diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7752 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9965075669383,
+  "eval_steps": 500,
+  "global_step": 1287,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 0.8856,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 0.8834,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 0.8061,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 0.8177,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 0.8807,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.8335,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 0.8102,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 0.7847,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.7545,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 0.8027,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 0.7502,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.8542,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.8134,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 0.7582,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.7644,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 0.8194,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 0.7587,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.7776,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 0.7041,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.7404,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.7741,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 0.7662,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 0.7667,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.7431,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 0.7801,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.729,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.7712,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 0.7548,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 0.7692,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.7798,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 0.7421,
+      "step": 31
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 0.7973,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.7481,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 0.7601,
+      "step": 34
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.7442,
+      "step": 35
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.7078,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 0.7348,
+      "step": 37
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 0.6966,
+      "step": 38
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.7396,
+      "step": 39
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999968315974827e-05,
+      "loss": 0.7457,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999873264100076e-05,
+      "loss": 0.8337,
+      "step": 41
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999971484497808e-05,
+      "loss": 0.778,
+      "step": 42
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99994930596127e-05,
+      "loss": 0.7444,
+      "step": 43
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999207909409354e-05,
+      "loss": 0.7344,
+      "step": 44
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998859396174982e-05,
+      "loss": 0.7424,
+      "step": 45
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999844752211804e-05,
+      "loss": 0.7425,
+      "step": 46
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997972289848505e-05,
+      "loss": 0.7307,
+      "step": 47
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999743370237782e-05,
+      "loss": 0.7106,
+      "step": 48
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999683176311891e-05,
+      "loss": 0.7966,
+      "step": 49
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999616647588615e-05,
+      "loss": 0.7336,
+      "step": 50
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995437844895337e-05,
+      "loss": 0.7263,
+      "step": 51
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994645874763657e-05,
+      "loss": 0.8042,
+      "step": 52
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993790570509678e-05,
+      "loss": 0.7462,
+      "step": 53
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992871937553292e-05,
+      "loss": 0.7796,
+      "step": 54
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991889981715696e-05,
+      "loss": 0.8143,
+      "step": 55
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990844709219357e-05,
+      "loss": 0.7513,
+      "step": 56
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998973612668796e-05,
+      "loss": 0.7069,
+      "step": 57
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998856424114638e-05,
+      "loss": 0.7357,
+      "step": 58
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987329060020616e-05,
+      "loss": 0.7408,
+      "step": 59
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986030591137785e-05,
+      "loss": 0.7381,
+      "step": 60
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998466884272602e-05,
+      "loss": 0.761,
+      "step": 61
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998324382341446e-05,
+      "loss": 0.7462,
+      "step": 62
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981755542233175e-05,
+      "loss": 0.7614,
+      "step": 63
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980204008613112e-05,
+      "loss": 0.7148,
+      "step": 64
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978589232386036e-05,
+      "loss": 0.7451,
+      "step": 65
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976911223784473e-05,
+      "loss": 0.7229,
+      "step": 66
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975169993441627e-05,
+      "loss": 0.7264,
+      "step": 67
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973365552391346e-05,
+      "loss": 0.7643,
+      "step": 68
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971497912068014e-05,
+      "loss": 0.7341,
+      "step": 69
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969567084306506e-05,
+      "loss": 0.7651,
+      "step": 70
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967573081342103e-05,
+      "loss": 0.7713,
+      "step": 71
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965515915810408e-05,
+      "loss": 0.6778,
+      "step": 72
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963395600747285e-05,
+      "loss": 0.7484,
+      "step": 73
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996121214958875e-05,
+      "loss": 0.7411,
+      "step": 74
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995896557617091e-05,
+      "loss": 0.7596,
+      "step": 75
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956655894729866e-05,
+      "loss": 0.7661,
+      "step": 76
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954283119901613e-05,
+      "loss": 0.6651,
+      "step": 77
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995184726672197e-05,
+      "loss": 0.7518,
+      "step": 78
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949348350626456e-05,
+      "loss": 0.7212,
+      "step": 79
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946786387450223e-05,
+      "loss": 0.7507,
+      "step": 80
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944161393427923e-05,
+      "loss": 0.7249,
+      "step": 81
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994147338519364e-05,
+      "loss": 0.7747,
+      "step": 82
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938722379780747e-05,
+      "loss": 0.6898,
+      "step": 83
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935908394621844e-05,
+      "loss": 0.7072,
+      "step": 84
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933031447548593e-05,
+      "loss": 0.7016,
+      "step": 85
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930091556791656e-05,
+      "loss": 0.7033,
+      "step": 86
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992708874098054e-05,
+      "loss": 0.7209,
+      "step": 87
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992402301914351e-05,
+      "loss": 0.6539,
+      "step": 88
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920894410707446e-05,
+      "loss": 0.7334,
+      "step": 89
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917702935497725e-05,
+      "loss": 0.7615,
+      "step": 90
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914448613738107e-05,
+      "loss": 0.7591,
+      "step": 91
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911131466050594e-05,
+      "loss": 0.7106,
+      "step": 92
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99077515134553e-05,
+      "loss": 0.7583,
+      "step": 93
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990430877737034e-05,
+      "loss": 0.7018,
+      "step": 94
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900803279611643e-05,
+      "loss": 0.7175,
+      "step": 95
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897235042392876e-05,
+      "loss": 0.6996,
+      "step": 96
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893604088325258e-05,
+      "loss": 0.7738,
+      "step": 97
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988991044041744e-05,
+      "loss": 0.704,
+      "step": 98
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886154122075344e-05,
+      "loss": 0.73,
+      "step": 99
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882335157102033e-05,
+      "loss": 0.6545,
+      "step": 100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987845356969754e-05,
+      "loss": 0.7122,
+      "step": 101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874509384458726e-05,
+      "loss": 0.7811,
+      "step": 102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870502626379127e-05,
+      "loss": 0.7283,
+      "step": 103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866433320848793e-05,
+      "loss": 0.7103,
+      "step": 104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986230149365411e-05,
+      "loss": 0.6781,
+      "step": 105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985810717097766e-05,
+      "loss": 0.7584,
+      "step": 106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985385037939806e-05,
+      "loss": 0.6629,
+      "step": 107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849531145889758e-05,
+      "loss": 0.7587,
+      "step": 108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845149497822898e-05,
+      "loss": 0.6596,
+      "step": 109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984070546296313e-05,
+      "loss": 0.6613,
+      "step": 110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983619906947144e-05,
+      "loss": 0.7364,
+      "step": 111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831630345903953e-05,
+      "loss": 0.7328,
+      "step": 112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826999321211793e-05,
+      "loss": 0.7765,
+      "step": 113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822306024740855e-05,
+      "loss": 0.6444,
+      "step": 114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817550486231645e-05,
+      "loss": 0.7316,
+      "step": 115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812732735819078e-05,
+      "loss": 0.7047,
+      "step": 116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 0.7178,
+      "step": 117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802910721794505e-05,
+      "loss": 0.6992,
+      "step": 118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979790652042268e-05,
+      "loss": 0.7188,
+      "step": 119
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792840231627482e-05,
+      "loss": 0.754,
+      "step": 120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787711887513e-05,
+      "loss": 0.7017,
+      "step": 121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978252152057655e-05,
+      "loss": 0.6756,
+      "step": 122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977726916370847e-05,
+      "loss": 0.6991,
+      "step": 123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771954850191927e-05,
+      "loss": 0.7786,
+      "step": 124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766578613702683e-05,
+      "loss": 0.6866,
+      "step": 125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976114048830891e-05,
+      "loss": 0.7181,
+      "step": 126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975564050847094e-05,
+      "loss": 0.7016,
+      "step": 127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750078709041077e-05,
+      "loss": 0.7472,
+      "step": 128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974445512526336e-05,
+      "loss": 0.6819,
+      "step": 129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738769792773338e-05,
+      "loss": 0.7272,
+      "step": 130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973302274759786e-05,
+      "loss": 0.711,
+      "step": 131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727214026154827e-05,
+      "loss": 0.7268,
+      "step": 132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972134366525298e-05,
+      "loss": 0.712,
+      "step": 133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971541170209165e-05,
+      "loss": 0.662,
+      "step": 134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 0.729,
+      "step": 135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703363119739424e-05,
+      "loss": 0.7034,
+      "step": 136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969724657689805e-05,
+      "loss": 0.7096,
+      "step": 137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691068584495744e-05,
+      "loss": 0.7043,
+      "step": 138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684829181681236e-05,
+      "loss": 0.7244,
+      "step": 139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678528407992404e-05,
+      "loss": 0.714,
+      "step": 140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967216630335603e-05,
+      "loss": 0.6707,
+      "step": 141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966574290808752e-05,
+      "loss": 0.7164,
+      "step": 142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.7373,
+      "step": 143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652712408857452e-05,
+      "loss": 0.7256,
+      "step": 144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646105387467627e-05,
+      "loss": 0.699,
+      "step": 145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639437240588618e-05,
+      "loss": 0.751,
+      "step": 146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632708010475166e-05,
+      "loss": 0.7514,
+      "step": 147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962591773976909e-05,
+      "loss": 0.6713,
+      "step": 148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619066471499017e-05,
+      "loss": 0.7549,
+      "step": 149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961215424908009e-05,
+      "loss": 0.6804,
+      "step": 150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605181116313725e-05,
+      "loss": 0.7296,
+      "step": 151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598147117387298e-05,
+      "loss": 0.7234,
+      "step": 152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959105229687389e-05,
+      "loss": 0.7569,
+      "step": 153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583896699731996e-05,
+      "loss": 0.7144,
+      "step": 154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957668037130524e-05,
+      "loss": 0.7082,
+      "step": 155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956940335732209e-05,
+      "loss": 0.7199,
+      "step": 156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956206570389556e-05,
+      "loss": 0.751,
+      "step": 157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554667457522936e-05,
+      "loss": 0.7553,
+      "step": 158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954720866508546e-05,
+      "loss": 0.7216,
+      "step": 159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953968937384804e-05,
+      "loss": 0.7342,
+      "step": 160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532109631458963e-05,
+      "loss": 0.766,
+      "step": 161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524469485949586e-05,
+      "loss": 0.7734,
+      "step": 162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516768985734008e-05,
+      "loss": 0.777,
+      "step": 163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509008179608804e-05,
+      "loss": 0.7557,
+      "step": 164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501187116752694e-05,
+      "loss": 0.6586,
+      "step": 165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949330584672622e-05,
+      "loss": 0.6987,
+      "step": 166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485364419471454e-05,
+      "loss": 0.6163,
+      "step": 167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477362885311684e-05,
+      "loss": 0.7489,
+      "step": 168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946930129495106e-05,
+      "loss": 0.6906,
+      "step": 169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946117969947431e-05,
+      "loss": 0.7249,
+      "step": 170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452998150346403e-05,
+      "loss": 0.6988,
+      "step": 171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444756699412225e-05,
+      "loss": 0.6643,
+      "step": 172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436455398896234e-05,
+      "loss": 0.6682,
+      "step": 173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9428094301402164e-05,
+      "loss": 0.7114,
+      "step": 174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419673459912652e-05,
+      "loss": 0.7416,
+      "step": 175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941119292778893e-05,
+      "loss": 0.6653,
+      "step": 176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402652758770476e-05,
+      "loss": 0.7231,
+      "step": 177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939405300697468e-05,
