diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18780 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999000099990001,
+  "eval_steps": 500,
+  "global_step": 3125,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0638297872340426e-07,
+      "loss": 2.651,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1276595744680852e-07,
+      "loss": 2.4004,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1914893617021275e-07,
+      "loss": 2.4933,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2553191489361704e-07,
+      "loss": 2.3427,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.319148936170213e-07,
+      "loss": 2.4238,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.382978723404255e-07,
+      "loss": 2.3589,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.446808510638298e-07,
+      "loss": 2.4474,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.510638297872341e-07,
+      "loss": 2.3338,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.574468085106384e-07,
+      "loss": 2.4259,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0638297872340427e-06,
+      "loss": 2.6739,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.170212765957447e-06,
+      "loss": 2.6681,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.276595744680851e-06,
+      "loss": 2.632,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3829787234042555e-06,
+      "loss": 2.368,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4893617021276596e-06,
+      "loss": 2.3466,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.595744680851064e-06,
+      "loss": 2.4533,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7021276595744682e-06,
+      "loss": 2.5377,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8085106382978727e-06,
+      "loss": 2.4521,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9148936170212767e-06,
+      "loss": 2.1765,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.021276595744681e-06,
+      "loss": 2.3687,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1276595744680853e-06,
+      "loss": 2.5292,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2340425531914894e-06,
+      "loss": 2.5995,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.340425531914894e-06,
+      "loss": 2.6008,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.446808510638298e-06,
+      "loss": 2.3642,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.553191489361702e-06,
+      "loss": 2.4682,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6595744680851065e-06,
+      "loss": 2.516,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.765957446808511e-06,
+      "loss": 2.4319,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8723404255319155e-06,
+      "loss": 2.5353,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.978723404255319e-06,
+      "loss": 2.5288,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0851063829787237e-06,
+      "loss": 2.5427,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.191489361702128e-06,
+      "loss": 2.4972,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.297872340425532e-06,
+      "loss": 2.5212,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4042553191489363e-06,
+      "loss": 2.8224,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.510638297872341e-06,
+      "loss": 2.6402,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6170212765957453e-06,
+      "loss": 2.4022,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.723404255319149e-06,
+      "loss": 2.609,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8297872340425535e-06,
+      "loss": 2.6311,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.936170212765958e-06,
+      "loss": 2.4299,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.042553191489362e-06,
+      "loss": 2.5973,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.148936170212766e-06,
+      "loss": 2.5149,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 2.3426,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.361702127659575e-06,
+      "loss": 2.3869,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.468085106382979e-06,
+      "loss": 2.4518,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.574468085106383e-06,
+      "loss": 2.5953,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.680851063829788e-06,
+      "loss": 2.4407,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.787234042553192e-06,
+      "loss": 2.4744,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.893617021276596e-06,
+      "loss": 2.4609,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 2.3709,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.106382978723404e-06,
+      "loss": 2.4096,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.212765957446809e-06,
+      "loss": 2.4081,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.319148936170213e-06,
+      "loss": 2.3029,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.425531914893617e-06,
+      "loss": 2.3843,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.531914893617022e-06,
+      "loss": 2.6355,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.638297872340426e-06,
+      "loss": 2.3764,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.744680851063831e-06,
+      "loss": 2.3649,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.851063829787235e-06,
+      "loss": 2.3193,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.957446808510638e-06,
+      "loss": 2.298,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.063829787234044e-06,
+      "loss": 2.3387,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.170212765957447e-06,
+      "loss": 2.2963,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.276595744680851e-06,
+      "loss": 2.1387,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.382978723404256e-06,
+      "loss": 2.392,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.48936170212766e-06,
+      "loss": 2.1749,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.595744680851064e-06,
+      "loss": 2.3058,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.702127659574469e-06,
+      "loss": 2.3141,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.808510638297873e-06,
+      "loss": 2.2407,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.914893617021278e-06,
+      "loss": 2.3009,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.021276595744682e-06,
+      "loss": 2.1976,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.127659574468085e-06,
+      "loss": 1.9817,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.234042553191491e-06,
+      "loss": 2.0944,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.340425531914894e-06,
+      "loss": 2.0039,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.446808510638298e-06,
+      "loss": 1.8105,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.553191489361703e-06,
+      "loss": 2.0203,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.659574468085107e-06,
+      "loss": 1.9992,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.765957446808511e-06,
+      "loss": 1.9564,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.872340425531916e-06,
+      "loss": 2.0312,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.97872340425532e-06,
+      "loss": 1.9039,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.085106382978723e-06,
+      "loss": 2.0643,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.191489361702128e-06,
+      "loss": 2.1061,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.297872340425532e-06,
+      "loss": 1.8455,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.404255319148937e-06,
+      "loss": 2.0594,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 1.964,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.617021276595746e-06,
+      "loss": 2.0406,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.72340425531915e-06,
+      "loss": 2.0131,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.829787234042555e-06,
+      "loss": 2.0538,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.936170212765958e-06,
+      "loss": 1.9504,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.042553191489362e-06,
+      "loss": 2.029,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.148936170212767e-06,
+      "loss": 1.942,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.255319148936171e-06,
+      "loss": 1.7639,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.361702127659576e-06,
+      "loss": 2.0619,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.46808510638298e-06,
+      "loss": 1.9878,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.574468085106385e-06,
+      "loss": 1.8802,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.680851063829787e-06,
+      "loss": 2.0544,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.787234042553192e-06,
+      "loss": 1.8512,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.893617021276596e-06,
+      "loss": 2.0012,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 2.0257,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999997314236036e-06,
+      "loss": 1.8799,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999989256947029e-06,
+      "loss": 1.9646,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999975828141635e-06,
+      "loss": 1.9381,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999957027834282e-06,
+      "loss": 1.8833,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999932856045164e-06,
+      "loss": 1.8755,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99990331280025e-06,
+      "loss": 1.8335,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999868398131282e-06,
+      "loss": 1.8557,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999828112075764e-06,
+      "loss": 1.8033,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99978245467698e-06,
+      "loss": 1.8602,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999731425983975e-06,
+      "loss": 1.6957,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999675026051576e-06,
+      "loss": 1.8498,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999613254940368e-06,
+      "loss": 1.7582,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999546112716715e-06,
+      "loss": 1.8517,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999473599452746e-06,
+      "loss": 1.8381,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999395715226365e-06,
+      "loss": 1.9382,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999312460121242e-06,
+      "loss": 1.8322,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999223834226817e-06,
+      "loss": 1.8451,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999129837638303e-06,
+      "loss": 1.8422,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999030470456684e-06,
+      "loss": 1.8487,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998925732788706e-06,
+      "loss": 1.8712,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99881562474689e-06,
+      "loss": 1.7937,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998700146449528e-06,
+      "loss": 1.8627,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998579298020676e-06,
+      "loss": 1.8512,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998453079590167e-06,
+      "loss": 1.7684,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998321491293592e-06,
+      "loss": 1.7062,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998184533272321e-06,
+      "loss": 1.7993,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998042205673489e-06,
+      "loss": 1.9294,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997894508649995e-06,
+      "loss": 1.733,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997741442360515e-06,
+      "loss": 1.8085,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99758300696949e-06,
+      "loss": 1.754,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997419202647124e-06,
+      "loss": 1.7942,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997250029569395e-06,
+      "loss": 1.8257,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997075487918047e-06,
+      "loss": 1.7591,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99689557788059e-06,
+      "loss": 1.8277,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996710299650302e-06,
+      "loss": 1.8232,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996519653426229e-06,
+      "loss": 1.7878,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996323639413185e-06,
+      "loss": 1.8788,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996122257821746e-06,
+      "loss": 1.8268,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99591550886826e-06,
+      "loss": 1.7563,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995703392774836e-06,
+      "loss": 1.8308,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995485909769354e-06,
+      "loss": 1.7709,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995263060085456e-06,
+      "loss": 1.7351,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99503484396255e-06,
+      "loss": 1.7174,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99480126164581e-06,
+      "loss": 1.8137,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994562313386177e-06,
+      "loss": 1.8221,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994317999440351e-06,
+      "loss": 1.6661,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994068320070805e-06,
+      "loss": 1.8035,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993813275545764e-06,
+      "loss": 1.7718,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99355286613923e-06,
+      "loss": 1.7583,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993287092130956e-06,
+      "loss": 1.7783,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993015953806472e-06,
+      "loss": 1.7476,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992739451457058e-06,
+      "loss": 1.8032,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992457585379764e-06,
+      "loss": 1.8038,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992170355877398e-06,
+      "loss": 1.7615,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991877763258538e-06,
+      "loss": 1.7914,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991579807837511e-06,
+      "loss": 1.7995,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991276489934416e-06,
+      "loss": 1.7739,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990967809875107e-06,
+      "loss": 1.9179,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990653767991203e-06,
+      "loss": 1.725,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99033436462008e-06,
+      "loss": 1.8729,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990009600104875e-06,
+      "loss": 1.8004,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989679474794484e-06,
+      "loss": 1.8444,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989343989043563e-06,
+      "loss": 1.8067,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989003143212526e-06,
+      "loss": 1.7493,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988656937667544e-06,
+      "loss": 1.8416,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98830537278055e-06,
+      "loss": 1.7412,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987948448929232e-06,
+      "loss": 1.7681,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987586166497032e-06,
+      "loss": 1.6424,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987218525873155e-06,
+      "loss": 1.6834,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98684552745256e-06,
+      "loss": 1.8012,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98646717163596e-06,
+      "loss": 1.7224,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986083458829824e-06,
+      "loss": 1.8329,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985694389446378e-06,
+      "loss": 1.7853,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9852999639036e-06,
+      "loss": 1.7374,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984900182625226e-06,
+      "loss": 1.7672,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98449504604074e-06,
+      "loss": 1.7901,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984084554585387e-06,
+      "loss": 1.5986,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983668708700156e-06,
+      "loss": 1.7909,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983247508831795e-06,
+      "loss": 1.7336,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9828209554328e-06,
+      "loss": 1.6962,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982389048961421e-06,
+      "loss": 1.7748,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981951789881657e-06,
