
+      "loss": 0.6229,
+      "step": 2426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.166100263592986e-09,
+      "loss": 0.6136,
+      "step": 2427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8416122819452384e-09,
+      "loss": 0.6242,
+      "step": 2428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5346609604326446e-09,
+      "loss": 0.6293,
+      "step": 2429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.245246837498938e-09,
+      "loss": 0.5637,
+      "step": 2430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.973370420823573e-09,
+      "loss": 0.5801,
+      "step": 2431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.719032187321723e-09,
+      "loss": 0.6315,
+      "step": 2432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4822325831453932e-09,
+      "loss": 0.6365,
+      "step": 2433
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2629720236800868e-09,
+      "loss": 0.5855,
+      "step": 2434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0612508935459176e-09,
+      "loss": 0.6802,
+      "step": 2435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.770695465920576e-10,
+      "loss": 0.5994,
+      "step": 2436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.104283059056194e-10,
+      "loss": 0.5722,
+      "step": 2437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.613274638005539e-10,
+      "loss": 0.4984,
+      "step": 2438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2976728182653193e-10,
+      "loss": 0.6701,
+      "step": 2439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1574799076006247e-10,
+      "loss": 0.5628,
+      "step": 2440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1926979061004382e-10,
+      "loss": 0.6036,
+      "step": 2441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4033285061554325e-10,
+      "loss": 0.6475,
+      "step": 2442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.893730924468657e-11,
+      "loss": 0.6088,
+      "step": 2443
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.508327419576851e-11,
+      "loss": 0.6262,
+      "step": 2444
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.770822395032241e-12,
+      "loss": 0.5745,
+      "step": 2445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.6366,
+      "step": 2446
+    },
+    {
+      "epoch": 1.0,
+      "step": 2446,
+      "total_flos": 191695072002048.0,
+      "train_loss": 0.6746118150815941,
+      "train_runtime": 25368.0481,
+      "train_samples_per_second": 12.339,
+      "train_steps_per_second": 0.096
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2446,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 191695072002048.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}