+      "loss": 0.7067,
+      "step": 178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385393726896492e-05,
+      "loss": 0.672,
+      "step": 179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376674973408077e-05,
+      "loss": 0.7218,
+      "step": 180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367896801758478e-05,
+      "loss": 0.7046,
+      "step": 181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935905926757326e-05,
+      "loss": 0.6373,
+      "step": 182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350162426854152e-05,
+      "loss": 0.69,
+      "step": 183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341206335978695e-05,
+      "loss": 0.7118,
+      "step": 184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93321910516999e-05,
+      "loss": 0.7281,
+      "step": 185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932311663114586e-05,
+      "loss": 0.7024,
+      "step": 186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313983131819406e-05,
+      "loss": 0.7186,
+      "step": 187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930479061159775e-05,
+      "loss": 0.6939,
+      "step": 188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295539128732096e-05,
+      "loss": 0.7606,
+      "step": 189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286228741847286e-05,
+      "loss": 0.7478,
+      "step": 190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927685950994143e-05,
+      "loss": 0.7428,
+      "step": 191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267431492385524e-05,
+      "loss": 0.7229,
+      "step": 192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257944748923074e-05,
+      "loss": 0.6717,
+      "step": 193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924839933966972e-05,
+      "loss": 0.7366,
+      "step": 194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.6973,
+      "step": 195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922913276611128e-05,
+      "loss": 0.709,
+      "step": 196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219411723894717e-05,
+      "loss": 0.6879,
+      "step": 197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920963226006352e-05,
+      "loss": 0.6869,
+      "step": 198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199794436588244e-05,
+      "loss": 0.691,
+      "step": 199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918989831580926e-05,
+      "loss": 0.7009,
+      "step": 200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917994396043636e-05,
+      "loss": 0.6796,
+      "step": 201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916993143354835e-05,
+      "loss": 0.7341,
+      "step": 202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915986079859266e-05,
+      "loss": 0.6873,
+      "step": 203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149732119384942e-05,
+      "loss": 0.7652,
+      "step": 204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139545460108658e-05,
+      "loss": 0.7347,
+      "step": 205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912930088531469e-05,
+      "loss": 0.7757,
+      "step": 206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91189984599209e-05,
+      "loss": 0.6867,
+      "step": 207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910863824921176e-05,
+      "loss": 0.7747,
+      "step": 208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098220318837897e-05,
+      "loss": 0.7027,
+      "step": 209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908774473481571e-05,
+      "loss": 0.697,
+      "step": 210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077211563526928e-05,
+      "loss": 0.6824,
+      "step": 211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066620871718207e-05,
+      "loss": 0.7084,
+      "step": 212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055972726500696e-05,
+      "loss": 0.7515,
+      "step": 213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904526719534961e-05,
+      "loss": 0.7149,
+      "step": 214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9034504346103825e-05,
+      "loss": 0.6818,
+      "step": 215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023684246965407e-05,
+      "loss": 0.7422,
+      "step": 216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012806966499218e-05,
+      "loss": 0.7077,
+      "step": 217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001872573632462e-05,
+      "loss": 0.7033,
+      "step": 218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899088113765426e-05,
+      "loss": 0.716,
+      "step": 219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89798327282152e-05,
+      "loss": 0.7025,
+      "step": 220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968727415326885e-05,
+      "loss": 0.702,
+      "step": 221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895756526936153e-05,
+      "loss": 0.7393,
+      "step": 222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946346361051474e-05,
+      "loss": 0.6504,
+      "step": 223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8935070761488754e-05,
+      "loss": 0.6957,
+      "step": 224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8923738542124644e-05,
+      "loss": 0.6663,
+      "step": 225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8912349774769208e-05,
+      "loss": 0.6942,
+      "step": 226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900904531590845e-05,
+      "loss": 0.6908,
+      "step": 227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8889402885115834e-05,
+      "loss": 0.6553,
+      "step": 228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8877844908227857e-05,
+      "loss": 0.6926,
+      "step": 229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886623067416757e-05,
+      "loss": 0.6946,
+      "step": 230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 0.7009,
+      "step": 231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842833729274617e-05,
+      "loss": 0.6793,
+      "step": 232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8831051166703832e-05,
+      "loss": 0.7145,
+      "step": 233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881921264348355e-05,
+      "loss": 0.6865,
+      "step": 234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807318234632187e-05,
+      "loss": 0.6813,
+      "step": 235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795368015522287e-05,
+      "loss": 0.7086,
+      "step": 236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783362061880063e-05,
+      "loss": 0.6379,
+      "step": 237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8771300449784898e-05,
+      "loss": 0.699,
+      "step": 238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875918325566888e-05,
+      "loss": 0.6959,
+      "step": 239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747010556316304e-05,
+      "loss": 0.7168,
+      "step": 240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8734782428863196e-05,
+      "loss": 0.6704,
+      "step": 241
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872249895079681e-05,
+      "loss": 0.7426,
+      "step": 242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8710160199955158e-05,
+      "loss": 0.6919,
+      "step": 243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869776625452649e-05,
+      "loss": 0.7646,
+      "step": 244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685317193048833e-05,
+      "loss": 0.705,
+      "step": 245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672813094409453e-05,
+      "loss": 0.6666,
+      "step": 246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.7231,
+      "step": 247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647640102937932e-05,
+      "loss": 0.758,
+      "step": 248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863497136962213e-05,
+      "loss": 0.6631,
+      "step": 249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8622247918176276e-05,
+      "loss": 0.6951,
+      "step": 250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86094698292264e-05,
+      "loss": 0.6584,
+      "step": 251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8596637183744762e-05,
+      "loss": 0.7485,
+      "step": 252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8583750063049338e-05,
+      "loss": 0.671,
+      "step": 253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857080854880329e-05,
+      "loss": 0.6642,
+      "step": 254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557812723014476e-05,
+      "loss": 0.6836,
+      "step": 255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854476266803491e-05,
+      "loss": 0.7021,
+      "step": 256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8531658466560248e-05,
+      "loss": 0.6727,
+      "step": 257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851850020162926e-05,
+      "loss": 0.6651,
+      "step": 258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8505287956623298e-05,
+      "loss": 0.6619,
+      "step": 259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849202181526579e-05,
+      "loss": 0.6892,
+      "step": 260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478701861621686e-05,
+      "loss": 0.6941,
+      "step": 261
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8465328180096936e-05,
+      "loss": 0.7294,
+      "step": 262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845190085543795e-05,
+      "loss": 0.7293,
+      "step": 263
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8438419972731066e-05,
+      "loss": 0.6863,
+      "step": 264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8424885617402016e-05,
+      "loss": 0.6956,
+      "step": 265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8411297875215363e-05,
+      "loss": 0.712,
+      "step": 266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8397656832273982e-05,
+      "loss": 0.6647,
+      "step": 267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383962575018503e-05,
+      "loss": 0.7132,
+      "step": 268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8370215190226768e-05,
+      "loss": 0.6865,
+      "step": 269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356414765013267e-05,
+      "loss": 0.7043,
+      "step": 270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8342561386828613e-05,
+      "loss": 0.677,
+      "step": 271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8328655143458958e-05,
+      "loss": 0.7306,
+      "step": 272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 0.6944,
+      "step": 273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300684413983692e-05,
+      "loss": 0.7091,
+      "step": 274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828662010512314e-05,
+      "loss": 0.6644,
+      "step": 275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8272503285566587e-05,
+      "loss": 0.7102,
+      "step": 276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8258334044769558e-05,
+      "loss": 0.6988,
+      "step": 277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244112472519772e-05,
+      "loss": 0.632,
+      "step": 278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229838658936566e-05,
+      "loss": 0.6531,
+      "step": 279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8215512694470304e-05,
+      "loss": 0.6507,
+      "step": 280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8201134669901838e-05,
+      "loss": 0.7119,
+      "step": 281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81867046763419e-05,
+      "loss": 0.6645,
+      "step": 282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817222280523055e-05,
+      "loss": 0.7481,
+      "step": 283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157689148336578e-05,
+      "loss": 0.6979,
+      "step": 284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8143103797756942e-05,
+      "loss": 0.6906,
+      "step": 285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8128466845916156e-05,
+      "loss": 0.6546,
+      "step": 286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811377838556573e-05,
+      "loss": 0.639,
+      "step": 287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8099038509783586e-05,
+      "loss": 0.6769,
+      "step": 288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8084247311973426e-05,
+      "loss": 0.6531,
+      "step": 289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069404885864198e-05,
+      "loss": 0.6508,
+      "step": 290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805451132550946e-05,
+      "loss": 0.6991,
+      "step": 291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80395667252868e-05,
+      "loss": 0.6991,
+      "step": 292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8024571179897236e-05,
+      "loss": 0.6984,
+      "step": 293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8009524784364615e-05,
+      "loss": 0.7872,
+      "step": 294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7994427634035016e-05,
+      "loss": 0.6956,
+      "step": 295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7979279824576127e-05,
+      "loss": 0.6801,
+      "step": 296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7964081451976673e-05,
+      "loss": 0.6579,
+      "step": 297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794883261254577e-05,
+      "loss": 0.7602,
+      "step": 298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7933533402912354e-05,
+      "loss": 0.7077,
+      "step": 299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791818392002452e-05,
+      "loss": 0.7342,
+      "step": 300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7902784261148952e-05,
+      "loss": 0.7036,
+      "step": 301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788733452387029e-05,
+      "loss": 0.6809,
+      "step": 302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7871834806090502e-05,
+      "loss": 0.6297,
+      "step": 303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7856285206028282e-05,
+      "loss": 0.7425,
+      "step": 304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784068582221841e-05,
+      "loss": 0.6649,
+      "step": 305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7825036753511143e-05,
+      "loss": 0.6832,
+      "step": 306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.780933809907158e-05,
+      "loss": 0.69,
+      "step": 307
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779358995837902e-05,
+      "loss": 0.6768,
+      "step": 308
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777792431226384e-05,
+      "loss": 0.7612,
+      "step": 309
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761945617719502e-05,
+      "loss": 0.6974,
+      "step": 310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7746049618276545e-05,
+      "loss": 0.6653,
+      "step": 311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773010453362737e-05,
+      "loss": 0.6423,
+      "step": 312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7714110464812865e-05,
+      "loss": 0.6814,
+      "step": 313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698067513184326e-05,
+      "loss": 0.695,
+      "step": 314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7681975780402807e-05,
+      "loss": 0.6734,
+      "step": 315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7665835368438483e-05,
+      "loss": 0.6903,
+      "step": 316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649646379570003e-05,
+      "loss": 0.6607,
+      "step": 317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7633408916383826e-05,
+      "loss": 0.6263,
+      "step": 318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761712308177359e-05,
+      "loss": 0.6718,
+      "step": 319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600788978939463e-05,
+      "loss": 0.6668,
+      "step": 320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7584406711387462e-05,
+      "loss": 0.7012,
+      "step": 321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7567976382928823e-05,
+      "loss": 0.668,
+      "step": 322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755149809767934e-05,
+      "loss": 0.6368,
+      "step": 323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7534971960058684e-05,
+      "loss": 0.7012,
+      "step": 324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7518398074789776e-05,
+      "loss": 0.6769,
+      "step": 325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750177654689809e-05,
+      "loss": 0.6835,
+      "step": 326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 0.7005,
+      "step": 327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7468390984857157e-05,
+      "loss": 0.7436,
+      "step": 328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7451627162265703e-05,
+      "loss": 0.6826,
+      "step": 329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743481612016573e-05,
+      "loss": 0.7138,
+      "step": 330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7417957965085532e-05,
+      "loss": 0.6944,
+      "step": 331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740105280385195e-05,
+      "loss": 0.669,
+      "step": 332
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7384100743589698e-05,
+      "loss": 0.6957,
+      "step": 333
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7367101891720672e-05,
+      "loss": 0.6974,
+      "step": 334
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7350056355963287e-05,
+      "loss": 0.6979,
+      "step": 335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733296424433178e-05,
+      "loss": 0.6704,
+      "step": 336
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731582566513552e-05,
+      "loss": 0.6833,
+      "step": 337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298640726978357e-05,
+      "loss": 0.6944,
+      "step": 338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7281409538757886e-05,
+      "loss": 0.6986,