+      "loss": 1.7922,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981509178663256e-06,
+      "loss": 1.8702,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98106121578172e-06,
+      "loss": 1.7221,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980607901718297e-06,
+      "loss": 1.8153,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980149236959986e-06,
+      "loss": 1.7762,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979685221999532e-06,
+      "loss": 1.8802,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97921585733543e-06,
+      "loss": 1.8059,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97874114347192e-06,
+      "loss": 1.6782,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978261080918988e-06,
+      "loss": 1.807,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977775670192373e-06,
+      "loss": 1.8344,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977284911813549e-06,
+      "loss": 1.6362,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976788806309742e-06,
+      "loss": 1.8029,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976287354213924e-06,
+      "loss": 1.7458,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975780556064806e-06,
+      "loss": 1.7918,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975268412406842e-06,
+      "loss": 1.7913,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974750923790234e-06,
+      "loss": 1.8464,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97422809077092e-06,
+      "loss": 1.6998,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973699913910584e-06,
+      "loss": 1.8007,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97316639377665e-06,
+      "loss": 1.7287,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97262753094228e-06,
+      "loss": 1.7644,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972083325986377e-06,
+      "loss": 1.706,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971533779493586e-06,
+      "loss": 1.7447,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970978892054286e-06,
+      "loss": 1.7924,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970418664264596e-06,
+      "loss": 1.7637,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969853096726372e-06,
+      "loss": 1.6764,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969282190047207e-06,
+      "loss": 1.746,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968705944840428e-06,
+      "loss": 1.7779,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968124361725098e-06,
+      "loss": 1.8048,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967537441326018e-06,
+      "loss": 1.8066,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966945184273716e-06,
+      "loss": 1.8003,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966347591204459e-06,
+      "loss": 1.8642,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965744662760246e-06,
+      "loss": 1.8044,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965136399588803e-06,
+      "loss": 1.6783,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964522802343593e-06,
+      "loss": 1.7793,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963903871683806e-06,
+      "loss": 1.7159,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963279608274364e-06,
+      "loss": 1.7482,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962650012785917e-06,
+      "loss": 1.6607,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962015085894838e-06,
+      "loss": 1.7147,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961374828283239e-06,
+      "loss": 1.6376,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960729240638947e-06,
+      "loss": 1.691,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960078323655524e-06,
+      "loss": 1.6669,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959422078032253e-06,
+      "loss": 1.7556,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958760504474144e-06,
+      "loss": 1.8202,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958093603691923e-06,
+      "loss": 1.6916,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957421376402053e-06,
+      "loss": 1.5906,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956743823326704e-06,
+      "loss": 1.6485,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956060945193781e-06,
+      "loss": 1.7653,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955372742736903e-06,
+      "loss": 1.7309,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954679216695406e-06,
+      "loss": 1.6663,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953980367814354e-06,
+      "loss": 1.6583,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953276196844519e-06,
+      "loss": 1.769,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9525667045424e-06,
+      "loss": 1.7377,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951851891670206e-06,
+      "loss": 1.6518,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951131758995866e-06,
+      "loss": 1.6743,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950406307293023e-06,
+      "loss": 1.76,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949675537341031e-06,
+      "loss": 1.6835,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948939449924964e-06,
+      "loss": 1.7242,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948198045835601e-06,
+      "loss": 1.8006,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94745132586944e-06,
+      "loss": 1.7477,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946699290828683e-06,
+      "loss": 1.7631,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94594194152125e-06,
+      "loss": 1.8108,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945179278760759e-06,
+      "loss": 1.6784,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94441130336655e-06,
+      "loss": 1.7795,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943638016163658e-06,
+      "loss": 1.694,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942859417982833e-06,
+      "loss": 1.6576,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942075509660527e-06,
+      "loss": 1.7555,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941286292038894e-06,
+      "loss": 1.7107,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940491765965798e-06,
+      "loss": 1.7697,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939691932294804e-06,
+      "loss": 1.7247,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938886791885172e-06,
+      "loss": 1.7173,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938076345601875e-06,
+      "loss": 1.6966,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937260594315578e-06,
+      "loss": 1.7562,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936439538902644e-06,
+      "loss": 1.8112,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935613180245143e-06,
+      "loss": 1.6941,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934781519230832e-06,
+      "loss": 1.7425,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933944556753173e-06,
+      "loss": 1.6841,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933102293711314e-06,
+      "loss": 1.6951,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932254731010108e-06,
+      "loss": 1.7045,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931401869560096e-06,
+      "loss": 1.6628,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93054371027751e-06,
+      "loss": 1.8037,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929680254084273e-06,
+      "loss": 1.6148,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928811501908006e-06,
+      "loss": 1.6917,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92793745468201e-06,
+      "loss": 1.7002,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927058113345282e-06,
+      "loss": 1.7342,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926173478842502e-06,
+      "loss": 1.7739,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925283552124039e-06,
+      "loss": 1.6155,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924388334145943e-06,
+      "loss": 1.6434,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923487825869955e-06,
+      "loss": 1.7202,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922582028263495e-06,
+      "loss": 1.7412,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921670942299664e-06,
+      "loss": 1.8052,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92075456895725e-06,
+      "loss": 1.722,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919832909220717e-06,
+      "loss": 1.8543,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91890596408021e-06,
+      "loss": 1.7788,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917973734531549e-06,
+      "loss": 1.7856,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917036221576235e-06,
+      "loss": 1.7479,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916093426221445e-06,
+      "loss": 1.6891,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915145349480027e-06,
+      "loss": 1.7325,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914191992370504e-06,
+      "loss": 1.8139,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913233355917075e-06,
+      "loss": 1.7722,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91226944114961e-06,
+      "loss": 1.7284,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911300249103646e-06,
+      "loss": 1.685,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910325780820391e-06,
+      "loss": 1.7423,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90934603734672e-06,
+      "loss": 1.6864,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908361019735181e-06,
+      "loss": 1.797,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907370729043984e-06,
+      "loss": 1.7033,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906375166336998e-06,
+      "loss": 1.6409,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905374332683768e-06,
+      "loss": 1.6727,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904368229159494e-06,
+      "loss": 1.6517,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903356856845035e-06,
+      "loss": 1.7009,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902340216826915e-06,
+      "loss": 1.7839,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90131831019732e-06,
+      "loss": 1.665,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900291138054086e-06,
+      "loss": 1.7989,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899258701500712e-06,
+      "loss": 1.6963,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89822100164635e-06,
+      "loss": 1.743,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897178039605803e-06,
+      "loss": 1.7103,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896129816499535e-06,
+      "loss": 1.5754,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89507633345366e-06,
+      "loss": 1.6976,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894017591599934e-06,
+      "loss": 1.7608,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892953592075776e-06,
+      "loss": 1.6915,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891884336024242e-06,
+      "loss": 1.648,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890809824594041e-06,
+      "loss": 1.6183,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889730058939529e-06,
+      "loss": 1.8362,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8886450402207e-06,
+      "loss": 1.6633,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8875547696032e-06,
+      "loss": 1.7289,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88645924825831e-06,
+      "loss": 1.6346,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885358477362956e-06,
+      "loss": 1.5483,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8842524580997e-06,
+      "loss": 1.7838,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883141191656748e-06,
+      "loss": 1.8372,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88202467922794e-06,
+      "loss": 1.8388,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880902922012747e-06,
+      "loss": 1.6653,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879775921216284e-06,
+      "loss": 1.7297,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87864367804929e-06,
+      "loss": 1.6776,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877506193728144e-06,
+      "loss": 1.7087,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876363469474848e-06,
+      "loss": 1.7207,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87521550651704e-06,
+      "loss": 1.8512,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874062306087983e-06,
+      "loss": 1.7535,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872903869426564e-06,
+      "loss": 1.5693,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8717401977773e-06,
+      "loss": 1.7237,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870571292390331e-06,
+      "loss": 1.8489,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869397154521418e-06,
+      "loss": 1.7989,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868217785431942e-06,
+      "loss": 1.6084,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867033186388906e-06,
+      "loss": 1.7534,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865843358664933e-06,
+      "loss": 1.6955,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86464830353826e-06,
+      "loss": 1.7656,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863448022292742e-06,
+      "loss": 1.7786,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86224251621785e-06,
+      "loss": 1.7406,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861031786608663e-06,
+      "loss": 1.7005,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859815834765875e-06,
+      "loss": 1.7332,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858594661995792e-06,
+      "loss": 1.8024,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857368269610325e-06,
+      "loss": 1.6857,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856136658926993e-06,
+      "loss": 1.7378,
+      "step": 326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854899831268926e-06,
+      "loss": 1.8801,
+      "step": 327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85365778796485e-06,
+      "loss": 1.6551,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852410530349102e-06,
+      "loss": 1.6253,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851158059761617e-06,
+      "loss": 1.6853,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849900377547933e-06,
+      "loss": 1.6277,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848637485059183e-06,
+      "loss": 1.6785,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8473693836521e-06,
+      "loss": 1.7196,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846096074689012e-06,
+      "loss": 1.6358,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844817559537841e-06,
+      "loss": 1.7292,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843533839572105e-06,
+      "loss": 1.622,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842244916170913e-06,
+      "loss": 1.7749,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840950790718959e-06,
+      "loss": 1.6924,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83965146460653e-06,
+      "loss": 1.7013,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838346939229501e-06,
+      "loss": 1.7429,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83703721598933e-06,
+      "loss": 1.6255,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835722296293058e-06,
+      "loss": 1.7166,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834402181553314e-06,
+      "loss": 1.7175,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833076873188303e-06,
+      "loss": 1.6857,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831746372621811e-06,
+      "loss": 1.7024,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830410681283203e-06,
+      "loss": 1.7783,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829069800607418e-06,
+      "loss": 1.677,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827723732034972e-06,
+      "loss": 1.6971,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826372477011956e-06,
+      "loss": 1.7409,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825016036990029e-06,
+      "loss": 1.7133,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823654413426424e-06,
+      "loss": 1.7899,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822287607783938e-06,
+      "loss": 1.7277,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820915621530939e-06,