+      "step": 339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264132209664785e-05,
+      "loss": 0.6958,
+      "step": 340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7246808849182125e-05,
+      "loss": 0.6992,
+      "step": 341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722943956708466e-05,
+      "loss": 0.6826,
+      "step": 342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7212024473438145e-05,
+      "loss": 0.6613,
+      "step": 343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7194563678598638e-05,
+      "loss": 0.6874,
+      "step": 344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7177057293211786e-05,
+      "loss": 0.7067,
+      "step": 345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715950542821214e-05,
+      "loss": 0.7072,
+      "step": 346
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7141908194822448e-05,
+      "loss": 0.7222,
+      "step": 347
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124265704552948e-05,
+      "loss": 0.7074,
+      "step": 348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710657806920066e-05,
+      "loss": 0.6994,
+      "step": 349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7088845400848684e-05,
+      "loss": 0.7031,
+      "step": 350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.6955,
+      "step": 351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705324541490416e-05,
+      "loss": 0.6589,
+      "step": 352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703537832290178e-05,
+      "loss": 0.7225,
+      "step": 353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701746664907862e-05,
+      "loss": 0.7101,
+      "step": 354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6999510506937465e-05,
+      "loss": 0.697,
+      "step": 355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981510010262892e-05,
+      "loss": 0.7174,
+      "step": 356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696346527312053e-05,
+      "loss": 0.7193,
+      "step": 357
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6945376409856366e-05,
+      "loss": 0.7524,
+      "step": 358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6927243535095995e-05,
+      "loss": 0.6654,
+      "step": 359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909066763743914e-05,
+      "loss": 0.7009,
+      "step": 360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6890846210982775e-05,
+      "loss": 0.6664,
+      "step": 361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6872581992272673e-05,
+      "loss": 0.7077,
+      "step": 362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68542742233504e-05,
+      "loss": 0.6771,
+      "step": 363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6835923020228714e-05,
+      "loss": 0.6233,
+      "step": 364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6817528499195614e-05,
+      "loss": 0.7673,
+      "step": 365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6799090776813597e-05,
+      "loss": 0.6986,
+      "step": 366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678060996991891e-05,
+      "loss": 0.627,
+      "step": 367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6762086195620824e-05,
+      "loss": 0.7031,
+      "step": 368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674351957130089e-05,
+      "loss": 0.6813,
+      "step": 369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672491021461218e-05,
+      "loss": 0.6627,
+      "step": 370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670625824347856e-05,
+      "loss": 0.7178,
+      "step": 371
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687563776093943e-05,
+      "loss": 0.6907,
+      "step": 372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666882693092151e-05,
+      "loss": 0.6967,
+      "step": 373
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6650047826693006e-05,
+      "loss": 0.7048,
+      "step": 374
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6631226582407954e-05,
+      "loss": 0.7083,
+      "step": 375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6612363317332904e-05,
+      "loss": 0.6811,
+      "step": 376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659345815100069e-05,
+      "loss": 0.7092,
+      "step": 377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6574511203209667e-05,
+      "loss": 0.6646,
+      "step": 378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655552259402295e-05,
+      "loss": 0.6314,
+      "step": 379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6536492443767654e-05,
+      "loss": 0.6779,
+      "step": 380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651742087303412e-05,
+      "loss": 0.701,
+      "step": 381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649830800267519e-05,
+      "loss": 0.6662,
+      "step": 382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647915395380539e-05,
+      "loss": 0.6409,
+      "step": 383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645995884780019e-05,
+      "loss": 0.6548,
+      "step": 384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6440722806295237e-05,
+      "loss": 0.6869,
+      "step": 385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6421445951185577e-05,
+      "loss": 0.6725,
+      "step": 386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640212840462488e-05,
+      "loss": 0.7371,
+      "step": 387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382770289024677e-05,
+      "loss": 0.6615,
+      "step": 388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6363371727053567e-05,
+      "loss": 0.653,
+      "step": 389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343932841636455e-05,
+      "loss": 0.7082,
+      "step": 390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6324453755953772e-05,
+      "loss": 0.6532,
+      "step": 391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6304934593440685e-05,
+      "loss": 0.6479,
+      "step": 392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285375477786322e-05,
+      "loss": 0.7503,
+      "step": 393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6265776532932985e-05,
+      "loss": 0.6287,
+      "step": 394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6246137883075363e-05,
+      "loss": 0.7058,
+      "step": 395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6226459652659752e-05,
+      "loss": 0.7019,
+      "step": 396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6206741966383268e-05,
+      "loss": 0.6444,
+      "step": 397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618698494919304e-05,
+      "loss": 0.7104,
+      "step": 398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6167188726285433e-05,
+      "loss": 0.6838,
+      "step": 399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614735342310525e-05,
+      "loss": 0.6939,
+      "step": 400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6127479165344948e-05,
+      "loss": 0.7191,
+      "step": 401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6107566078943818e-05,
+      "loss": 0.7079,
+      "step": 402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608761429008721e-05,
+      "loss": 0.6884,
+      "step": 403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606762392520571e-05,
+      "loss": 0.6939,
+      "step": 404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6047595110974376e-05,
+      "loss": 0.752,
+      "step": 405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602752797431189e-05,
+      "loss": 0.6645,
+      "step": 406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600742264237979e-05,
+      "loss": 0.695,
+      "step": 407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598727924258164e-05,
+      "loss": 0.6907,
+      "step": 408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5967097902562242e-05,
+      "loss": 0.7086,
+      "step": 409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.594687875020681e-05,
+      "loss": 0.6273,
+      "step": 410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592662191364017e-05,
+      "loss": 0.6622,
+      "step": 411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906327521225947e-05,
+      "loss": 0.6826,
+      "step": 412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5885995701565748e-05,
+      "loss": 0.6566,
+      "step": 413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5865626583498355e-05,
+      "loss": 0.7096,
+      "step": 414
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584522029609889e-05,
+      "loss": 0.6077,
+      "step": 415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5824776968678024e-05,
+      "loss": 0.702,
+      "step": 416
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5804296730781134e-05,
+      "loss": 0.751,
+      "step": 417
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5783779712187496e-05,
+      "loss": 0.7292,
+      "step": 418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5763226042909455e-05,
+      "loss": 0.7162,
+      "step": 419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574263585319161e-05,
+      "loss": 0.6971,
+      "step": 420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5722009273509972e-05,
+      "loss": 0.7252,
+      "step": 421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5701346434571162e-05,
+      "loss": 0.6367,
+      "step": 422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 0.7049,
+      "step": 423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5659912502896483e-05,
+      "loss": 0.6549,
+      "step": 424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563914167271936e-05,
+      "loss": 0.6976,
+      "step": 425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5618335108400893e-05,
+      "loss": 0.6406,
+      "step": 426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5597492941788225e-05,
+      "loss": 0.6826,
+      "step": 427
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5576615304954094e-05,
+      "loss": 0.6795,
+      "step": 428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 0.6854,
+      "step": 429
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5534754150035454e-05,
+      "loss": 0.5909,
+      "step": 430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551377089721692e-05,
+      "loss": 0.5324,
+      "step": 431
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492752704707198e-05,
+      "loss": 0.5304,
+      "step": 432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547169970569449e-05,
+      "loss": 0.5566,
+      "step": 433
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5450612033587526e-05,
+      "loss": 0.599,
+      "step": 434
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542948982201479e-05,
+      "loss": 0.5728,
+      "step": 435
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540833320482361e-05,
+      "loss": 0.6194,
+      "step": 436
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5387142316079345e-05,
+      "loss": 0.5917,
+      "step": 437
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536591729006453e-05,
+      "loss": 0.5592,
+      "step": 438
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5344658261278013e-05,
+      "loss": 0.5705,
+      "step": 439
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5323365364434114e-05,
+      "loss": 0.605,
+      "step": 440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530203873446177e-05,
+      "loss": 0.5901,
+      "step": 441
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528067850650368e-05,
+      "loss": 0.6697,
+      "step": 442
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.525928481591544e-05,
+      "loss": 0.5809,
+      "step": 443
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5237857798264701e-05,
+      "loss": 0.5935,
+      "step": 444
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5216397589330289e-05,
+      "loss": 0.6314,
+      "step": 445
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194904325101374e-05,
+      "loss": 0.6142,
+      "step": 446
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5173378141776569e-05,
+      "loss": 0.5897,
+      "step": 447
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5151819175763112e-05,
+      "loss": 0.5823,
+      "step": 448
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5130227563675957e-05,
+      "loss": 0.5797,
+      "step": 449
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5108603442336949e-05,
+      "loss": 0.5766,
+      "step": 450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5086946948773924e-05,
+      "loss": 0.6463,
+      "step": 451
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5065258220219864e-05,
+      "loss": 0.6017,
+      "step": 452
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5043537394112008e-05,
+      "loss": 0.6247,
+      "step": 453
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5021784608091e-05,
+      "loss": 0.6028,
+      "step": 454
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.5747,
+      "step": 455
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4978183707883828e-05,
+      "loss": 0.6469,
+      "step": 456
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.495633586998807e-05,
+      "loss": 0.5838,
+      "step": 457
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4934456624758216e-05,
+      "loss": 0.6178,
+      "step": 458
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4912546110838775e-05,
+      "loss": 0.6495,
+      "step": 459
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4890604467072406e-05,
+      "loss": 0.6027,
+      "step": 460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4868631832499026e-05,
+      "loss": 0.6109,
+      "step": 461
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4846628346354934e-05,
+      "loss": 0.6282,
+      "step": 462
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4824594148071936e-05,
+      "loss": 0.6119,
+      "step": 463
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4802529377276447e-05,
+      "loss": 0.5984,
+      "step": 464
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4780434173788617e-05,
+      "loss": 0.5983,
+      "step": 465
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4758308677621451e-05,
+      "loss": 0.5657,
+      "step": 466
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4736153028979893e-05,
+      "loss": 0.5983,
+      "step": 467
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713967368259981e-05,
+      "loss": 0.5957,
+      "step": 468
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4691751836047907e-05,
+      "loss": 0.5964,
+      "step": 469
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4669506573119179e-05,
+      "loss": 0.6677,
+      "step": 470
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4647231720437687e-05,
+      "loss": 0.6465,
+      "step": 471
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.462492741915483e-05,
+      "loss": 0.6358,
+      "step": 472
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4602593810608619e-05,
+      "loss": 0.583,
+      "step": 473
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.458023103632277e-05,
+      "loss": 0.6337,
+      "step": 474
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4557839238005831e-05,
+      "loss": 0.6057,
+      "step": 475
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453541855755026e-05,
+      "loss": 0.5934,
+      "step": 476
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4512969137031538e-05,
+      "loss": 0.5838,
+      "step": 477
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4490491118707265e-05,
+      "loss": 0.6365,
+      "step": 478
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4467984645016259e-05,
+      "loss": 0.5854,
+      "step": 479
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444544985857766e-05,
+      "loss": 0.6424,
+      "step": 480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4422886902190014e-05,
+      "loss": 0.6091,
+      "step": 481
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4400295918830378e-05,
+      "loss": 0.6494,
+      "step": 482
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4377677051653404e-05,
+      "loss": 0.6019,
+      "step": 483
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4355030443990452e-05,
+      "loss": 0.6584,
+      "step": 484
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332356239348654e-05,
+      "loss": 0.6543,
+      "step": 485
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4309654581410024e-05,
+      "loss": 0.6072,
+      "step": 486
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4286925614030542e-05,
+      "loss": 0.601,
+      "step": 487
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4264169481239245e-05,
+      "loss": 0.6171,
+      "step": 488
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4241386327237312e-05,
+      "loss": 0.5987,
+      "step": 489
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4218576296397143e-05,
+      "loss": 0.5854,
+      "step": 490
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419573953326146e-05,
+      "loss": 0.6481,
+      "step": 491
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4172876182542372e-05,
+      "loss": 0.6432,
+      "step": 492
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4149986389120485e-05,
+      "loss": 0.5956,
+      "step": 493
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4127070298043949e-05,
+      "loss": 0.6269,
+      "step": 494
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410412805452757e-05,
+      "loss": 0.6016,
+      "step": 495
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.408115980395187e-05,
+      "loss": 0.5826,
+      "step": 496
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4058165691862176e-05,
+      "loss": 0.623,
+      "step": 497
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4035145863967692e-05,
+      "loss": 0.6192,
+      "step": 498
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4012100466140579e-05,
+      "loss": 0.6052,