+      "loss": 1.7343,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81953845614136e-06,
+      "loss": 1.754,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818156113094699e-06,
+      "loss": 1.6684,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816768593876012e-06,
+      "loss": 1.6446,
+      "step": 356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81537589997592e-06,
+      "loss": 1.7473,
+      "step": 357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8139780328906e-06,
+      "loss": 1.6668,
+      "step": 358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812574994121791e-06,
+      "loss": 1.8095,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811166785176785e-06,
+      "loss": 1.741,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809753407568427e-06,
+      "loss": 1.6371,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80833486281512e-06,
+      "loss": 1.7454,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80691115244081e-06,
+      "loss": 1.6965,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805482277974999e-06,
+      "loss": 1.7628,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804048240952736e-06,
+      "loss": 1.6845,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802609042914614e-06,
+      "loss": 1.701,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80116468540677e-06,
+      "loss": 1.7207,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79971516998089e-06,
+      "loss": 1.509,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79826049819419e-06,
+      "loss": 1.6471,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796800671609436e-06,
+      "loss": 1.5177,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795335691794929e-06,
+      "loss": 1.7073,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793865560324503e-06,
+      "loss": 1.5895,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792390278777527e-06,
+      "loss": 1.6819,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790909848738907e-06,
+      "loss": 1.5657,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789424271799075e-06,
+      "loss": 1.6086,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787933549553996e-06,
+      "loss": 1.5719,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786437683605161e-06,
+      "loss": 1.6742,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78493667555959e-06,
+      "loss": 1.6348,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783430527029818e-06,
+      "loss": 1.62,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781919239633912e-06,
+      "loss": 1.6231,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780402814995458e-06,
+      "loss": 1.7355,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77888125474356e-06,
+      "loss": 1.7294,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777354560512835e-06,
+      "loss": 1.7304,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77582273394342e-06,
+      "loss": 1.6239,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774285776680967e-06,
+      "loss": 1.7646,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772743690376636e-06,
+      "loss": 1.6867,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7711964766871e-06,
+      "loss": 1.7557,
+      "step": 387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76964413727454e-06,
+      "loss": 1.7433,
+      "step": 388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768086673806638e-06,
+      "loss": 1.659,
+      "step": 389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766524087956592e-06,
+      "loss": 1.6823,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764956381403095e-06,
+      "loss": 1.7398,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76338355583034e-06,
+      "loss": 1.6802,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761805612928025e-06,
+      "loss": 1.7687,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760222554391343e-06,
+      "loss": 1.7171,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758634381920982e-06,
+      "loss": 1.6782,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757041097223123e-06,
+      "loss": 1.6743,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755442702009443e-06,
+      "loss": 1.6624,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753839197997105e-06,
+      "loss": 1.7969,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752230586908767e-06,
+      "loss": 1.762,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75061687047256e-06,
+      "loss": 1.7237,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748998050422117e-06,
+      "loss": 1.6622,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747374128496541e-06,
+      "loss": 1.7124,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745745106440422e-06,
+      "loss": 1.6679,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744110986003826e-06,
+      "loss": 1.6886,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742471768942299e-06,
+      "loss": 1.613,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740827457016863e-06,
+      "loss": 1.7046,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739178051994008e-06,
+      "loss": 1.5909,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7375235556457e-06,
+      "loss": 1.7518,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735863969749373e-06,
+      "loss": 1.6134,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734199296087932e-06,
+      "loss": 1.6798,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732529536449741e-06,
+      "loss": 1.6985,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730854692628637e-06,
+      "loss": 1.671,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729174766423912e-06,
+      "loss": 1.758,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72748975964032e-06,
+      "loss": 1.7807,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725799674088072e-06,
+      "loss": 1.7571,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724104511582838e-06,
+      "loss": 1.6439,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72240427394574e-06,
+      "loss": 1.7175,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720698963003351e-06,
+      "loss": 1.732,
+      "step": 418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7189885805877e-06,
+      "loss": 1.657,
+      "step": 419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717273128536259e-06,
+      "loss": 1.7151,
+      "step": 420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715552608691944e-06,
+      "loss": 1.7349,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713827022903124e-06,
+      "loss": 1.6649,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712096373023603e-06,
+      "loss": 1.7311,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710360660912629e-06,
+      "loss": 1.5709,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708619888434887e-06,
+      "loss": 1.7688,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706874057460497e-06,
+      "loss": 1.6778,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705123169865016e-06,
+      "loss": 1.7095,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703367227529432e-06,
+      "loss": 1.6586,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701606232340165e-06,
+      "loss": 1.6535,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699840186189061e-06,
+      "loss": 1.6333,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698069090973391e-06,
+      "loss": 1.7181,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696292948595857e-06,
+      "loss": 1.808,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694511760964578e-06,
+      "loss": 1.6899,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69272552999309e-06,
+      "loss": 1.6908,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690934257600353e-06,
+      "loss": 1.6878,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689137945710742e-06,
+      "loss": 1.6547,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687336596254045e-06,
+      "loss": 1.5962,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685530211165459e-06,
+      "loss": 1.6643,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683718792385595e-06,
+      "loss": 1.6707,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681902341860471e-06,
+      "loss": 1.5834,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680080861541511e-06,
+      "loss": 1.6795,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678254353385538e-06,
+      "loss": 1.6583,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676422819354785e-06,
+      "loss": 1.6996,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674586261416874e-06,
+      "loss": 1.7052,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672744681544834e-06,
+      "loss": 1.6858,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670898081717079e-06,
+      "loss": 1.6946,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669046463917427e-06,
+      "loss": 1.643,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667189830135078e-06,
+      "loss": 1.7447,
+      "step": 448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665328182364627e-06,
+      "loss": 1.7659,
+      "step": 449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663461522606049e-06,
+      "loss": 1.7281,
+      "step": 450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66158985286471e-06,
+      "loss": 1.6406,
+      "step": 451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659713175151352e-06,
+      "loss": 1.5899,
+      "step": 452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657831491482103e-06,
+      "loss": 1.6914,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655944803878467e-06,
+      "loss": 1.6477,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654053114367321e-06,
+      "loss": 1.674,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65215642498092e-06,
+      "loss": 1.6972,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650254737756883e-06,
+      "loss": 1.6471,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648348054738208e-06,
+      "loss": 1.6628,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646436377973253e-06,
+      "loss": 1.7067,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644519709515746e-06,
+      "loss": 1.5877,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642598051424772e-06,
+      "loss": 1.7224,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640671405764777e-06,
+      "loss": 1.5512,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638739774605572e-06,
+      "loss": 1.6244,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636803160022314e-06,
+      "loss": 1.6312,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634861564095525e-06,
+      "loss": 1.7351,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632914988911066e-06,
+      "loss": 1.7559,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63096343656016e-06,
+      "loss": 1.6726,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629006909139363e-06,
+      "loss": 1.5844,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62704540875059e-06,
+      "loss": 1.6048,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625078937501089e-06,
+      "loss": 1.6805,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62310749750345e-06,
+      "loss": 1.6908,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621131090875603e-06,
+      "loss": 1.7334,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619149719740817e-06,
+      "loss": 1.8061,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617163386227683e-06,
+      "loss": 1.64,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615172092470134e-06,
+      "loss": 1.7089,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613175840607428e-06,
+      "loss": 1.7109,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611174632784147e-06,
+      "loss": 1.6033,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609168471150202e-06,
+      "loss": 1.7853,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607157357860823e-06,
+      "loss": 1.6755,
+      "step": 479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605141295076561e-06,
+      "loss": 1.621,
+      "step": 480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603120284963284e-06,
+      "loss": 1.683,
+      "step": 481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601094329692173e-06,
+      "loss": 1.5792,
+      "step": 482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599063431439721e-06,
+      "loss": 1.6919,
+      "step": 483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597027592387739e-06,
+      "loss": 1.6206,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594986814723335e-06,
+      "loss": 1.6596,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59294110063893e-06,
+      "loss": 1.674,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590890452332249e-06,
+      "loss": 1.6804,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588834872006308e-06,
+      "loss": 1.7044,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586774361869436e-06,
+      "loss": 1.7794,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584708924135245e-06,
+      "loss": 1.7797,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582638561022646e-06,
+      "loss": 1.6997,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580563274755848e-06,
+      "loss": 1.6487,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578483067564335e-06,
+      "loss": 1.7352,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576397941682891e-06,
+      "loss": 1.6135,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574307899351574e-06,
+      "loss": 1.6328,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572212942815734e-06,
+      "loss": 1.5449,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570113074325986e-06,
+      "loss": 1.6395,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568008296138238e-06,
+      "loss": 1.7314,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565898610513661e-06,
+      "loss": 1.6972,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563784019718704e-06,
+      "loss": 1.7302,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561664526025082e-06,
+      "loss": 1.822,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55954013170978e-06,
+      "loss": 1.5555,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557410839055047e-06,
+      "loss": 1.6642,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555276650348393e-06,
+      "loss": 1.7137,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55313756788259e-06,
+      "loss": 1.7119,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550993593955665e-06,
+      "loss": 1.6566,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548844730870903e-06,
+      "loss": 1.8257,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546690980936836e-06,
+      "loss": 1.6663,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544532346467254e-06,
+      "loss": 1.6819,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542368829781186e-06,
+      "loss": 1.6969,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540200433202913e-06,
+      "loss": 1.6454,
+      "step": 511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538027159061955e-06,
+      "loss": 1.53,
+      "step": 512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535849009693072e-06,
+      "loss": 1.7466,
+      "step": 513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533665987436262e-06,
+      "loss": 1.6924,
+      "step": 514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531478094636758e-06,
+      "loss": 1.6239,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529285333645027e-06,
+      "loss": 1.8069,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527087706816762e-06,
+      "loss": 1.6793,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524885216512887e-06,
+      "loss": 1.7355,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522677865099548e-06,
+      "loss": 1.6477,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520465654948119e-06,
+      "loss": 1.6219,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518248588435185e-06,
+      "loss": 1.7696,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516026667942557e-06,
+      "loss": 1.6319,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513799895857252e-06,
+      "loss": 1.697,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511568274571508e-06,
+      "loss": 1.6275,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509331806482767e-06,
+      "loss": 1.6908,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507090493993677e-06,