+      "step": 499
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3989029644415035e-05,
+      "loss": 0.6726,
+      "step": 500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3965933544986351e-05,
+      "loss": 0.6019,
+      "step": 501
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3942812314210015e-05,
+      "loss": 0.5643,
+      "step": 502
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3919666098600753e-05,
+      "loss": 0.6092,
+      "step": 503
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3896495044831622e-05,
+      "loss": 0.6284,
+      "step": 504
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3873299299733068e-05,
+      "loss": 0.645,
+      "step": 505
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3850079010292015e-05,
+      "loss": 0.5992,
+      "step": 506
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.5849,
+      "step": 507
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3803565387106775e-05,
+      "loss": 0.6387,
+      "step": 508
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3780272348110356e-05,
+      "loss": 0.5909,
+      "step": 509
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3756955354265085e-05,
+      "loss": 0.6192,
+      "step": 510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3733614553326211e-05,
+      "loss": 0.6146,
+      "step": 511
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3710250093199835e-05,
+      "loss": 0.6196,
+      "step": 512
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.368686212194199e-05,
+      "loss": 0.6136,
+      "step": 513
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3663450787757685e-05,
+      "loss": 0.5698,
+      "step": 514
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640016238999984e-05,
+      "loss": 0.5568,
+      "step": 515
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361655862416905e-05,
+      "loss": 0.6229,
+      "step": 516
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3593078091911218e-05,
+      "loss": 0.5576,
+      "step": 517
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3569574791018044e-05,
+      "loss": 0.5973,
+      "step": 518
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 0.6043,
+      "step": 519
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3522500479212337e-05,
+      "loss": 0.6735,
+      "step": 520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3498929766600543e-05,
+      "loss": 0.5867,
+      "step": 521
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3475336881952988e-05,
+      "loss": 0.6248,
+      "step": 522
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3451721974773179e-05,
+      "loss": 0.6047,
+      "step": 523
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342808519470418e-05,
+      "loss": 0.5915,
+      "step": 524
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.340442669152766e-05,
+      "loss": 0.5845,
+      "step": 525
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3380746615162943e-05,
+      "loss": 0.6146,
+      "step": 526
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.335704511566605e-05,
+      "loss": 0.6125,
+      "step": 527
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3333322343228763e-05,
+      "loss": 0.6096,
+      "step": 528
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3309578448177674e-05,
+      "loss": 0.5743,
+      "step": 529
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3285813580973207e-05,
+      "loss": 0.5816,
+      "step": 530
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3262027892208696e-05,
+      "loss": 0.6084,
+      "step": 531
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3238221532609412e-05,
+      "loss": 0.6414,
+      "step": 532
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3214394653031616e-05,
+      "loss": 0.6084,
+      "step": 533
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.31905474044616e-05,
+      "loss": 0.5642,
+      "step": 534
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3166679938014728e-05,
+      "loss": 0.6187,
+      "step": 535
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3142792404934477e-05,
+      "loss": 0.5916,
+      "step": 536
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311888495659149e-05,
+      "loss": 0.5934,
+      "step": 537
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309495774448261e-05,
+      "loss": 0.5949,
+      "step": 538
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3071010920229909e-05,
+      "loss": 0.5749,
+      "step": 539
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3047044635579748e-05,
+      "loss": 0.5782,
+      "step": 540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3023059042401797e-05,
+      "loss": 0.6343,
+      "step": 541
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2999054292688084e-05,
+      "loss": 0.6519,
+      "step": 542
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.297503053855203e-05,
+      "loss": 0.5608,
+      "step": 543
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2950987932227482e-05,
+      "loss": 0.5568,
+      "step": 544
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2926926626067745e-05,
+      "loss": 0.6437,
+      "step": 545
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2902846772544625e-05,
+      "loss": 0.5968,
+      "step": 546
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2878748524247462e-05,
+      "loss": 0.5771,
+      "step": 547
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2854632033882155e-05,
+      "loss": 0.606,
+      "step": 548
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2830497454270206e-05,
+      "loss": 0.624,
+      "step": 549
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2806344938347734e-05,
+      "loss": 0.6084,
+      "step": 550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2782174639164528e-05,
+      "loss": 0.5783,
+      "step": 551
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2757986709883059e-05,
+      "loss": 0.5804,
+      "step": 552
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2733781303777524e-05,
+      "loss": 0.6284,
+      "step": 553
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2709558574232852e-05,
+      "loss": 0.6072,
+      "step": 554
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2685318674743769e-05,
+      "loss": 0.575,
+      "step": 555
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2661061758913787e-05,
+      "loss": 0.5946,
+      "step": 556
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263678798045425e-05,
+      "loss": 0.5829,
+      "step": 557
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2612497493183365e-05,
+      "loss": 0.6373,
+      "step": 558
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.6017,
+      "step": 559
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2563867008008773e-05,
+      "loss": 0.6126,
+      "step": 560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2539527318266971e-05,
+      "loss": 0.5987,
+      "step": 561
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.251517153603567e-05,
+      "loss": 0.6088,
+      "step": 562
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2490799815652719e-05,
+      "loss": 0.5663,
+      "step": 563
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2466412311556952e-05,
+      "loss": 0.5867,
+      "step": 564
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.244200917828723e-05,
+      "loss": 0.6184,
+      "step": 565
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2417590570481454e-05,
+      "loss": 0.6088,
+      "step": 566
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2393156642875579e-05,
+      "loss": 0.5844,
+      "step": 567
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2368707550302638e-05,
+      "loss": 0.6123,
+      "step": 568
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2344243447691767e-05,
+      "loss": 0.588,
+      "step": 569
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2319764490067212e-05,
+      "loss": 0.6188,
+      "step": 570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2295270832547358e-05,
+      "loss": 0.5706,
+      "step": 571
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2270762630343734e-05,
+      "loss": 0.5963,
+      "step": 572
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2246240038760042e-05,
+      "loss": 0.6494,
+      "step": 573
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221703213191161e-05,
+      "loss": 0.5995,
+      "step": 574
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2197152309122173e-05,
+      "loss": 0.58,
+      "step": 575
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217258748212737e-05,
+      "loss": 0.6639,
+      "step": 576
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.214800888786927e-05,
+      "loss": 0.6121,
+      "step": 577
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2123416682097635e-05,
+      "loss": 0.6036,
+      "step": 578
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2098811020648475e-05,
+      "loss": 0.6038,
+      "step": 579
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2074192059443077e-05,
+      "loss": 0.5994,
+      "step": 580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204955995448699e-05,
+      "loss": 0.6153,
+      "step": 581
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2024914861869064e-05,
+      "loss": 0.6163,
+      "step": 582
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2000256937760446e-05,
+      "loss": 0.6273,
+      "step": 583
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1975586338413594e-05,
+      "loss": 0.5966,
+      "step": 584
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 0.6354,
+      "step": 585
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1926207739415624e-05,
+      "loss": 0.6356,
+      "step": 586
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1901500052667068e-05,
+      "loss": 0.5992,
+      "step": 587
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1876780316483401e-05,
+      "loss": 0.6105,
+      "step": 588
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852048687508775e-05,
+      "loss": 0.6566,
+      "step": 589
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1827305322462702e-05,
+      "loss": 0.6522,
+      "step": 590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.180255037813906e-05,
+      "loss": 0.6053,
+      "step": 591
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1777784011405104e-05,
+      "loss": 0.678,
+      "step": 592
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1753006379200478e-05,
+      "loss": 0.6909,
+      "step": 593
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1728217638536196e-05,
+      "loss": 0.5869,
+      "step": 594
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1703417946493682e-05,
+      "loss": 0.573,
+      "step": 595
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1678607460223752e-05,
+      "loss": 0.5586,
+      "step": 596
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1653786336945614e-05,
+      "loss": 0.59,
+      "step": 597
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162895473394589e-05,
+      "loss": 0.622,
+      "step": 598
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1604112808577603e-05,
+      "loss": 0.6025,
+      "step": 599
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1579260718259197e-05,
+      "loss": 0.5679,
+      "step": 600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1554398620473516e-05,
+      "loss": 0.5793,
+      "step": 601
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1529526672766838e-05,
+      "loss": 0.5893,
+      "step": 602
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1504645032747832e-05,
+      "loss": 0.5826,
+      "step": 603
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1479753858086608e-05,
+      "loss": 0.6054,
+      "step": 604
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1454853306513687e-05,
+      "loss": 0.6011,
+      "step": 605
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1429943535819005e-05,
+      "loss": 0.5916,
+      "step": 606
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1405024703850929e-05,
+      "loss": 0.6192,
+      "step": 607
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1380096968515233e-05,
+      "loss": 0.5894,
+      "step": 608
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1355160487774119e-05,
+      "loss": 0.5864,
+      "step": 609
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1330215419645202e-05,
+      "loss": 0.6241,
+      "step": 610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130526192220052e-05,
+      "loss": 0.5776,
+      "step": 611
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.128030015356551e-05,
+      "loss": 0.5451,
+      "step": 612
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1255330271918048e-05,
+      "loss": 0.6144,
+      "step": 613
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1230352435487393e-05,
+      "loss": 0.6065,
+      "step": 614
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 0.6133,
+      "step": 615
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1180373531444648e-05,
+      "loss": 0.6113,
+      "step": 616
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1155372780539125e-05,
+      "loss": 0.6083,
+      "step": 617
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1130364708261552e-05,
+      "loss": 0.6032,
+      "step": 618
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11053494730832e-05,
+      "loss": 0.6884,
+      "step": 619
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1080327233520744e-05,
+      "loss": 0.6117,
+      "step": 620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1055298148135236e-05,
+      "loss": 0.6232,
+      "step": 621
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.103026237553111e-05,
+      "loss": 0.5988,
+      "step": 622
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1005220074355172e-05,
+      "loss": 0.5933,
+      "step": 623
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.098017140329561e-05,
+      "loss": 0.6088,
+      "step": 624
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0955116521080959e-05,
+      "loss": 0.6253,
+      "step": 625
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0930055586479132e-05,
+      "loss": 0.5863,
+      "step": 626
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090498875829638e-05,
+      "loss": 0.5868,
+      "step": 627
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0879916195376307e-05,
+      "loss": 0.5791,
+      "step": 628
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.085483805659886e-05,
+      "loss": 0.5852,
+      "step": 629
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0829754500879308e-05,
+      "loss": 0.5868,
+      "step": 630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0804665687167262e-05,
+      "loss": 0.5748,
+      "step": 631
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0779571774445634e-05,
+      "loss": 0.6086,
+      "step": 632
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0754472921729661e-05,
+      "loss": 0.6666,
+      "step": 633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0729369288065878e-05,
+      "loss": 0.635,
+      "step": 634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070426103253112e-05,
+      "loss": 0.613,
+      "step": 635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0679148314231504e-05,
+      "loss": 0.6311,
+      "step": 636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0654031292301432e-05,
+      "loss": 0.6307,
+      "step": 637
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0628910125902574e-05,
+      "loss": 0.6386,
+      "step": 638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0603784974222862e-05,
+      "loss": 0.5951,
+      "step": 639
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0578655996475486e-05,
+      "loss": 0.5882,
+      "step": 640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0553523351897879e-05,
+      "loss": 0.606,
+      "step": 641
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0528387199750706e-05,
+      "loss": 0.5884,
+      "step": 642
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0503247699316867e-05,
+      "loss": 0.6499,
+      "step": 643
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0478105009900473e-05,
+      "loss": 0.6043,
+      "step": 644
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0452959290825846e-05,
+      "loss": 0.5872,
+      "step": 645
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0427810701436502e-05,
+      "loss": 0.619,
+      "step": 646
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0402659401094154e-05,
+      "loss": 0.6257,
+      "step": 647
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0377505549177683e-05,
+      "loss": 0.6082,
+      "step": 648
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0352349305082148e-05,
+      "loss": 0.5901,
+      "step": 649
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0327190828217763e-05,
+      "loss": 0.6132,
+      "step": 650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030203027800889e-05,
+      "loss": 0.6139,
+      "step": 651
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.027686781389303e-05,
+      "loss": 0.6135,
+      "step": 652
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0251703595319816e-05,
+      "loss": 0.5351,
+      "step": 653
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0226537781749988e-05,
+      "loss": 0.5973,
+      "step": 654
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0201370532654404e-05,
+      "loss": 0.5795,
+      "step": 655
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0176202007513016e-05,
+      "loss": 0.5887,
+      "step": 656
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015103236581386e-05,
+      "loss": 0.5916,
+      "step": 657
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.012586176705205e-05,
+      "loss": 0.5375,
+      "step": 658
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0100690370728756e-05,
+      "loss": 0.6302,
+      "step": 659
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0075518336350218e-05,
+      "loss": 0.5781,
+      "step": 660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050345823426704e-05,
+      "loss": 0.5958,
+      "step": 661
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0025172991471523e-05,
+      "loss": 0.6014,
+      "step": 662
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.6239,
+      "step": 663
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.97482700852848e-06,
+      "loss": 0.5937,
+      "step": 664
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9496541765733e-06,
+      "loss": 0.587,
+      "step": 665
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.924481663649785e-06,
+      "loss": 0.569,
+      "step": 666
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.899309629271246e-06,
+      "loss": 0.5875,
+      "step": 667
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.874138232947953e-06,
+      "loss": 0.6128,
+      "step": 668
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.848967634186142e-06,
+      "loss": 0.595,
+      "step": 669
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.823797992486986e-06,
+      "loss": 0.5747,
+      "step": 670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.7986294673456e-06,
+      "loss": 0.6317,
+      "step": 671
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.773462218250014e-06,
+      "loss": 0.5908,
+      "step": 672
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.748296404680189e-06,
+      "loss": 0.6548,
+      "step": 673
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.723132186106973e-06,
+      "loss": 0.6308,
+      "step": 674
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697969721991114e-06,
+      "loss": 0.5918,
+      "step": 675
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.67280917178224e-06,
+      "loss": 0.5854,
+      "step": 676
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.647650694917857e-06,
+      "loss": 0.6282,
+      "step": 677
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62249445082232e-06,
+      "loss": 0.6365,
+      "step": 678
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597340598905851e-06,
+      "loss": 0.6052,
+      "step": 679
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.572189298563501e-06,
+      "loss": 0.5366,
+      "step": 680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.547040709174159e-06,
+      "loss": 0.6306,
+      "step": 681
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52189499009953e-06,
+      "loss": 0.5875,
+      "step": 682
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.496752300683133e-06,
+      "loss": 0.5811,
+      "step": 683
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.471612800249295e-06,
+      "loss": 0.5936,
+      "step": 684
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.446476648102123e-06,
+      "loss": 0.6115,
+      "step": 685
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.421344003524517e-06,
+      "loss": 0.5799,
+      "step": 686
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.39621502577714e-06,
+      "loss": 0.6122,
+      "step": 687
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.37108987409743e-06,
+      "loss": 0.5786,
+      "step": 688
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.34596870769857e-06,
+      "loss": 0.5509,
+      "step": 689
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.320851685768498e-06,
+      "loss": 0.5976,
+      "step": 690
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295738967468881e-06,
+      "loss": 0.6013,
+      "step": 691
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.270630711934124e-06,
+      "loss": 0.674,
+      "step": 692
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.24552707827034e-06,
+      "loss": 0.5796,
+      "step": 693
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22042822555437e-06,
+      "loss": 0.6119,
+      "step": 694
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.195334312832742e-06,
+      "loss": 0.6009,
+      "step": 695
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170245499120694e-06,
+      "loss": 0.6007,
+      "step": 696
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.145161943401143e-06,
+      "loss": 0.601,
+      "step": 697
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.120083804623694e-06,
+      "loss": 0.5624,
+      "step": 698
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.095011241703623e-06,
+      "loss": 0.6645,
+      "step": 699
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.069944413520873e-06,
+      "loss": 0.561,
+      "step": 700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.044883478919043e-06,
+      "loss": 0.6389,
+      "step": 701
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.019828596704394e-06,
+      "loss": 0.5974,
+      "step": 702
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.994779925644832e-06,
+      "loss": 0.6121,
+      "step": 703
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.969737624468894e-06,
+      "loss": 0.5613,
+      "step": 704
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.944701851864767e-06,
+      "loss": 0.5967,
+      "step": 705
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.919672766479257e-06,
+      "loss": 0.6217,
+      "step": 706
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.894650526916802e-06,
+      "loss": 0.5319,
+      "step": 707
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.869635291738452e-06,
+      "loss": 0.588,
+      "step": 708
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.844627219460879e-06,
+      "loss": 0.5376,
+      "step": 709
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.819626468555356e-06,
+      "loss": 0.5743,
+      "step": 710
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 0.6467,
+      "step": 711
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76964756451261e-06,
+      "loss": 0.5932,
+      "step": 712
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.744669728081959e-06,
+      "loss": 0.6343,
+      "step": 713
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.719699846434493e-06,
+      "loss": 0.5945,
+      "step": 714
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.694738077799487e-06,
+      "loss": 0.6006,
+      "step": 715
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.669784580354801e-06,
+      "loss": 0.5621,
+      "step": 716
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.644839512225886e-06,
+      "loss": 0.5545,
+      "step": 717
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61990303148477e-06,
+      "loss": 0.5601,
+      "step": 718
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.594975296149076e-06,
+      "loss": 0.5791,
+      "step": 719
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.570056464180998e-06,
+      "loss": 0.5561,
+      "step": 720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54514669348632e-06,
+      "loss": 0.6331,
+      "step": 721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.520246141913393e-06,
+      "loss": 0.6097,
+      "step": 722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.49535496725217e-06,
+      "loss": 0.6355,
+      "step": 723
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.470473327233167e-06,
+      "loss": 0.65,
+      "step": 724
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.445601379526482e-06,
+      "loss": 0.5875,
+      "step": 725
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.420739281740806e-06,
+      "loss": 0.5474,
+      "step": 726
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.395887191422397e-06,
+      "loss": 0.6169,
+      "step": 727
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.371045266054114e-06,
+      "loss": 0.6755,
+      "step": 728
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.346213663054388e-06,
+      "loss": 0.627,
+      "step": 729
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.321392539776251e-06,
+      "loss": 0.6129,
+      "step": 730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.296582053506318e-06,
+      "loss": 0.5795,
+      "step": 731
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.271782361463806e-06,
+      "loss": 0.5766,
+      "step": 732
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.246993620799526e-06,
+      "loss": 0.604,
+      "step": 733
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.222215988594899e-06,
+      "loss": 0.6171,
+      "step": 734
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.197449621860944e-06,
+      "loss": 0.5751,
+      "step": 735
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.172694677537303e-06,
+      "loss": 0.5633,
+      "step": 736
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147951312491227e-06,
+      "loss": 0.6157,
+      "step": 737
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.123219683516604e-06,
+      "loss": 0.6559,
+      "step": 738
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.098499947332935e-06,
+      "loss": 0.5719,
+      "step": 739
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.073792260584377e-06,
+      "loss": 0.6274,
+      "step": 740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04909677983872e-06,
+      "loss": 0.5692,
+      "step": 741
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.02441366158641e-06,
+      "loss": 0.5767,
+      "step": 742
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.999743062239557e-06,
+      "loss": 0.5698,
+      "step": 743
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.975085138130938e-06,
+      "loss": 0.6432,
+      "step": 744
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.950440045513014e-06,
+      "loss": 0.574,
+      "step": 745
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.925807940556927e-06,
+      "loss": 0.5924,
+      "step": 746
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.901188979351527e-06,
+      "loss": 0.5194,
+      "step": 747
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.876583317902368e-06,
+      "loss": 0.5652,
+      "step": 748
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.851991112130734e-06,
+      "loss": 0.5605,
+      "step": 749
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.827412517872634e-06,
+      "loss": 0.6194,
+      "step": 750
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.802847690877832e-06,
+      "loss": 0.609,
+      "step": 751
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.778296786808842e-06,
+      "loss": 0.5565,
+      "step": 752
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.753759961239965e-06,
+      "loss": 0.5755,
+      "step": 753
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72923736965627e-06,
+      "loss": 0.5936,
+      "step": 754
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.704729167452647e-06,
+      "loss": 0.5775,
+      "step": 755
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680235509932791e-06,
+      "loss": 0.6103,
+      "step": 756
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.655756552308238e-06,
+      "loss": 0.6325,
+      "step": 757
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.631292449697365e-06,
+      "loss": 0.6272,
+      "step": 758
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.606843357124426e-06,
+      "loss": 0.5829,
+      "step": 759
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.582409429518548e-06,
+      "loss": 0.6285,
+      "step": 760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.557990821712769e-06,
+      "loss": 0.596,
+      "step": 761
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5335876884430495e-06,
+      "loss": 0.5685,
+      "step": 762
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5092001843472825e-06,
+      "loss": 0.6278,
+      "step": 763
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484828463964331e-06,
+      "loss": 0.5941,
+      "step": 764
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.460472681733031e-06,
+      "loss": 0.6203,
+      "step": 765
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.43613299199123e-06,
+      "loss": 0.5668,
+      "step": 766
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.6239,
+      "step": 767
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.387502506816638e-06,
+      "loss": 0.5554,
+      "step": 768
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3632120195457515e-06,
+      "loss": 0.6296,
+      "step": 769
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.338938241086217e-06,
+      "loss": 0.6107,
+      "step": 770
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.314681325256232e-06,
+      "loss": 0.6263,
+      "step": 771
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.29044142576715e-06,
+      "loss": 0.5571,
+      "step": 772
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.26621869622248e-06,
+      "loss": 0.6224,
+      "step": 773
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.242013290116944e-06,
+      "loss": 0.5864,
+      "step": 774
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.217825360835475e-06,
+      "loss": 0.5982,
+      "step": 775
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.193655061652269e-06,
+      "loss": 0.6405,
+      "step": 776
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.169502545729798e-06,
+      "loss": 0.602,
+      "step": 777
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.145367966117846e-06,
+      "loss": 0.6393,
+      "step": 778
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.12125147575254e-06,
+      "loss": 0.6259,
+      "step": 779
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.097153227455379e-06,
+      "loss": 0.6047,
+      "step": 780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.07307337393226e-06,
+      "loss": 0.595,
+      "step": 781
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049012067772521e-06,
+      "loss": 0.5885,
+      "step": 782
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.024969461447973e-06,
+      "loss": 0.6052,
+      "step": 783
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0009457073119166e-06,
+      "loss": 0.6139,
+      "step": 784
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.976940957598206e-06,
+      "loss": 0.6271,
+      "step": 785
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.952955364420255e-06,
+      "loss": 0.63,
+      "step": 786
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.928989079770094e-06,
+      "loss": 0.6221,
+      "step": 787
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.905042255517393e-06,
+      "loss": 0.6069,
+      "step": 788
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.881115043408512e-06,
+      "loss": 0.6313,
+      "step": 789
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8572075950655255e-06,
+      "loss": 0.583,
+      "step": 790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.833320061985278e-06,
+      "loss": 0.6004,
+      "step": 791
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.809452595538403e-06,
+      "loss": 0.619,
+      "step": 792
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.785605346968387e-06,
+      "loss": 0.5648,
+      "step": 793
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.761778467390592e-06,
+      "loss": 0.6056,
+      "step": 794
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7379721077913095e-06,
+      "loss": 0.6451,
+      "step": 795
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.714186419026798e-06,
+      "loss": 0.5826,
+      "step": 796
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.690421551822333e-06,
+      "loss": 0.5715,
+      "step": 797
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.666677656771239e-06,
+      "loss": 0.5941,
+      "step": 798
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6429548843339554e-06,
+      "loss": 0.6039,
+      "step": 799
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.619253384837059e-06,
+      "loss": 0.5799,
+      "step": 800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.595573308472338e-06,
+      "loss": 0.6366,
+      "step": 801
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.57191480529582e-06,
+      "loss": 0.5695,
+      "step": 802
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.548278025226822e-06,
+      "loss": 0.6168,
+      "step": 803
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.524663118047015e-06,
+      "loss": 0.6392,
+      "step": 804
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.501070233399458e-06,
+      "loss": 0.6004,
+      "step": 805
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4774995207876654e-06,
+      "loss": 0.6051,
+      "step": 806
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 0.5985,
+      "step": 807
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4304252089819605e-06,
+      "loss": 0.5747,
+      "step": 808
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.406921908088783e-06,
+      "loss": 0.5517,
+      "step": 809
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3834413758309525e-06,
+      "loss": 0.5953,
+      "step": 810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3599837610000194e-06,
+      "loss": 0.5978,
+      "step": 811
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.336549212242318e-06,
+      "loss": 0.5733,
+      "step": 812
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3131378780580134e-06,
+      "loss": 0.5442,
+      "step": 813
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.289749906800166e-06,
+      "loss": 0.5958,
+      "step": 814
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.266385446673791e-06,
+      "loss": 0.5907,
+      "step": 815
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.243044645734917e-06,
+      "loss": 0.6512,
+      "step": 816
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.219727651889647e-06,
+      "loss": 0.5674,
+      "step": 817
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.196434612893229e-06,