+      "loss": 1.6443,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504844339512096e-06,
+      "loss": 1.5952,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502593345451078e-06,
+      "loss": 1.6992,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500337514228878e-06,
+      "loss": 1.7055,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49807684826895e-06,
+      "loss": 1.8455,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495811349999941e-06,
+      "loss": 1.5761,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493541021855685e-06,
+      "loss": 1.7289,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49126586627521e-06,
+      "loss": 1.7057,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488985885702728e-06,
+      "loss": 1.6906,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486701082587635e-06,
+      "loss": 1.7373,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484411459384508e-06,
+      "loss": 1.5439,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482117018553101e-06,
+      "loss": 1.7382,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479817762558345e-06,
+      "loss": 1.6473,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477513693870347e-06,
+      "loss": 1.5964,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475204814964374e-06,
+      "loss": 1.5634,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472891128320874e-06,
+      "loss": 1.6401,
+      "step": 541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470572636425451e-06,
+      "loss": 1.7356,
+      "step": 542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46824934176887e-06,
+      "loss": 1.7242,
+      "step": 543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465921246847067e-06,
+      "loss": 1.7241,
+      "step": 544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463588354161122e-06,
+      "loss": 1.8465,
+      "step": 545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461250666217277e-06,
+      "loss": 1.6192,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458908185526921e-06,
+      "loss": 1.5597,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456560914606594e-06,
+      "loss": 1.5834,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454208855977986e-06,
+      "loss": 1.5809,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451852012167924e-06,
+      "loss": 1.7136,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449490385708378e-06,
+      "loss": 1.6345,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447123979136457e-06,
+      "loss": 1.6974,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444752794994408e-06,
+      "loss": 1.761,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4423768358296e-06,
+      "loss": 1.5988,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439996104194546e-06,
+      "loss": 1.6949,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437610602646878e-06,
+      "loss": 1.7229,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43522033374935e-06,
+      "loss": 1.602,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432825300069848e-06,
+      "loss": 1.6957,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430425504181361e-06,
+      "loss": 1.6405,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428020948662012e-06,
+      "loss": 1.865,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425611636095023e-06,
+      "loss": 1.6169,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423197569068733e-06,
+      "loss": 1.5917,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420778750176588e-06,
+      "loss": 1.6931,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418355182017138e-06,
+      "loss": 1.6626,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41592686719404e-06,
+      "loss": 1.7617,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413493808316038e-06,
+      "loss": 1.6361,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411056007996989e-06,
+      "loss": 1.7683,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408613468855829e-06,
+      "loss": 1.6592,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406166193516596e-06,
+      "loss": 1.7436,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403714184608411e-06,
+      "loss": 1.613,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40125744476548e-06,
+      "loss": 1.7193,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398795976627091e-06,
+      "loss": 1.6348,
+      "step": 572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396329782837614e-06,
+      "loss": 1.5797,
+      "step": 573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393858866046494e-06,
+      "loss": 1.6836,
+      "step": 574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391383228908253e-06,
+      "loss": 1.693,
+      "step": 575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388902874082482e-06,
+      "loss": 1.6932,
+      "step": 576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386417804233836e-06,
+      "loss": 1.7571,
+      "step": 577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.383928022032044e-06,
+      "loss": 1.7845,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381433530151887e-06,
+      "loss": 1.655,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37893433127322e-06,
+      "loss": 1.6516,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376430428080939e-06,
+      "loss": 1.6923,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373921823265004e-06,
+      "loss": 1.6464,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371408519520421e-06,
+      "loss": 1.6161,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36889051954725e-06,
+      "loss": 1.5796,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366367826050593e-06,
+      "loss": 1.6534,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36384044174059e-06,
+      "loss": 1.757,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361308369332426e-06,
+      "loss": 1.6041,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358771611546319e-06,
+      "loss": 1.644,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356230171107524e-06,
+      "loss": 1.7657,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353684050746323e-06,
+      "loss": 1.6517,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351133253198027e-06,
+      "loss": 1.6702,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348577781202976e-06,
+      "loss": 1.6703,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.346017637506523e-06,
+      "loss": 1.5987,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343452824859048e-06,
+      "loss": 1.9405,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340883346015941e-06,
+      "loss": 1.7368,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338309203737609e-06,
+      "loss": 1.6937,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335730400789466e-06,
+      "loss": 1.6786,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333146939941938e-06,
+      "loss": 1.6284,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330558823970448e-06,
+      "loss": 1.7631,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327966055655424e-06,
+      "loss": 1.7365,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325368637782292e-06,
+      "loss": 1.6349,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322766573141473e-06,
+      "loss": 1.6427,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320159864528378e-06,
+      "loss": 1.6622,
+      "step": 603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31754851474341e-06,
+      "loss": 1.5653,
+      "step": 604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314932526591956e-06,
+      "loss": 1.5644,
+      "step": 605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312311902884388e-06,
+      "loss": 1.5908,
+      "step": 606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309686646436053e-06,
+      "loss": 1.6398,
+      "step": 607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307056760067284e-06,
+      "loss": 1.7208,
+      "step": 608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30442224660338e-06,
+      "loss": 1.6371,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301783108874611e-06,
+      "loss": 1.732,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299139349716221e-06,
+      "loss": 1.7115,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296490971968416e-06,
+      "loss": 1.6152,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293837978476359e-06,
+      "loss": 1.6696,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291180372090178e-06,
+      "loss": 1.5402,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288518155664956e-06,
+      "loss": 1.7045,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285851332060722e-06,
+      "loss": 1.6317,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283179904142465e-06,
+      "loss": 1.6823,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280503874780112e-06,
+      "loss": 1.7649,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277823246848537e-06,
+      "loss": 1.7668,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275138023227555e-06,
+      "loss": 1.6933,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272448206801912e-06,
+      "loss": 1.5835,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269753800461299e-06,
+      "loss": 1.6333,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267054807100327e-06,
+      "loss": 1.603,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264351229618541e-06,
+      "loss": 1.7097,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261643070920409e-06,
+      "loss": 1.5347,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.258930333915325e-06,
+      "loss": 1.5401,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256213021517593e-06,
+      "loss": 1.6538,
+      "step": 627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253491136646437e-06,
+      "loss": 1.7349,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250764682225997e-06,
+      "loss": 1.7587,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248033661185313e-06,
+      "loss": 1.7165,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24529807645834e-06,
+      "loss": 1.6709,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24255793098393e-06,
+      "loss": 1.7697,
+      "step": 632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23981322770584e-06,
+      "loss": 1.646,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237063969572713e-06,
+      "loss": 1.6591,
+      "step": 634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2343101595381e-06,
+      "loss": 1.7666,
+      "step": 635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23155180056043e-06,
+      "loss": 1.6118,
+      "step": 636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228788895603024e-06,
+      "loss": 1.7188,
+      "step": 637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226021447634085e-06,
+      "loss": 1.6766,
+      "step": 638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223249459626704e-06,
+      "loss": 1.756,
+      "step": 639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220472934558838e-06,
+      "loss": 1.591,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217691875413323e-06,
+      "loss": 1.6234,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214906285177867e-06,
+      "loss": 1.5921,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212116166845048e-06,
+      "loss": 1.6858,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209321523412303e-06,
+      "loss": 1.7072,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206522357881931e-06,
+      "loss": 1.6384,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203718673261098e-06,
+      "loss": 1.6046,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20091047256181e-06,
+      "loss": 1.7566,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198097758800938e-06,
+      "loss": 1.6069,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195280535000196e-06,
+      "loss": 1.6106,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19245880418614e-06,
+      "loss": 1.6747,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189632569390172e-06,
+      "loss": 1.6496,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186801833648535e-06,
+      "loss": 1.6808,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183966600002301e-06,
+      "loss": 1.6754,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181126871497378e-06,
+      "loss": 1.6842,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.178282651184506e-06,
+      "loss": 1.7304,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175433942119238e-06,
+      "loss": 1.5301,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172580747361968e-06,
+      "loss": 1.6294,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169723069977892e-06,
+      "loss": 1.6115,
+      "step": 658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166860913037032e-06,
+      "loss": 1.6056,
+      "step": 659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163994279614218e-06,
+      "loss": 1.6914,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161123172789091e-06,
+      "loss": 1.6928,
+      "step": 661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158247595646098e-06,
+      "loss": 1.6577,
+      "step": 662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155367551274485e-06,
+      "loss": 1.6872,
+      "step": 663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152483042768302e-06,
+      "loss": 1.6102,
+      "step": 664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149594073226391e-06,
+      "loss": 1.7545,
+      "step": 665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14670064575239e-06,
+      "loss": 1.6717,
+      "step": 666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143802763454723e-06,
+      "loss": 1.6195,
+      "step": 667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140900429446601e-06,
+      "loss": 1.6969,
+      "step": 668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137993646846018e-06,
+      "loss": 1.5573,
+      "step": 669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135082418775746e-06,
+      "loss": 1.5481,
+      "step": 670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132166748363335e-06,
+      "loss": 1.7011,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.129246638741108e-06,
+      "loss": 1.6015,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126322093046149e-06,
+      "loss": 1.7264,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123393114420318e-06,
+      "loss": 1.7024,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120459706010233e-06,
+      "loss": 1.7152,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11752187096727e-06,
+      "loss": 1.667,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114579612447562e-06,
+      "loss": 1.7626,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111632933611993e-06,
+      "loss": 1.5775,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108681837626199e-06,
+      "loss": 1.7652,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105726327660556e-06,
+      "loss": 1.6548,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102766406890185e-06,
+      "loss": 1.6851,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099802078494947e-06,
+      "loss": 1.6092,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.096833345659437e-06,
+      "loss": 1.6174,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09386021157298e-06,
+      "loss": 1.7192,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09088267942963e-06,
+      "loss": 1.6354,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087900752428168e-06,
+      "loss": 1.5762,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.084914433772094e-06,
+      "loss": 1.6651,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081923726669626e-06,
+      "loss": 1.5661,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0789286343337e-06,
+      "loss": 1.5861,
+      "step": 689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075929159981957e-06,
+      "loss": 1.675,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072925306836751e-06,
+      "loss": 1.6076,
+      "step": 691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06991707812514e-06,
+      "loss": 1.6368,
+      "step": 692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066904477078875e-06,
+      "loss": 1.7312,
+      "step": 693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063887506934417e-06,
+      "loss": 1.5562,
+      "step": 694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06086617093291e-06,