+      "loss": 0.5565,
+      "step": 818
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.6233,
+      "step": 819
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.149920989707989e-06,
+      "loss": 0.5866,
+      "step": 820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1267007002669345e-06,
+      "loss": 0.5471,
+      "step": 821
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.103504955168382e-06,
+      "loss": 0.6304,
+      "step": 822
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.080333901399252e-06,
+      "loss": 0.613,
+      "step": 823
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.057187685789987e-06,
+      "loss": 0.5563,
+      "step": 824
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0340664550136494e-06,
+      "loss": 0.5999,
+      "step": 825
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.010970355584968e-06,
+      "loss": 0.5908,
+      "step": 826
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.987899533859422e-06,
+      "loss": 0.6025,
+      "step": 827
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.96485413603231e-06,
+      "loss": 0.6176,
+      "step": 828
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.941834308137828e-06,
+      "loss": 0.5807,
+      "step": 829
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.918840196048132e-06,
+      "loss": 0.6489,
+      "step": 830
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.895871945472434e-06,
+      "loss": 0.6051,
+      "step": 831
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.872929701956054e-06,
+      "loss": 0.5922,
+      "step": 832
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.85001361087952e-06,
+      "loss": 0.6391,
+      "step": 833
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.827123817457631e-06,
+      "loss": 0.682,
+      "step": 834
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8042604667385475e-06,
+      "loss": 0.5695,
+      "step": 835
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.781423703602863e-06,
+      "loss": 0.6103,
+      "step": 836
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.75861367276269e-06,
+      "loss": 0.6449,
+      "step": 837
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.735830518760757e-06,
+      "loss": 0.6074,
+      "step": 838
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.713074385969457e-06,
+      "loss": 0.5809,
+      "step": 839
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.690345418589978e-06,
+      "loss": 0.5481,
+      "step": 840
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.667643760651349e-06,
+      "loss": 0.5846,
+      "step": 841
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.644969556009547e-06,
+      "loss": 0.588,
+      "step": 842
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.622322948346595e-06,
+      "loss": 0.5981,
+      "step": 843
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.599704081169626e-06,
+      "loss": 0.556,
+      "step": 844
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5771130978099896e-06,
+      "loss": 0.5936,
+      "step": 845
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.55455014142234e-06,
+      "loss": 0.6398,
+      "step": 846
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.532015354983742e-06,
+      "loss": 0.6353,
+      "step": 847
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.50950888129274e-06,
+      "loss": 0.6306,
+      "step": 848
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4870308629684675e-06,
+      "loss": 0.6219,
+      "step": 849
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.464581442449741e-06,
+      "loss": 0.643,
+      "step": 850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.442160761994171e-06,
+      "loss": 0.5953,
+      "step": 851
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.419768963677233e-06,
+      "loss": 0.6063,
+      "step": 852
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.397406189391387e-06,
+      "loss": 0.5566,
+      "step": 853
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.375072580845171e-06,
+      "loss": 0.5964,
+      "step": 854
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.352768279562315e-06,
+      "loss": 0.5946,
+      "step": 855
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.330493426880825e-06,
+      "loss": 0.5622,
+      "step": 856
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.308248163952097e-06,
+      "loss": 0.6286,
+      "step": 857
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286032631740023e-06,
+      "loss": 0.58,
+      "step": 858
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.263846971020108e-06,
+      "loss": 0.6134,
+      "step": 859
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.241691322378554e-06,
+      "loss": 0.4445,
+      "step": 860
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.219565826211382e-06,
+      "loss": 0.487,
+      "step": 861
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.197470622723555e-06,
+      "loss": 0.4901,
+      "step": 862
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.175405851928068e-06,
+      "loss": 0.4907,
+      "step": 863
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.15337165364507e-06,
+      "loss": 0.5451,
+      "step": 864
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.131368167500977e-06,
+      "loss": 0.5117,
+      "step": 865
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.109395532927597e-06,
+      "loss": 0.5078,
+      "step": 866
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.087453889161229e-06,
+      "loss": 0.5252,
+      "step": 867
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.06554337524179e-06,
+      "loss": 0.4818,
+      "step": 868
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.043664130011932e-06,
+      "loss": 0.534,
+      "step": 869
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.021816292116175e-06,
+      "loss": 0.5066,
+      "step": 870
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.5677,
+      "step": 871
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.978215391909006e-06,
+      "loss": 0.5421,
+      "step": 872
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.956462605887994e-06,
+      "loss": 0.518,
+      "step": 873
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.93474177978014e-06,
+      "loss": 0.5264,
+      "step": 874
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.913053051226079e-06,
+      "loss": 0.5488,
+      "step": 875
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.891396557663056e-06,
+      "loss": 0.5089,
+      "step": 876
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.869772436324045e-06,
+      "loss": 0.5344,
+      "step": 877
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.84818082423689e-06,
+      "loss": 0.4785,
+      "step": 878
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.826621858223431e-06,
+      "loss": 0.5041,
+      "step": 879
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.80509567489863e-06,
+      "loss": 0.5255,
+      "step": 880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.783602410669712e-06,
+      "loss": 0.555,
+      "step": 881
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.762142201735299e-06,
+      "loss": 0.5502,
+      "step": 882
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.740715184084559e-06,
+      "loss": 0.5164,
+      "step": 883
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7193214934963204e-06,
+      "loss": 0.5448,
+      "step": 884
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697961265538231e-06,
+      "loss": 0.5195,
+      "step": 885
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6766346355658856e-06,
+      "loss": 0.5352,
+      "step": 886
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655341738721989e-06,
+      "loss": 0.5274,
+      "step": 887
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6340827099354734e-06,
+      "loss": 0.5492,
+      "step": 888
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.612857683920658e-06,
+      "loss": 0.5505,
+      "step": 889
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5916667951763924e-06,
+      "loss": 0.5151,
+      "step": 890
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.570510177985213e-06,
+      "loss": 0.5665,
+      "step": 891
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.549387966412478e-06,
+      "loss": 0.5246,
+      "step": 892
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.528300294305519e-06,
+      "loss": 0.5372,
+      "step": 893
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5072472952928015e-06,
+      "loss": 0.5232,
+      "step": 894
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.486229102783084e-06,
+      "loss": 0.4925,
+      "step": 895
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.465245849964549e-06,
+      "loss": 0.5257,
+      "step": 896
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.444297669803981e-06,
+      "loss": 0.4918,
+      "step": 897
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.423384695045907e-06,
+      "loss": 0.5386,
+      "step": 898
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.40250705821178e-06,
+      "loss": 0.5298,
+      "step": 899
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.381664891599111e-06,
+      "loss": 0.6077,
+      "step": 900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.360858327280641e-06,
+      "loss": 0.56,
+      "step": 901
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.340087497103519e-06,
+      "loss": 0.5327,
+      "step": 902
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 0.5317,
+      "step": 903
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.298653565428841e-06,
+      "loss": 0.5274,
+      "step": 904
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.277990726490029e-06,
+      "loss": 0.5385,
+      "step": 905
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.257364146808394e-06,
+      "loss": 0.4947,
+      "step": 906
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.236773957090549e-06,
+      "loss": 0.5215,
+      "step": 907
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2162202878125096e-06,
+      "loss": 0.5532,
+      "step": 908
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1957032692188685e-06,
+      "loss": 0.5275,
+      "step": 909
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.17522303132198e-06,
+      "loss": 0.5356,
+      "step": 910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.154779703901114e-06,
+      "loss": 0.5623,
+      "step": 911
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.134373416501652e-06,
+      "loss": 0.5231,
+      "step": 912
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1140042984342535e-06,
+      "loss": 0.5465,
+      "step": 913
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.093672478774057e-06,
+      "loss": 0.5816,
+      "step": 914
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.073378086359834e-06,
+      "loss": 0.5451,
+      "step": 915
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.053121249793196e-06,
+      "loss": 0.5197,
+      "step": 916
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.03290209743776e-06,
+      "loss": 0.5299,
+      "step": 917
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0127207574183576e-06,
+      "loss": 0.5236,
+      "step": 918
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99257735762021e-06,
+      "loss": 0.5299,
+      "step": 919
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.97247202568811e-06,
+      "loss": 0.5324,
+      "step": 920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.952404889025626e-06,
+      "loss": 0.5556,
+      "step": 921
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932376074794288e-06,
+      "loss": 0.5376,
+      "step": 922
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.912385709912794e-06,
+      "loss": 0.5153,
+      "step": 923
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.892433921056184e-06,
+      "loss": 0.5493,
+      "step": 924
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8725208346550535e-06,
+      "loss": 0.5224,
+      "step": 925
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.852646576894749e-06,
+      "loss": 0.5098,
+      "step": 926
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.832811273714569e-06,
+      "loss": 0.5721,
+      "step": 927
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.813015050806964e-06,
+      "loss": 0.5172,
+      "step": 928
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7932580336167356e-06,
+      "loss": 0.543,
+      "step": 929
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7735403473402475e-06,
+      "loss": 0.5789,
+      "step": 930
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.75386211692464e-06,
+      "loss": 0.4872,
+      "step": 931
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7342234670670196e-06,
+      "loss": 0.5147,
+      "step": 932
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.714624522213681e-06,
+      "loss": 0.5391,
+      "step": 933
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.695065406559315e-06,
+      "loss": 0.5496,
+      "step": 934
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6755462440462288e-06,
+      "loss": 0.5518,
+      "step": 935
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6560671583635467e-06,
+      "loss": 0.483,
+      "step": 936
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.636628272946434e-06,
+      "loss": 0.5466,
+      "step": 937
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6172297109753253e-06,
+      "loss": 0.5291,
+      "step": 938
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5978715953751207e-06,
+      "loss": 0.5239,
+      "step": 939
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5785540488144265e-06,
+      "loss": 0.5207,
+      "step": 940
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5592771937047633e-06,
+      "loss": 0.5671,
+      "step": 941
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5400411521998123e-06,
+      "loss": 0.5146,
+      "step": 942
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5208460461946136e-06,
+      "loss": 0.5273,
+      "step": 943
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.501691997324813e-06,
+      "loss": 0.5048,
+      "step": 944
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4825791269658782e-06,
+      "loss": 0.509,
+      "step": 945
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4635075562323516e-06,
+      "loss": 0.5316,
+      "step": 946
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4444774059770537e-06,
+      "loss": 0.4805,
+      "step": 947
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4254887967903373e-06,
+      "loss": 0.5112,
+      "step": 948
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4065418489993118e-06,
+      "loss": 0.5422,
+      "step": 949
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3876366826670993e-06,
+      "loss": 0.5588,
+      "step": 950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3687734175920505e-06,
+      "loss": 0.5253,
+      "step": 951
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3499521733069962e-06,
+      "loss": 0.5337,
+      "step": 952
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3311730690784916e-06,
+      "loss": 0.5209,
+      "step": 953
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3124362239060627e-06,
+      "loss": 0.5151,
+      "step": 954
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2937417565214426e-06,
+      "loss": 0.5119,
+      "step": 955
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2750897853878215e-06,
+      "loss": 0.574,
+      "step": 956
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2564804286991137e-06,
+      "loss": 0.5086,
+      "step": 957
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2379138043791748e-06,
+      "loss": 0.5353,
+      "step": 958
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2193900300810908e-06,
+      "loss": 0.467,
+      "step": 959
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2009092231864047e-06,
+      "loss": 0.5257,
+      "step": 960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1824715008043874e-06,
+      "loss": 0.5688,
+      "step": 961
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1640769797712865e-06,
+      "loss": 0.5211,
+      "step": 962
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.145725776649602e-06,
+      "loss": 0.5151,
+      "step": 963
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1274180077273287e-06,
+      "loss": 0.4875,
+      "step": 964
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1091537890172263e-06,
+      "loss": 0.5437,
+      "step": 965
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0909332362560875e-06,
+      "loss": 0.4882,
+      "step": 966
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0727564649040066e-06,
+      "loss": 0.5668,
+      "step": 967
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0546235901436383e-06,
+      "loss": 0.4976,
+      "step": 968
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.036534726879473e-06,
+      "loss": 0.4815,
+      "step": 969
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.01848998973711e-06,
+      "loss": 0.5437,
+      "step": 970
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.000489493062535e-06,
+      "loss": 0.5355,
+      "step": 971
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.982533350921383e-06,
+      "loss": 0.5673,
+      "step": 972
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.964621677098225e-06,
+      "loss": 0.4698,
+      "step": 973
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9467545850958445e-06,
+      "loss": 0.5215,
+      "step": 974
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.5175,
+      "step": 975
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9111545991513202e-06,
+      "loss": 0.5298,
+      "step": 976
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8934219307993393e-06,
+      "loss": 0.5101,
+      "step": 977
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8757342954470537e-06,
+      "loss": 0.5413,
+      "step": 978
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8580918051775542e-06,
+      "loss": 0.5443,
+      "step": 979
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.840494571787864e-06,
+      "loss": 0.5075,
+      "step": 980
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8229427067882165e-06,
+      "loss": 0.5034,
+      "step": 981
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8054363214013647e-06,
+      "loss": 0.5186,
+      "step": 982
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7879755265618558e-06,
+      "loss": 0.5471,
+      "step": 983
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7705604329153434e-06,
+      "loss": 0.4909,
+      "step": 984
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7531911508178764e-06,
+      "loss": 0.5202,
+      "step": 985
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7358677903352173e-06,
+      "loss": 0.5045,
+      "step": 986
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7185904612421177e-06,
+      "loss": 0.568,
+      "step": 987
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7013592730216464e-06,
+      "loss": 0.5011,
+      "step": 988
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6841743348644788e-06,
+      "loss": 0.5333,
+      "step": 989
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6670357556682245e-06,
+      "loss": 0.5383,
+      "step": 990
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6499436440367165e-06,
+      "loss": 0.5226,
+      "step": 991
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6328981082793313e-06,
+      "loss": 0.5272,
+      "step": 992
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.615899256410306e-06,
+      "loss": 0.4873,
+      "step": 993
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.598947196148053e-06,
+      "loss": 0.5314,
+      "step": 994
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5820420349144693e-06,
+      "loss": 0.5098,
+      "step": 995
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.565183879834272e-06,
+      "loss": 0.524,
+      "step": 996
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.548372837734299e-06,
+      "loss": 0.5327,
+      "step": 997
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.531609015142844e-06,
+      "loss": 0.56,
+      "step": 998
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 0.5382,
+      "step": 999
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.498223453101911e-06,
+      "loss": 0.5109,
+      "step": 1000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4816019252102274e-06,
+      "loss": 0.4958,
+      "step": 1001
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.465028039941316e-06,
+      "loss": 0.5627,
+      "step": 1002
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4485019023206637e-06,
+      "loss": 0.5343,
+      "step": 1003
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4320236170711785e-06,
+      "loss": 0.5031,
+      "step": 1004
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.415593288612541e-06,
+      "loss": 0.5579,
+      "step": 1005
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3992110210605378e-06,
+      "loss": 0.5601,
+      "step": 1006
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.382876918226409e-06,
+      "loss": 0.5039,
+      "step": 1007
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.366591083616178e-06,
+      "loss": 0.5444,
+      "step": 1008
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3503536204300017e-06,
+      "loss": 0.5011,
+      "step": 1009
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.334164631561516e-06,
+      "loss": 0.5428,
+      "step": 1010
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.318024219597196e-06,
+      "loss": 0.53,
+      "step": 1011
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3019324868156777e-06,
+      "loss": 0.5698,
+      "step": 1012
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.285889535187139e-06,
+      "loss": 0.5188,
+      "step": 1013
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.26989546637263e-06,
+      "loss": 0.5344,
+      "step": 1014
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2539503817234553e-06,
+      "loss": 0.5446,
+      "step": 1015
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2380543822805023e-06,
+      "loss": 0.5432,
+      "step": 1016
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.222207568773619e-06,
+      "loss": 0.5523,
+      "step": 1017
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2064100416209777e-06,
+      "loss": 0.546,
+      "step": 1018
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.190661900928426e-06,
+      "loss": 0.5414,
+      "step": 1019
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1749632464888594e-06,
+      "loss": 0.5612,
+      "step": 1020
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1593141777815906e-06,
+      "loss": 0.5683,
+      "step": 1021
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1437147939717197e-06,
+      "loss": 0.5879,
+      "step": 1022
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1281651939094996e-06,
+      "loss": 0.5669,
+      "step": 1023
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.112665476129714e-06,
+      "loss": 0.495,
+      "step": 1024
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0972157388510484e-06,
+      "loss": 0.5303,
+      "step": 1025
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0818160799754826e-06,
+      "loss": 0.4482,
+      "step": 1026
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0664665970876496e-06,
+      "loss": 0.571,
+      "step": 1027
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0511673874542302e-06,
+      "loss": 0.5204,
+      "step": 1028
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.03591854802333e-06,
+      "loss": 0.5217,
+      "step": 1029
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.020720175423876e-06,
+      "loss": 0.5163,
+      "step": 1030
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0055723659649907e-06,
+      "loss": 0.4873,
+      "step": 1031
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.990475215635388e-06,
+      "loss": 0.494,
+      "step": 1032
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.975428820102766e-06,
+      "loss": 0.5291,
+      "step": 1033
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9604332747131993e-06,
+      "loss": 0.5433,
+      "step": 1034
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.94548867449054e-06,
+      "loss": 0.5057,
+      "step": 1035
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9305951141358027e-06,
+      "loss": 0.5485,
+      "step": 1036
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9157526880265763e-06,
+      "loss": 0.5347,
+      "step": 1037
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9009614902164175e-06,
+      "loss": 0.4946,
+      "step": 1038
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8862216144342692e-06,
+      "loss": 0.5363,
+      "step": 1039
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8715331540838488e-06,
+      "loss": 0.5287,
+      "step": 1040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8568962022430637e-06,
+      "loss": 0.4871,
+      "step": 1041
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8423108516634214e-06,
+      "loss": 0.5014,
+      "step": 1042
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8277771947694523e-06,
+      "loss": 0.5344,
+      "step": 1043
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.813295323658103e-06,
+      "loss": 0.5432,
+      "step": 1044
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7988653300981663e-06,
+      "loss": 0.5338,
+      "step": 1045
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7844873055296963e-06,
+      "loss": 0.5297,
+      "step": 1046
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7701613410634367e-06,
+      "loss": 0.5296,
+      "step": 1047
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7558875274802278e-06,
+      "loss": 0.5851,
+      "step": 1048
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7416659552304449e-06,
+      "loss": 0.5531,
+      "step": 1049
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7274967144334155e-06,
+      "loss": 0.5526,
+      "step": 1050
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7133798948768598e-06,
+      "loss": 0.534,
+      "step": 1051
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6993155860163102e-06,
+      "loss": 0.511,
+      "step": 1052
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6853038769745466e-06,
+      "loss": 0.5456,
+      "step": 1053
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6713448565410428e-06,
+      "loss": 0.5174,
+      "step": 1054
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6574386131713872e-06,
+      "loss": 0.5368,
+      "step": 1055
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.643585234986733e-06,
+      "loss": 0.524,
+      "step": 1056
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6297848097732338e-06,
+      "loss": 0.4975,
+      "step": 1057
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.616037424981498e-06,
+      "loss": 0.5085,
+      "step": 1058
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6023431677260215e-06,
+      "loss": 0.5203,
+      "step": 1059
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5887021247846424e-06,
+      "loss": 0.4904,
+      "step": 1060
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5751143825979875e-06,
+      "loss": 0.5573,
+      "step": 1061
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5615800272689352e-06,
+      "loss": 0.5114,
+      "step": 1062
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5480991445620541e-06,
+      "loss": 0.589,
+      "step": 1063
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5346718199030687e-06,
+      "loss": 0.5433,
+      "step": 1064
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5212981383783154e-06,
+      "loss": 0.5849,
+      "step": 1065
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5079781847342122e-06,
+      "loss": 0.5115,
+      "step": 1066
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4947120433767048e-06,
+      "loss": 0.5738,
+      "step": 1067
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4814997983707457e-06,
+      "loss": 0.533,
+      "step": 1068
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4683415334397532e-06,
+      "loss": 0.5087,
+      "step": 1069
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4552373319650914e-06,
+      "loss": 0.5113,
+      "step": 1070
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4421872769855262e-06,
+      "loss": 0.5177,
+      "step": 1071
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4291914511967141e-06,
+      "loss": 0.5573,
+      "step": 1072
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4162499369506655e-06,
+      "loss": 0.5431,
+      "step": 1073
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.403362816255236e-06,
+      "loss": 0.5109,
+      "step": 1074
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3905301707735986e-06,
+      "loss": 0.5452,
+      "step": 1075
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3777520818237244e-06,
+      "loss": 0.5398,
+      "step": 1076
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3650286303778715e-06,
+      "loss": 0.5277,
+      "step": 1077
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3523598970620687e-06,
+      "loss": 0.5233,
+      "step": 1078
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.527,
+      "step": 1079
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3271869055905495e-06,
+      "loss": 0.5268,
+      "step": 1080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3146828069511708e-06,
+      "loss": 0.5579,
+      "step": 1081
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.302233745473509e-06,
+      "loss": 0.4865,
+      "step": 1082
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2898398000448441e-06,
+      "loss": 0.5064,
+      "step": 1083
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2775010492031915e-06,
+      "loss": 0.5106,
+      "step": 1084
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2652175711368075e-06,
+      "loss": 0.5242,
+      "step": 1085
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2529894436836965e-06,
+      "loss": 0.5229,
+      "step": 1086
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2408167443311215e-06,
+      "loss": 0.4811,
+      "step": 1087
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2286995502151045e-06,
+      "loss": 0.5107,
+      "step": 1088
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2166379381199423e-06,
+      "loss": 0.5165,
+      "step": 1089
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2046319844777144e-06,
+      "loss": 0.5222,
+      "step": 1090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1926817653678159e-06,
+      "loss": 0.5134,
+      "step": 1091
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1807873565164507e-06,
+      "loss": 0.5449,
+      "step": 1092
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1689488332961684e-06,
+      "loss": 0.5073,
+      "step": 1093
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.157166270725385e-06,
+      "loss": 0.5012,
+      "step": 1094
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1454397434679022e-06,
+      "loss": 0.5389,
+      "step": 1095
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1337693258324344e-06,
+      "loss": 0.4932,
+      "step": 1096
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1221550917721434e-06,
+      "loss": 0.5297,
+      "step": 1097
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.11059711488417e-06,
+      "loss": 0.528,
+      "step": 1098
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.099095468409156e-06,
+      "loss": 0.5401,
+      "step": 1099
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0876502252307942e-06,
+      "loss": 0.5006,
+      "step": 1100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0762614578753571e-06,
+      "loss": 0.5715,
+      "step": 1101
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0649292385112474e-06,
+      "loss": 0.5373,
+      "step": 1102
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0536536389485275e-06,
+      "loss": 0.5585,
+      "step": 1103
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.042434730638473e-06,
+      "loss": 0.5343,
+      "step": 1104
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0312725846731174e-06,
+      "loss": 0.4908,
+      "step": 1105
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0201672717848065e-06,
+      "loss": 0.5571,
+      "step": 1106
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0091188623457415e-06,
+      "loss": 0.5796,
+      "step": 1107
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.981274263675388e-07,
+      "loss": 0.4938,
+      "step": 1108
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.87193033500784e-07,
+      "loss": 0.5294,
+      "step": 1109
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.763157530345957e-07,
+      "loss": 0.4903,
+      "step": 1110
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.65495653896179e-07,
+      "loss": 0.5719,
+      "step": 1111
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.547328046503889e-07,
+      "loss": 0.5182,
+      "step": 1112
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.440272734993072e-07,
+      "loss": 0.4994,
+      "step": 1113
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.333791282817939e-07,
+      "loss": 0.5292,
+      "step": 1114
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.227884364730743e-07,
+      "loss": 0.5275,
+      "step": 1115
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.122552651842931e-07,
+      "loss": 0.5171,
+      "step": 1116
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.017796811621049e-07,
+      "loss": 0.5007,
+      "step": 1117
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.91361750788241e-07,
+      "loss": 0.4969,
+      "step": 1118
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.810015400790994e-07,
+      "loss": 0.5202,
+      "step": 1119
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.706991146853128e-07,
+      "loss": 0.5295,
+      "step": 1120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.604545398913422e-07,
+      "loss": 0.5608,
+      "step": 1121
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.502678806150588e-07,
+      "loss": 0.5585,
+      "step": 1122
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.401392014073406e-07,
+      "loss": 0.5148,
+      "step": 1123
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.300685664516516e-07,
+      "loss": 0.5317,
+      "step": 1124
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.200560395636414e-07,
+      "loss": 0.5315,
+      "step": 1125
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.101016841907394e-07,
+      "loss": 0.5128,
+      "step": 1126
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.002055634117578e-07,
+      "loss": 0.5457,
+      "step": 1127
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.903677399364839e-07,
+      "loss": 0.5104,
+      "step": 1128
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.805882761052852e-07,
+      "loss": 0.5365,
+      "step": 1129
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.70867233888718e-07,
+      "loss": 0.5182,
+      "step": 1130
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.5384,
+      "step": 1131
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.516006603302816e-07,
+      "loss": 0.5168,
+      "step": 1132
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.420552510769308e-07,
+      "loss": 0.5276,
+      "step": 1133
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.325685076144795e-07,
+      "loss": 0.5046,
+      "step": 1134