+      "loss": 1.7073,
+      "step": 695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057840472320192e-06,
+      "loss": 1.6598,
+      "step": 696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054810414346789e-06,
+      "loss": 1.52,
+      "step": 697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05177600026791e-06,
+      "loss": 1.6168,
+      "step": 698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048737233343442e-06,
+      "loss": 1.691,
+      "step": 699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045694116837948e-06,
+      "loss": 1.6401,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042646654020667e-06,
+      "loss": 1.5933,
+      "step": 701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039594848165507e-06,
+      "loss": 1.7813,
+      "step": 702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.036538702551037e-06,
+      "loss": 1.6201,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03347822046049e-06,
+      "loss": 1.6496,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03041340518176e-06,
+      "loss": 1.7,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027344260007401e-06,
+      "loss": 1.5947,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024270788234606e-06,
+      "loss": 1.6787,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021192993165224e-06,
+      "loss": 1.6247,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01811087810575e-06,
+      "loss": 1.6361,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015024446367315e-06,
+      "loss": 1.6269,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01193370126569e-06,
+      "loss": 1.5569,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008838646121282e-06,
+      "loss": 1.6577,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005739284259123e-06,
+      "loss": 1.5408,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002635619008877e-06,
+      "loss": 1.6749,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999527653704829e-06,
+      "loss": 1.6086,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996415391685882e-06,
+      "loss": 1.647,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993298836295556e-06,
+      "loss": 1.6982,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990177990881986e-06,
+      "loss": 1.6503,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987052858797914e-06,
+      "loss": 1.6984,
+      "step": 719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983923443400682e-06,
+      "loss": 1.6863,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980789748052245e-06,
+      "loss": 1.7796,
+      "step": 721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977651776119145e-06,
+      "loss": 1.7507,
+      "step": 722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974509530972523e-06,
+      "loss": 1.6957,
+      "step": 723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971363015988115e-06,
+      "loss": 1.6172,
+      "step": 724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968212234546235e-06,
+      "loss": 1.5231,
+      "step": 725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965057190031785e-06,
+      "loss": 1.6761,
+      "step": 726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961897885834247e-06,
+      "loss": 1.4835,
+      "step": 727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958734325347684e-06,
+      "loss": 1.7527,
+      "step": 728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955566511970721e-06,
+      "loss": 1.6619,
+      "step": 729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95239444910656e-06,
+      "loss": 1.6382,
+      "step": 730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949218140162965e-06,
+      "loss": 1.5619,
+      "step": 731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946037588552266e-06,
+      "loss": 1.6889,
+      "step": 732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94285279769134e-06,
+      "loss": 1.6706,
+      "step": 733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939663771001632e-06,
+      "loss": 1.6436,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93647051190913e-06,
+      "loss": 1.6462,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93327302384437e-06,
+      "loss": 1.727,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930071310242429e-06,
+      "loss": 1.5808,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926865374542928e-06,
+      "loss": 1.7614,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92365522019002e-06,
+      "loss": 1.6766,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920440850632395e-06,
+      "loss": 1.6317,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917222269323263e-06,
+      "loss": 1.57,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91399947972037e-06,
+      "loss": 1.5668,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91077248528597e-06,
+      "loss": 1.6273,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907541289486847e-06,
+      "loss": 1.5773,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904305895794292e-06,
+      "loss": 1.6244,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901066307684102e-06,
+      "loss": 1.7234,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89782252863659e-06,
+      "loss": 1.6636,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894574562136561e-06,
+      "loss": 1.6889,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89132241167333e-06,
+      "loss": 1.5958,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888066080740692e-06,
+      "loss": 1.683,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88480557283695e-06,
+      "loss": 1.6079,
+      "step": 751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88154089146488e-06,
+      "loss": 1.5908,
+      "step": 752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.878272040131748e-06,
+      "loss": 1.5926,
+      "step": 753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874999022349303e-06,
+      "loss": 1.5953,
+      "step": 754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871721841633762e-06,
+      "loss": 1.6405,
+      "step": 755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868440501505822e-06,
+      "loss": 1.5675,
+      "step": 756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865155005490643e-06,
+      "loss": 1.6822,
+      "step": 757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861865357117852e-06,
+      "loss": 1.6775,
+      "step": 758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858571559921539e-06,
+      "loss": 1.4975,
+      "step": 759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855273617440243e-06,
+      "loss": 1.6837,
+      "step": 760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851971533216968e-06,
+      "loss": 1.7066,
+      "step": 761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848665310799156e-06,
+      "loss": 1.6112,
+      "step": 762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845354953738706e-06,
+      "loss": 1.6492,
+      "step": 763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84204046559195e-06,
+      "loss": 1.6053,
+      "step": 764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.83872184991966e-06,
+      "loss": 1.6591,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835399110287046e-06,
+      "loss": 1.6255,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832072250263746e-06,
+      "loss": 1.6519,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82874127342382e-06,
+      "loss": 1.6764,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82540618334576e-06,
+      "loss": 1.566,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82206698361247e-06,
+      "loss": 1.7474,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818723677811269e-06,
+      "loss": 1.6626,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815376269533893e-06,
+      "loss": 1.6228,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812024762376477e-06,
+      "loss": 1.5379,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808669159939568e-06,
+      "loss": 1.6888,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.805309465828105e-06,
+      "loss": 1.5877,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80194568365143e-06,
+      "loss": 1.7083,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798577817023269e-06,
+      "loss": 1.6247,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795205869561742e-06,
+      "loss": 1.5508,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79182984488935e-06,
+      "loss": 1.5436,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788449746632976e-06,
+      "loss": 1.6403,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78506557842388e-06,
+      "loss": 1.5443,
+      "step": 781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781677343897687e-06,
+      "loss": 1.6498,
+      "step": 782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778285046694403e-06,
+      "loss": 1.5833,
+      "step": 783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77488869045839e-06,
+      "loss": 1.6773,
+      "step": 784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771488278838368e-06,
+      "loss": 1.7334,
+      "step": 785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768083815487428e-06,
+      "loss": 1.5927,
+      "step": 786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764675304062992e-06,
+      "loss": 1.6573,
+      "step": 787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76126274822685e-06,
+      "loss": 1.6621,
+      "step": 788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75784615164513e-06,
+      "loss": 1.6251,
+      "step": 789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754425517988298e-06,
+      "loss": 1.7836,
+      "step": 790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751000850931162e-06,
+      "loss": 1.5745,
+      "step": 791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74757215415286e-06,
+      "loss": 1.6233,
+      "step": 792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74413943133686e-06,
+      "loss": 1.7662,
+      "step": 793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740702686170955e-06,
+      "loss": 1.6231,
+      "step": 794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73726192234726e-06,
+      "loss": 1.6739,
+      "step": 795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733817143562207e-06,
+      "loss": 1.4668,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.73036835351654e-06,
+      "loss": 1.7034,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726915555915317e-06,
+      "loss": 1.5549,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723458754467893e-06,
+      "loss": 1.6249,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719997952887932e-06,
+      "loss": 1.6887,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71653315489339e-06,
+      "loss": 1.6135,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71306436420652e-06,
+      "loss": 1.6517,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709591584553865e-06,
+      "loss": 1.6234,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706114819666249e-06,
+      "loss": 1.5262,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702634073278784e-06,
+      "loss": 1.6573,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699149349130848e-06,
+      "loss": 1.6839,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695660650966109e-06,
+      "loss": 1.6135,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692167982532487e-06,
+      "loss": 1.6722,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688671347582178e-06,
+      "loss": 1.7136,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685170749871638e-06,
+      "loss": 1.6885,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681666193161578e-06,
+      "loss": 1.6717,
+      "step": 811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67815768121696e-06,
+      "loss": 1.6655,
+      "step": 812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.674645217807e-06,
+      "loss": 1.5274,
+      "step": 813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.671128806705159e-06,
+      "loss": 1.5268,
+      "step": 814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667608451689135e-06,
+      "loss": 1.5713,
+      "step": 815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664084156540864e-06,
+      "loss": 1.5606,
+      "step": 816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660555925046518e-06,
+      "loss": 1.7347,
+      "step": 817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657023760996497e-06,
+      "loss": 1.5717,
+      "step": 818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653487668185419e-06,
+      "loss": 1.6771,
+      "step": 819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649947650412135e-06,
+      "loss": 1.5636,
+      "step": 820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646403711479702e-06,
+      "loss": 1.5827,
+      "step": 821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642855855195394e-06,
+      "loss": 1.6808,
+      "step": 822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639304085370692e-06,
+      "loss": 1.6476,
+      "step": 823
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.635748405821285e-06,
+      "loss": 1.636,
+      "step": 824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632188820367056e-06,
+      "loss": 1.7191,
+      "step": 825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62862533283209e-06,
+      "loss": 1.62,
+      "step": 826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625057947044662e-06,
+      "loss": 1.5557,
+      "step": 827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62148666683723e-06,
+      "loss": 1.7489,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617911496046446e-06,
+      "loss": 1.6966,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614332438513132e-06,
+      "loss": 1.7291,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610749498082291e-06,
+      "loss": 1.6475,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607162678603097e-06,
+      "loss": 1.5812,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603571983928888e-06,
+      "loss": 1.5562,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599977417917169e-06,
+      "loss": 1.7772,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5963789844296e-06,
+      "loss": 1.577,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592776687332003e-06,
+      "loss": 1.6907,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58917053049434e-06,
+      "loss": 1.593,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58556051779073e-06,
+      "loss": 1.6642,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581946653099427e-06,
+      "loss": 1.6248,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.578328940302827e-06,
+      "loss": 1.6203,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574707383287459e-06,
+      "loss": 1.6054,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571081985943984e-06,
+      "loss": 1.6978,
+      "step": 842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567452752167183e-06,
+      "loss": 1.5844,
+      "step": 843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563819685855963e-06,
+      "loss": 1.5951,
+      "step": 844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.560182790913349e-06,
+      "loss": 1.6168,
+      "step": 845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556542071246476e-06,
+      "loss": 1.6609,
+      "step": 846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552897530766592e-06,
+      "loss": 1.6718,
+      "step": 847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549249173389045e-06,
+      "loss": 1.5932,
+      "step": 848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545597003033286e-06,
+      "loss": 1.5592,
+      "step": 849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54194102362286e-06,
+      "loss": 1.6942,
+      "step": 850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.538281239085411e-06,
+      "loss": 1.642,
+      "step": 851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534617653352661e-06,
+      "loss": 1.7223,
+      "step": 852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530950270360425e-06,
+      "loss": 1.6581,
+      "step": 853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52727909404859e-06,
+      "loss": 1.6021,
+      "step": 854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.523604128361123e-06,
+      "loss": 1.6454,
+      "step": 855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519925377246057e-06,
+      "loss": 1.6562,
+      "step": 856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516242844655498e-06,
+      "loss": 1.5912,
+      "step": 857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512556534545612e-06,
+      "loss": 1.7059,
+      "step": 858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.50886645087662e-06,
+      "loss": 1.5695,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.5051725976128e-06,
+      "loss": 1.6129,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50147497872248e-06,
+      "loss": 1.7793,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.497773598178033e-06,
+      "loss": 1.7364,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494068459955871e-06,
+      "loss": 1.5904,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490359568036446e-06,
+      "loss": 1.6493,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486646926404243e-06,
+      "loss": 1.6261,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48293053904777e-06,
+      "loss": 1.5795,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479210409959565e-06,
+      "loss": 1.6833,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.475486543136181e-06,
+      "loss": 1.6581,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471758942578193e-06,
+      "loss": 1.6373,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46802761229018e-06,
+      "loss": 1.5906,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464292556280734e-06,
+      "loss": 1.6514,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46055377856244e-06,
+      "loss": 1.5939,
+      "step": 872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456811283151896e-06,
+      "loss": 1.6573,
+      "step": 873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453065074069682e-06,
+      "loss": 1.6402,
+      "step": 874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449315155340369e-06,
+      "loss": 1.7516,
+      "step": 875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44556153099252e-06,
+      "loss": 1.6742,
+      "step": 876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441804205058672e-06,
+      "loss": 1.6514,
+      "step": 877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43804318157534e-06,
+      "loss": 1.6512,
+      "step": 878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434278464583018e-06,
+      "loss": 1.4881,
+      "step": 879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430510058126156e-06,
+      "loss": 1.6386,
+      "step": 880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426737966253176e-06,
+      "loss": 1.8296,
+      "step": 881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422962193016459e-06,
+      "loss": 1.711,
+      "step": 882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41918274247234e-06,
+      "loss": 1.5452,
+      "step": 883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.415399618681101e-06,
+      "loss": 1.6361,
+      "step": 884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411612825706976e-06,
+      "loss": 1.6508,
+      "step": 885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.407822367618135e-06,
+      "loss": 1.6964,
+      "step": 886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40402824848669e-06,
+      "loss": 1.5838,
+      "step": 887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400230472388684e-06,
+      "loss": 1.8096,
+      "step": 888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396429043404088e-06,
+      "loss": 1.6443,
+      "step": 889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.3926239656168e-06,
+      "loss": 1.7326,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388815243114637e-06,
+      "loss": 1.7111,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.385002879989328e-06,
+      "loss": 1.6244,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381186880336518e-06,
+      "loss": 1.7512,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377367248255757e-06,
+      "loss": 1.7644,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.373543987850494e-06,
+      "loss": 1.6964,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.369717103228084e-06,
+      "loss": 1.6417,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.365886598499766e-06,
+      "loss": 1.5767,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.362052477780677e-06,
+      "loss": 1.5865,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35821474518983e-06,
+      "loss": 1.656,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354373404850124e-06,
+      "loss": 1.7115,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.350528460888334e-06,
+      "loss": 1.7248,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.346679917435104e-06,
+      "loss": 1.5919,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.342827778624943e-06,
+      "loss": 1.5749,
+      "step": 903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33897204859623e-06,
+      "loss": 1.76,
+      "step": 904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335112731491192e-06,
+      "loss": 1.6716,
+      "step": 905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.331249831455921e-06,
+      "loss": 1.6643,
+      "step": 906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.327383352640347e-06,
+      "loss": 1.5616,
+      "step": 907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323513299198252e-06,
+      "loss": 1.6957,
+      "step": 908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319639675287255e-06,
+      "loss": 1.5866,
+      "step": 909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315762485068815e-06,
+      "loss": 1.6034,
+      "step": 910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311881732708213e-06,
+      "loss": 1.6345,
+      "step": 911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307997422374569e-06,
+      "loss": 1.6969,
+      "step": 912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304109558240817e-06,
+      "loss": 1.6597,
+      "step": 913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300218144483709e-06,
+      "loss": 1.5902,
+      "step": 914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.296323185283816e-06,
+      "loss": 1.6701,
+      "step": 915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292424684825514e-06,
+      "loss": 1.6232,
+      "step": 916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28852264729698e-06,
+      "loss": 1.6822,
+      "step": 917
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284617076890199e-06,
+      "loss": 1.6393,
+      "step": 918
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280707977800944e-06,
+      "loss": 1.7256,
+      "step": 919
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.276795354228785e-06,
+      "loss": 1.6975,
+      "step": 920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272879210377074e-06,
+      "loss": 1.6948,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268959550452946e-06,
+      "loss": 1.7446,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265036378667312e-06,
+      "loss": 1.6269,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261109699234862e-06,
+      "loss": 1.6505,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257179516374045e-06,
+      "loss": 1.7437,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.253245834307079e-06,
+      "loss": 1.7447,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249308657259943e-06,
+      "loss": 1.7016,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245367989462368e-06,
+      "loss": 1.7069,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241423835147833e-06,
+      "loss": 1.5923,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237476198553567e-06,
+      "loss": 1.6333,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233525083920536e-06,
+      "loss": 1.5749,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229570495493447e-06,
+      "loss": 1.6004,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225612437520736e-06,
+      "loss": 1.6613,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.221650914254566e-06,
+      "loss": 1.6855,
+      "step": 934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.217685929950823e-06,
+      "loss": 1.7077,
+      "step": 935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.213717488869113e-06,
+      "loss": 1.5797,
+      "step": 936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209745595272755e-06,
+      "loss": 1.7837,
+      "step": 937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205770253428775e-06,
+      "loss": 1.6896,
+      "step": 938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201791467607905e-06,
+      "loss": 1.6324,
+      "step": 939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197809242084575e-06,
+      "loss": 1.6203,
+      "step": 940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.193823581136919e-06,
+      "loss": 1.665,
+      "step": 941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189834489046746e-06,
+      "loss": 1.5472,
+      "step": 942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185841970099566e-06,
+      "loss": 1.6049,
+      "step": 943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.181846028584563e-06,
+      "loss": 1.6212,
+      "step": 944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177846668794598e-06,
+      "loss": 1.6824,
+      "step": 945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173843895026207e-06,
+      "loss": 1.6394,
+      "step": 946
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.169837711579591e-06,
+      "loss": 1.6075,
+      "step": 947
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.165828122758615e-06,
+      "loss": 1.7049,
+      "step": 948
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161815132870806e-06,
+      "loss": 1.6557,
+      "step": 949
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157798746227337e-06,
+      "loss": 1.6246,
+      "step": 950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153778967143035e-06,
+      "loss": 1.6691,
+      "step": 951
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149755799936377e-06,
+      "loss": 1.56,
+      "step": 952
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.145729248929466e-06,
+      "loss": 1.5636,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.141699318448053e-06,
+      "loss": 1.6683,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137666012821514e-06,
+      "loss": 1.5143,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13362933638285e-06,
+      "loss": 1.6105,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.129589293468689e-06,
+      "loss": 1.5265,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125545888419269e-06,
+      "loss": 1.6145,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12149912557844e-06,
+      "loss": 1.6604,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.117449009293668e-06,
+      "loss": 1.6113,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.113395543916012e-06,
+      "loss": 1.6889,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.109338733800132e-06,
+      "loss": 1.5489,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10527858330428e-06,
+      "loss": 1.6033,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101215096790305e-06,
+      "loss": 1.5472,
+      "step": 964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.097148278623628e-06,
+      "loss": 1.4815,
+      "step": 965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093078133173256e-06,
+      "loss": 1.5917,
+      "step": 966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089004664811767e-06,
+      "loss": 1.5097,
+      "step": 967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084927877915314e-06,
+      "loss": 1.6948,
+      "step": 968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.080847776863609e-06,
+      "loss": 1.696,
+      "step": 969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07676436603993e-06,
+      "loss": 1.6118,
+      "step": 970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072677649831107e-06,
+      "loss": 1.6456,
+      "step": 971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068587632627521e-06,
+      "loss": 1.6943,
+      "step": 972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.064494318823102e-06,
+      "loss": 1.5301,
+      "step": 973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060397712815318e-06,
+      "loss": 1.6907,
+      "step": 974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056297819005177e-06,
+      "loss": 1.6447,
+      "step": 975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052194641797217e-06,
+      "loss": 1.6076,
+      "step": 976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048088185599507e-06,
+      "loss": 1.7586,
+      "step": 977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043978454823632e-06,
+      "loss": 1.6094,
+      "step": 978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.0398654538847e-06,
+      "loss": 1.6471,
+      "step": 979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.035749187201333e-06,
+      "loss": 1.6513,
+      "step": 980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031629659195657e-06,
+      "loss": 1.6248,
+      "step": 981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027506874293304e-06,
+      "loss": 1.6602,
+      "step": 982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.023380836923404e-06,
+      "loss": 1.6221,
+      "step": 983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019251551518585e-06,
+      "loss": 1.5549,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.015119022514958e-06,
+      "loss": 1.7109,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010983254352127e-06,
+      "loss": 1.6258,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.006844251473165e-06,
+      "loss": 1.647,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002702018324629e-06,
+      "loss": 1.6241,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.998556559356543e-06,
+      "loss": 1.623,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994407879022397e-06,
+      "loss": 1.6229,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.990255981779139e-06,
+      "loss": 1.6345,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.986100872087177e-06,
+      "loss": 1.5853,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981942554410371e-06,
+      "loss": 1.6577,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97778103321602e-06,
+      "loss": 1.6058,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973616312974876e-06,
+      "loss": 1.498,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.969448398161115e-06,
+      "loss": 1.6153,
+      "step": 996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.965277293252354e-06,
+      "loss": 1.6881,
+      "step": 997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961103002729634e-06,
+      "loss": 1.6443,
+      "step": 998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.956925531077417e-06,
+      "loss": 1.6765,
+      "step": 999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.952744882783587e-06,
+      "loss": 1.6739,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948561062339435e-06,
+      "loss": 1.5991,
+      "step": 1001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944374074239665e-06,
+      "loss": 1.6868,
+      "step": 1002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.940183922982381e-06,
+      "loss": 1.6218,
+      "step": 1003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.935990613069087e-06,
+      "loss": 1.5527,
+      "step": 1004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.931794149004675e-06,
+      "loss": 1.6541,
+      "step": 1005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.927594535297433e-06,
+      "loss": 1.6607,
+      "step": 1006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923391776459031e-06,
+      "loss": 1.6174,
+      "step": 1007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.919185877004515e-06,
+      "loss": 1.6709,
+      "step": 1008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914976841452304e-06,
+      "loss": 1.5432,
+      "step": 1009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91076467432419e-06,
+      "loss": 1.6759,
+      "step": 1010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90654938014533e-06,
+      "loss": 1.654,
+      "step": 1011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.902330963444234e-06,
+      "loss": 1.6087,
+      "step": 1012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898109428752773e-06,
+      "loss": 1.7073,
+      "step": 1013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.893884780606164e-06,