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.231404900585714e-07,
+      "loss": 0.5232,
+      "step": 1135
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.137712581527157e-07,
+      "loss": 0.5289,
+      "step": 1136
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.044608712679058e-07,
+      "loss": 0.4747,
+      "step": 1137
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.952093884022515e-07,
+      "loss": 0.4885,
+      "step": 1138
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.860168681805946e-07,
+      "loss": 0.4941,
+      "step": 1139
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.768833688541443e-07,
+      "loss": 0.5454,
+      "step": 1140
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.678089483001016e-07,
+      "loss": 0.5212,
+      "step": 1141
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.587936640213055e-07,
+      "loss": 0.5718,
+      "step": 1142
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.498375731458529e-07,
+      "loss": 0.5347,
+      "step": 1143
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.409407324267448e-07,
+      "loss": 0.5294,
+      "step": 1144
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.321031982415237e-07,
+      "loss": 0.496,
+      "step": 1145
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.233250265919266e-07,
+      "loss": 0.4798,
+      "step": 1146
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.146062731035129e-07,
+      "loss": 0.5078,
+      "step": 1147
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.059469930253226e-07,
+      "loss": 0.4922,
+      "step": 1148
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.973472412295256e-07,
+      "loss": 0.4842,
+      "step": 1149
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.88807072211074e-07,
+      "loss": 0.5363,
+      "step": 1150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.803265400873514e-07,
+      "loss": 0.5368,
+      "step": 1151
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.719056985978388e-07,
+      "loss": 0.5265,
+      "step": 1152
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.63544601103767e-07,
+      "loss": 0.5912,
+      "step": 1153
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.552433005877778e-07,
+      "loss": 0.5244,
+      "step": 1154
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.470018496535967e-07,
+      "loss": 0.5146,
+      "step": 1155
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.38820300525692e-07,
+      "loss": 0.4954,
+      "step": 1156
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.306987050489442e-07,
+      "loss": 0.5772,
+      "step": 1157
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.22637114688318e-07,
+      "loss": 0.5465,
+      "step": 1158
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.146355805285452e-07,
+      "loss": 0.5659,
+      "step": 1159
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.066941532737834e-07,
+      "loss": 0.5397,
+      "step": 1160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.988128832473105e-07,
+      "loss": 0.5096,
+      "step": 1161
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.909918203911968e-07,
+      "loss": 0.4733,
+      "step": 1162
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.832310142659946e-07,
+      "loss": 0.5798,
+      "step": 1163
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.755305140504185e-07,
+      "loss": 0.4979,
+      "step": 1164
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.678903685410374e-07,
+      "loss": 0.5101,
+      "step": 1165
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6031062615196186e-07,
+      "loss": 0.5389,
+      "step": 1166
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5279133491454406e-07,
+      "loss": 0.5185,
+      "step": 1167
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4533254247706604e-07,
+      "loss": 0.5902,
+      "step": 1168
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3793429610444083e-07,
+      "loss": 0.5341,
+      "step": 1169
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.305966426779118e-07,
+      "loss": 0.5198,
+      "step": 1170
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.233196286947605e-07,
+      "loss": 0.4886,
+      "step": 1171
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.161033002680059e-07,
+      "loss": 0.5199,
+      "step": 1172
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.089477031261113e-07,
+      "loss": 0.5458,
+      "step": 1173
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.018528826127044e-07,
+      "loss": 0.5011,
+      "step": 1174
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9481888368627764e-07,
+      "loss": 0.5216,
+      "step": 1175
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.878457509199107e-07,
+      "loss": 0.4985,
+      "step": 1176
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.809335285009863e-07,
+      "loss": 0.4659,
+      "step": 1177
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.740822602309124e-07,
+      "loss": 0.4945,
+      "step": 1178
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6729198952483725e-07,
+      "loss": 0.516,
+      "step": 1179
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6056275941138474e-07,
+      "loss": 0.5352,
+      "step": 1180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.538946125323717e-07,
+      "loss": 0.5155,
+      "step": 1181
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4728759114254774e-07,
+      "loss": 0.4905,
+      "step": 1182
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 0.5385,
+      "step": 1183
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3425709191248325e-07,
+      "loss": 0.5162,
+      "step": 1184
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.278336966439744e-07,
+      "loss": 0.5106,
+      "step": 1185
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2147159200759723e-07,
+      "loss": 0.5354,
+      "step": 1186
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.151708183187674e-07,
+      "loss": 0.5781,
+      "step": 1187
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.089314155042589e-07,
+      "loss": 0.5132,
+      "step": 1188
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0275342310195133e-07,
+      "loss": 0.5541,
+      "step": 1189
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9663688026057793e-07,
+      "loss": 0.5424,
+      "step": 1190
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.905818257394799e-07,
+      "loss": 0.5082,
+      "step": 1191
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8458829790835474e-07,
+      "loss": 0.5351,
+      "step": 1192
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.786563347470217e-07,
+      "loss": 0.5152,
+      "step": 1193
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.727859738451721e-07,
+      "loss": 0.5417,
+      "step": 1194
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6697725240214076e-07,
+      "loss": 0.5221,
+      "step": 1195
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.612302072266637e-07,
+      "loss": 0.5327,
+      "step": 1196
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5554487473664404e-07,
+      "loss": 0.5198,
+      "step": 1197
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.499212909589244e-07,
+      "loss": 0.5331,
+      "step": 1198
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4435949152906144e-07,
+      "loss": 0.5349,
+      "step": 1199
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.388595116910919e-07,
+      "loss": 0.5253,
+      "step": 1200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.334213862973167e-07,
+      "loss": 0.524,
+      "step": 1201
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.280451498080749e-07,
+      "loss": 0.5217,
+      "step": 1202
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2273083629153148e-07,
+      "loss": 0.5278,
+      "step": 1203
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1747847942345191e-07,
+      "loss": 0.5448,
+      "step": 1204
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.122881124870002e-07,
+      "loss": 0.5466,
+      "step": 1205
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0715976837251793e-07,
+      "loss": 0.5684,
+      "step": 1206
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0209347957732328e-07,
+      "loss": 0.5433,
+      "step": 1207
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9708927820550007e-07,
+      "loss": 0.547,
+      "step": 1208
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.921471959676957e-07,
+      "loss": 0.5408,
+      "step": 1209
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8726726418092257e-07,
+      "loss": 0.536,
+      "step": 1210
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8244951376835907e-07,
+      "loss": 0.4941,
+      "step": 1211
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7769397525914668e-07,
+      "loss": 0.5647,
+      "step": 1212
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.730006787882077e-07,
+      "loss": 0.534,
+      "step": 1213
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6836965409604666e-07,
+      "loss": 0.535,
+      "step": 1214
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6380093052856482e-07,
+      "loss": 0.5618,
+      "step": 1215
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.592945370368715e-07,
+      "loss": 0.5552,
+      "step": 1216
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5485050217710295e-07,
+      "loss": 0.5793,
+      "step": 1217
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5046885411024393e-07,
+      "loss": 0.5114,
+      "step": 1218
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4614962060194303e-07,
+      "loss": 0.5431,
+      "step": 1219
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.418928290223398e-07,
+      "loss": 0.5152,
+      "step": 1220
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3769850634589356e-07,
+      "loss": 0.5397,
+      "step": 1221
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3356667915121025e-07,
+      "loss": 0.5259,
+      "step": 1222
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2949737362087156e-07,
+      "loss": 0.4808,
+      "step": 1223
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2549061554127494e-07,
+      "loss": 0.5815,
+      "step": 1224
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2154643030246273e-07,
+      "loss": 0.5499,
+      "step": 1225
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1766484289796987e-07,
+      "loss": 0.4773,
+      "step": 1226
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1384587792465873e-07,
+      "loss": 0.5283,
+      "step": 1227
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1008955958256462e-07,
+      "loss": 0.5238,
+      "step": 1228
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0639591167474483e-07,
+      "loss": 0.5141,
+      "step": 1229
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0276495760712768e-07,
+      "loss": 0.5436,
+      "step": 1230
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.919672038835926e-08,
+      "loss": 0.5246,
+      "step": 1231
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.569122262966468e-08,
+      "loss": 0.5326,
+      "step": 1232
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.224848654469932e-08,
+      "loss": 0.5491,
+      "step": 1233
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.886853394940887e-08,
+      "loss": 0.5598,
+      "step": 1234
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.555138626189619e-08,
+      "loss": 0.5204,
+      "step": 1235
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.229706450227804e-08,
+      "loss": 0.5298,
+      "step": 1236
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.910558929255852e-08,
+      "loss": 0.4973,
+      "step": 1237
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.59769808564914e-08,
+      "loss": 0.4827,
+      "step": 1238
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.291125901946027e-08,
+      "loss": 0.5202,
+      "step": 1239
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.990844320834633e-08,
+      "loss": 0.5392,
+      "step": 1240
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.696855245140743e-08,
+      "loss": 0.5167,
+      "step": 1241
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.409160537815818e-08,
+      "loss": 0.4814,
+      "step": 1242
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.127762021925221e-08,
+      "loss": 0.4992,
+      "step": 1243
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8526614806363456e-08,
+      "loss": 0.5211,
+      "step": 1244
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5838606572078404e-08,
+      "loss": 0.5153,
+      "step": 1245
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3213612549779524e-08,
+      "loss": 0.5662,
+      "step": 1246
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.0651649373544276e-08,
+      "loss": 0.5081,
+      "step": 1247
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.815273327803183e-08,
+      "loss": 0.4951,
+      "step": 1248
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.571688009838648e-08,
+      "loss": 0.5379,
+      "step": 1249
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.334410527013555e-08,
+      "loss": 0.5086,
+      "step": 1250
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.103442382909051e-08,
+      "loss": 0.4992,
+      "step": 1251
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8787850411252656e-08,
+      "loss": 0.5843,
+      "step": 1252
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.660439925271764e-08,
+      "loss": 0.4943,
+      "step": 1253
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.448408418959326e-08,
+      "loss": 0.5269,
+      "step": 1254
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.242691865790071e-08,
+      "loss": 0.5381,
+      "step": 1255
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.043291569349571e-08,
+      "loss": 0.5113,
+      "step": 1256
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.850208793198861e-08,
+      "loss": 0.5399,
+      "step": 1257
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6634447608657745e-08,
+      "loss": 0.517,
+      "step": 1258
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4830006558373975e-08,
+      "loss": 0.5314,
+      "step": 1259
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.308877621553185e-08,
+      "loss": 0.5512,
+      "step": 1260
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1410767613965212e-08,
+      "loss": 0.5497,
+      "step": 1261
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.979599138688948e-08,
+      "loss": 0.5278,
+      "step": 1262
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.824445776682504e-08,
+      "loss": 0.5838,
+      "step": 1263
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.675617658554063e-08,
+      "loss": 0.5841,
+      "step": 1264
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5331157273981155e-08,
+      "loss": 0.5178,
+      "step": 1265
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.396940886221776e-08,
+      "loss": 0.5418,
+      "step": 1266
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.267093997938451e-08,
+      "loss": 0.5459,
+      "step": 1267
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1435758853624025e-08,
+      "loss": 0.5416,
+      "step": 1268
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0263873312040818e-08,
+      "loss": 0.4738,
+      "step": 1269
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.155290780643588e-09,
+      "loss": 0.4997,
+      "step": 1270
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.110018284304132e-09,
+      "loss": 0.5079,
+      "step": 1271
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.128062446709605e-09,
+      "loss": 0.5233,
+      "step": 1272
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.209429490323659e-09,
+      "loss": 0.5468,
+      "step": 1273
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.354125236343155e-09,
+      "loss": 0.4702,
+      "step": 1274
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.562155104665955e-09,
+      "loss": 0.5532,
+      "step": 1275
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.833524113852072e-09,
+      "loss": 0.5876,
+      "step": 1276
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.168236881092579e-09,
+      "loss": 0.5518,
+      "step": 1277
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5662976221840772e-09,
+      "loss": 0.5515,
+      "step": 1278
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0277101514987184e-09,
+      "loss": 0.5418,
+      "step": 1279
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5524778819597797e-09,
+      "loss": 0.5568,
+      "step": 1280
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1406038250205699e-09,
+      "loss": 0.5101,
+      "step": 1281
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.920905906488862e-10,
+      "loss": 0.5384,
+      "step": 1282
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.069403873025902e-10,
+      "loss": 0.505,
+      "step": 1283
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.851550219240551e-10,
+      "loss": 0.5432,
+      "step": 1284
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2673589992462377e-10,
+      "loss": 0.5034,
+      "step": 1285
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.168402517461644e-11,
+      "loss": 0.5194,
+      "step": 1286
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.5364,
+      "step": 1287
+    },
+    {
+      "epoch": 3.0,
+      "step": 1287,
+      "total_flos": 1.6828647573018378e+18,
+      "train_loss": 0.6138159237634144,
+      "train_runtime": 12078.4868,
+      "train_samples_per_second": 13.654,
+      "train_steps_per_second": 0.107
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1287,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10000,
+  "total_flos": 1.6828647573018378e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}