+      "loss": 1.7368,
+      "step": 1014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889657023542973e-06,
+      "loss": 1.659,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885426162105101e-06,
+      "loss": 1.6443,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881192200837785e-06,
+      "loss": 1.7409,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876955144289594e-06,
+      "loss": 1.7426,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.872714997012421e-06,
+      "loss": 1.6249,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.868471763561482e-06,
+      "loss": 1.5635,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.864225448495304e-06,
+      "loss": 1.6441,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85997605637573e-06,
+      "loss": 1.5241,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.855723591767903e-06,
+      "loss": 1.5956,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85146805924027e-06,
+      "loss": 1.657,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847209463364574e-06,
+      "loss": 1.4867,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.842947808715848e-06,
+      "loss": 1.7553,
+      "step": 1026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83868309987241e-06,
+      "loss": 1.5502,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834415341415862e-06,
+      "loss": 1.7817,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.830144537931082e-06,
+      "loss": 1.7541,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.825870694006217e-06,
+      "loss": 1.665,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82159381423268e-06,
+      "loss": 1.6158,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817313903205148e-06,
+      "loss": 1.688,
+      "step": 1032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813030965521554e-06,
+      "loss": 1.74,
+      "step": 1033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80874500578308e-06,
+      "loss": 1.5366,
+      "step": 1034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804456028594158e-06,
+      "loss": 1.6822,
+      "step": 1035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80016403856246e-06,
+      "loss": 1.6723,
+      "step": 1036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.795869040298895e-06,
+      "loss": 1.7941,
+      "step": 1037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.791571038417602e-06,
+      "loss": 1.673,
+      "step": 1038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78727003753595e-06,
+      "loss": 1.5728,
+      "step": 1039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.782966042274529e-06,
+      "loss": 1.6606,
+      "step": 1040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.778659057257144e-06,
+      "loss": 1.7012,
+      "step": 1041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.774349087110813e-06,
+      "loss": 1.637,
+      "step": 1042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77003613646576e-06,
+      "loss": 1.6424,
+      "step": 1043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.765720209955414e-06,
+      "loss": 1.7211,
+      "step": 1044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761401312216398e-06,
+      "loss": 1.6917,
+      "step": 1045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757079447888529e-06,
+      "loss": 1.6091,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.752754621614807e-06,
+      "loss": 1.7085,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748426838041421e-06,
+      "loss": 1.6089,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.744096101817731e-06,
+      "loss": 1.5918,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73976241759627e-06,
+      "loss": 1.6516,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73542579003274e-06,
+      "loss": 1.7994,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.731086223786006e-06,
+      "loss": 1.5809,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.726743723518087e-06,
+      "loss": 1.6535,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.722398293894153e-06,
+      "loss": 1.7241,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.718049939582529e-06,
+      "loss": 1.6618,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713698665254669e-06,
+      "loss": 1.5355,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70934447558518e-06,
+      "loss": 1.7046,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.704987375251782e-06,
+      "loss": 1.5727,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70062736893534e-06,
+      "loss": 1.6436,
+      "step": 1059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.696264461319831e-06,
+      "loss": 1.5915,
+      "step": 1060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69189865709235e-06,
+      "loss": 1.6802,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687529960943107e-06,
+      "loss": 1.5917,
+      "step": 1062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.683158377565415e-06,
+      "loss": 1.6565,
+      "step": 1063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.678783911655691e-06,
+      "loss": 1.669,
+      "step": 1064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.674406567913447e-06,
+      "loss": 1.6652,
+      "step": 1065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67002635104129e-06,
+      "loss": 1.5721,
+      "step": 1066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66564326574491e-06,
+      "loss": 1.6099,
+      "step": 1067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.661257316733078e-06,
+      "loss": 1.6212,
+      "step": 1068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656868508717648e-06,
+      "loss": 1.5464,
+      "step": 1069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.652476846413537e-06,
+      "loss": 1.7025,
+      "step": 1070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.648082334538735e-06,
+      "loss": 1.7083,
+      "step": 1071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64368497781429e-06,
+      "loss": 1.6088,
+      "step": 1072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.639284780964307e-06,
+      "loss": 1.6713,
+      "step": 1073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634881748715941e-06,
+      "loss": 1.6539,
+      "step": 1074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630475885799395e-06,
+      "loss": 1.6635,
+      "step": 1075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.626067196947913e-06,
+      "loss": 1.6656,
+      "step": 1076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621655686897771e-06,
+      "loss": 1.6201,
+      "step": 1077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.617241360388282e-06,
+      "loss": 1.6279,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612824222161781e-06,
+      "loss": 1.7349,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608404276963623e-06,
+      "loss": 1.5416,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60398152954218e-06,
+      "loss": 1.5206,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599555984648836e-06,
+      "loss": 1.637,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595127647037976e-06,
+      "loss": 1.6668,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.590696521466992e-06,
+      "loss": 1.5278,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.586262612696263e-06,
+      "loss": 1.589,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5818259254891614e-06,
+      "loss": 1.5195,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577386464612049e-06,
+      "loss": 1.6853,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.572944234834261e-06,
+      "loss": 1.5993,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568499240928109e-06,
+      "loss": 1.6131,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5640514876688765e-06,
+      "loss": 1.6534,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559600979834809e-06,
+      "loss": 1.6009,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.555147722207111e-06,
+      "loss": 1.5805,
+      "step": 1092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.550691719569944e-06,
+      "loss": 1.7094,
+      "step": 1093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546232976710413e-06,
+      "loss": 1.6287,
+      "step": 1094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541771498418575e-06,
+      "loss": 1.7723,
+      "step": 1095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537307289487419e-06,
+      "loss": 1.6064,
+      "step": 1096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532840354712868e-06,
+      "loss": 1.5458,
+      "step": 1097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5283706988937765e-06,
+      "loss": 1.5997,
+      "step": 1098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.523898326831921e-06,
+      "loss": 1.6022,
+      "step": 1099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5194232433319955e-06,
+      "loss": 1.6143,
+      "step": 1100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.514945453201608e-06,
+      "loss": 1.7358,
+      "step": 1101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.510464961251271e-06,
+      "loss": 1.5876,
+      "step": 1102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505981772294404e-06,
+      "loss": 1.6603,
+      "step": 1103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.501495891147322e-06,
+      "loss": 1.6798,
+      "step": 1104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.497007322629231e-06,
+      "loss": 1.6493,
+      "step": 1105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492516071562226e-06,
+      "loss": 1.6369,
+      "step": 1106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488022142771282e-06,
+      "loss": 1.5248,
+      "step": 1107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483525541084253e-06,
+      "loss": 1.688,
+      "step": 1108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.479026271331864e-06,
+      "loss": 1.6389,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4745243383477055e-06,
+      "loss": 1.4062,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470019746968226e-06,
+      "loss": 1.683,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4655125020327376e-06,
+      "loss": 1.7601,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.461002608383396e-06,
+      "loss": 1.6138,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.456490070865206e-06,
+      "loss": 1.6043,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4519748943260126e-06,
+      "loss": 1.6143,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447457083616494e-06,
+      "loss": 1.781,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.44293664359016e-06,
+      "loss": 1.5903,
+      "step": 1117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.438413579103344e-06,
+      "loss": 1.5881,
+      "step": 1118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.433887895015199e-06,
+      "loss": 1.5927,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.429359596187694e-06,
+      "loss": 1.6153,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424828687485606e-06,
+      "loss": 1.6999,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.420295173776515e-06,
+      "loss": 1.7169,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.415759059930799e-06,
+      "loss": 1.464,
+      "step": 1123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411220350821631e-06,
+      "loss": 1.6429,
+      "step": 1124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.406679051324972e-06,
+      "loss": 1.5758,
+      "step": 1125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.402135166319567e-06,
+      "loss": 1.5826,
+      "step": 1126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397588700686933e-06,
+      "loss": 1.6571,
+      "step": 1127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.393039659311366e-06,
+      "loss": 1.5808,
+      "step": 1128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388488047079927e-06,
+      "loss": 1.6669,
+      "step": 1129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383933868882438e-06,
+      "loss": 1.572,
+      "step": 1130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.379377129611478e-06,
+      "loss": 1.7202,
+      "step": 1131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374817834162378e-06,
+      "loss": 1.6525,
+      "step": 1132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3702559874332125e-06,
+      "loss": 1.6816,
+      "step": 1133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3656915943247984e-06,
+      "loss": 1.7594,
+      "step": 1134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3611246597406925e-06,
+      "loss": 1.6414,
+      "step": 1135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356555188587178e-06,
+      "loss": 1.6813,
+      "step": 1136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.351983185773259e-06,
+      "loss": 1.5804,
+      "step": 1137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347408656210666e-06,
+      "loss": 1.7449,
+      "step": 1138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342831604813844e-06,
+      "loss": 1.7535,
+      "step": 1139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338252036499941e-06,
+      "loss": 1.5803,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333669956188815e-06,
+      "loss": 1.7141,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3290853688030196e-06,
+      "loss": 1.657,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324498279267803e-06,
+      "loss": 1.6082,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319908692511103e-06,
+      "loss": 1.6718,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315316613463535e-06,
+      "loss": 1.663,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.310722047058396e-06,
+      "loss": 1.6963,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.306124998231655e-06,
+      "loss": 1.6242,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.301525471921949e-06,
+      "loss": 1.7502,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.296923473070571e-06,
+      "loss": 1.5809,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292319006621477e-06,
+      "loss": 1.7039,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2877120775212685e-06,
+      "loss": 1.5961,
+      "step": 1151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.283102690719198e-06,
+      "loss": 1.6361,
+      "step": 1152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.278490851167155e-06,
+      "loss": 1.7643,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2738765638196625e-06,
+      "loss": 1.6035,
+      "step": 1154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269259833633877e-06,
+      "loss": 1.603,
+      "step": 1155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264640665569577e-06,
+      "loss": 1.7312,
+      "step": 1156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26001906458916e-06,
+      "loss": 1.542,
+      "step": 1157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.255395035657639e-06,
+      "loss": 1.5783,
+      "step": 1158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.250768583742634e-06,
+      "loss": 1.54,
+      "step": 1159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.246139713814365e-06,
+      "loss": 1.5958,
+      "step": 1160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241508430845656e-06,
+      "loss": 1.701,
+      "step": 1161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.236874739811921e-06,
+      "loss": 1.6266,
+      "step": 1162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.232238645691157e-06,
+      "loss": 1.7754,
+      "step": 1163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.227600153463947e-06,
+      "loss": 1.6153,
+      "step": 1164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.222959268113452e-06,
+      "loss": 1.6403,
+      "step": 1165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218315994625397e-06,
+      "loss": 1.588,
+      "step": 1166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.213670337988079e-06,
+      "loss": 1.6513,
+      "step": 1167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.209022303192351e-06,
+      "loss": 1.6723,
+      "step": 1168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.204371895231623e-06,
+      "loss": 1.6012,
+      "step": 1169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.199719119101858e-06,
+      "loss": 1.5164,
+      "step": 1170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.195063979801554e-06,
+      "loss": 1.5891,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.190406482331757e-06,
+      "loss": 1.5721,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18574663169604e-06,
+      "loss": 1.5743,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1810844329005095e-06,
+      "loss": 1.7068,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.176419890953788e-06,
+      "loss": 1.6473,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.171753010867023e-06,
+      "loss": 1.5883,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167083797653866e-06,
+      "loss": 1.6353,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.162412256330481e-06,
+      "loss": 1.6518,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157738391915531e-06,
+      "loss": 1.5339,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.153062209430174e-06,
+      "loss": 1.6174,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.148383713898058e-06,
+      "loss": 1.6658,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.143702910345318e-06,
+      "loss": 1.5673,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139019803800569e-06,
+      "loss": 1.5807,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.134334399294897e-06,
+      "loss": 1.7477,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129646701861858e-06,
+      "loss": 1.6218,
+      "step": 1185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.124956716537471e-06,
+      "loss": 1.6711,
+      "step": 1186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.120264448360214e-06,
+      "loss": 1.638,
+      "step": 1187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115569902371018e-06,
+      "loss": 1.6321,
+      "step": 1188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.110873083613259e-06,
+      "loss": 1.6191,
+      "step": 1189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106173997132755e-06,
+      "loss": 1.4842,
+      "step": 1190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.101472647977761e-06,
+      "loss": 1.6965,
+      "step": 1191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096769041198964e-06,
+      "loss": 1.6673,
+      "step": 1192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0920631818494745e-06,
+      "loss": 1.6358,
+      "step": 1193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087355074984823e-06,
+      "loss": 1.7363,
+      "step": 1194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.082644725662954e-06,
+      "loss": 1.6192,
+      "step": 1195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077932138944225e-06,
+      "loss": 1.6535,
+      "step": 1196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073217319891391e-06,
+      "loss": 1.655,
+      "step": 1197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.068500273569612e-06,
+      "loss": 1.5959,
+      "step": 1198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063781005046433e-06,
+      "loss": 1.7229,
+      "step": 1199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.059059519391794e-06,
+      "loss": 1.567,
+      "step": 1200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.054335821678012e-06,
+      "loss": 1.6309,
+      "step": 1201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.049609916979782e-06,
+      "loss": 1.6467,
+      "step": 1202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.044881810374169e-06,
+      "loss": 1.6157,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.040151506940605e-06,
+      "loss": 1.6358,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035419011760882e-06,
+      "loss": 1.5369,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0306843299191465e-06,
+      "loss": 1.6729,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0259474665018915e-06,
+      "loss": 1.6173,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0212084265979575e-06,
+      "loss": 1.683,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016467215298519e-06,
+      "loss": 1.632,
+      "step": 1209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.011723837697091e-06,
+      "loss": 1.6238,
+      "step": 1210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0069782988895056e-06,
+      "loss": 1.6256,
+      "step": 1211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002230603973924e-06,
+      "loss": 1.627,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9974807580508205e-06,
+      "loss": 1.5498,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.992728766222982e-06,
+      "loss": 1.5812,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.987974633595498e-06,
+      "loss": 1.6163,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9832183652757625e-06,
+      "loss": 1.59,
+      "step": 1216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978459966373458e-06,
+      "loss": 1.6932,
+      "step": 1217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.973699442000561e-06,
+      "loss": 1.5689,
+      "step": 1218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.96893679727133e-06,
+      "loss": 1.5826,
+      "step": 1219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9641720373022996e-06,
+      "loss": 1.5837,
+      "step": 1220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.959405167212278e-06,
+      "loss": 1.5108,
+      "step": 1221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954636192122339e-06,
+      "loss": 1.5627,
+      "step": 1222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.949865117155823e-06,
+      "loss": 1.5728,
+      "step": 1223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.94509194743832e-06,
+      "loss": 1.6462,
+      "step": 1224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940316688097675e-06,
+      "loss": 1.531,
+      "step": 1225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.935539344263971e-06,
+      "loss": 1.6016,
+      "step": 1226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93075992106954e-06,
+      "loss": 1.5628,
+      "step": 1227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.925978423648941e-06,
+      "loss": 1.7405,
+      "step": 1228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921194857138963e-06,
+      "loss": 1.6107,
+      "step": 1229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.91640922667862e-06,
+      "loss": 1.6891,
+      "step": 1230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.911621537409139e-06,
+      "loss": 1.6806,
+      "step": 1231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.906831794473963e-06,
+      "loss": 1.5559,
+      "step": 1232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9020400030187394e-06,
+      "loss": 1.6135,
+      "step": 1233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.897246168191317e-06,
+      "loss": 1.558,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.892450295141737e-06,
+      "loss": 1.701,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887652389022236e-06,
+      "loss": 1.595,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.88285245498723e-06,
+      "loss": 1.6353,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.878050498193314e-06,
+      "loss": 1.7457,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.873246523799256e-06,
+      "loss": 1.5948,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.868440536965997e-06,
+      "loss": 1.6247,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.863632542856632e-06,
+      "loss": 1.5904,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.858822546636417e-06,
+      "loss": 1.5387,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.854010553472757e-06,
+      "loss": 1.6547,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.849196568535201e-06,
+      "loss": 1.6166,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8443805969954445e-06,
+      "loss": 1.6275,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839562644027311e-06,
+      "loss": 1.6022,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.834742714806754e-06,
+      "loss": 1.4864,
+      "step": 1247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8299208145118475e-06,
+      "loss": 1.5629,
+      "step": 1248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.825096948322791e-06,
+      "loss": 1.5951,
+      "step": 1249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.820271121421889e-06,
+      "loss": 1.5712,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.815443338993554e-06,
+      "loss": 1.6558,
+      "step": 1251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810613606224299e-06,
+      "loss": 1.7031,
+      "step": 1252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.805781928302732e-06,
+      "loss": 1.5871,
+      "step": 1253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.800948310419554e-06,
+      "loss": 1.5576,
+      "step": 1254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796112757767547e-06,
+      "loss": 1.6629,
+      "step": 1255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7912752755415716e-06,
+      "loss": 1.5784,
+      "step": 1256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786435868938561e-06,
+      "loss": 1.668,
+      "step": 1257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78159454315752e-06,
+      "loss": 1.5803,
+      "step": 1258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.776751303399509e-06,
+      "loss": 1.6984,
+      "step": 1259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.771906154867649e-06,
+      "loss": 1.5681,
+      "step": 1260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.767059102767109e-06,
+      "loss": 1.6677,
+      "step": 1261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7622101523051045e-06,
+      "loss": 1.6682,
+      "step": 1262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757359308690889e-06,
+      "loss": 1.699,
+      "step": 1263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7525065771357546e-06,
+      "loss": 1.5778,
+      "step": 1264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7476519628530145e-06,
+      "loss": 1.5722,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742795471058009e-06,
+      "loss": 1.5944,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.737937106968094e-06,
+      "loss": 1.5986,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7330768758026374e-06,
+      "loss": 1.5262,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.728214782783013e-06,
+      "loss": 1.6351,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723350833132596e-06,
+      "loss": 1.5914,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7184850320767505e-06,
+      "loss": 1.6485,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7136173848428375e-06,
+      "loss": 1.6675,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.708747896660196e-06,
+      "loss": 1.6928,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.703876572760144e-06,
+      "loss": 1.6674,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6990034183759726e-06,
+      "loss": 1.6126,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694128438742939e-06,
+      "loss": 1.642,
+      "step": 1276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.689251639098261e-06,
+      "loss": 1.6183,
+      "step": 1277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684373024681112e-06,
+      "loss": 1.6983,
+      "step": 1278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.679492600732614e-06,
+      "loss": 1.7185,
+      "step": 1279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.674610372495832e-06,
+      "loss": 1.6813,
+      "step": 1280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669726345215776e-06,
+      "loss": 1.6136,
+      "step": 1281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.66484052413938e-06,
+      "loss": 1.6592,
+      "step": 1282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.659952914515508e-06,
+      "loss": 1.7093,
+      "step": 1283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65506352159495e-06,
+      "loss": 1.5995,
+      "step": 1284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.650172350630406e-06,
+      "loss": 1.6706,
+      "step": 1285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.645279406876488e-06,
+      "loss": 1.7152,
+      "step": 1286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640384695589714e-06,
+      "loss": 1.628,
+      "step": 1287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.635488222028497e-06,
+      "loss": 1.5641,
+      "step": 1288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.630589991453148e-06,
+      "loss": 1.5397,
+      "step": 1289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6256900091258644e-06,
+      "loss": 1.5584,
+      "step": 1290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.620788280310722e-06,
+      "loss": 1.6686,
+      "step": 1291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615884810273678e-06,
+      "loss": 1.6575,
+      "step": 1292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610979604282557e-06,
+      "loss": 1.5664,
+      "step": 1293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.606072667607048e-06,
+      "loss": 1.5524,
+      "step": 1294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.601164005518702e-06,
+      "loss": 1.6626,
+      "step": 1295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59625362329092e-06,
+      "loss": 1.7679,
+      "step": 1296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.591341526198955e-06,
+      "loss": 1.555,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.586427719519901e-06,
+      "loss": 1.6318,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581512208532685e-06,
+      "loss": 1.7193,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.576594998518071e-06,
+      "loss": 1.6205,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5716760947586425e-06,
+      "loss": 1.6356,
+      "step": 1301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.566755502538806e-06,
+      "loss": 1.621,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.561833227144784e-06,
+      "loss": 1.5574,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.556909273864601e-06,
+      "loss": 1.5964,
+      "step": 1304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.551983647988089e-06,
+      "loss": 1.715,
+      "step": 1305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.547056354806874e-06,
+      "loss": 1.646,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.542127399614376e-06,
+      "loss": 1.6208,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5371967877058e-06,
+      "loss": 1.6533,
+      "step": 1308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.532264524378128e-06,
+      "loss": 1.4628,
+      "step": 1309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52733061493012e-06,
+      "loss": 1.6643,
+      "step": 1310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.522395064662299e-06,
+      "loss": 1.6522,
+      "step": 1311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.517457878876958e-06,
+      "loss": 1.6412,
+      "step": 1312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.512519062878142e-06,
+      "loss": 1.5833,
+      "step": 1313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.507578621971646e-06,
+      "loss": 1.7039,
+      "step": 1314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502636561465018e-06,
+      "loss": 1.5547,
+      "step": 1315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.497692886667537e-06,
+      "loss": 1.7036,
+      "step": 1316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.492747602890223e-06,
+      "loss": 1.6507,
+      "step": 1317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.487800715445822e-06,
+      "loss": 1.7112,
+      "step": 1318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4828522296488014e-06,
+      "loss": 1.6341,
+      "step": 1319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.477902150815347e-06,
+      "loss": 1.7088,
+      "step": 1320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472950484263359e-06,
+      "loss": 1.5178,
+      "step": 1321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.467997235312437e-06,
+      "loss": 1.6346,
+      "step": 1322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.463042409283885e-06,
+      "loss": 1.6413,
+      "step": 1323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.458086011500703e-06,
+      "loss": 1.669,
+      "step": 1324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453128047287573e-06,
+      "loss": 1.5624,
+      "step": 1325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.448168521970865e-06,
+      "loss": 1.762,
+      "step": 1326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.443207440878624e-06,
+      "loss": 1.6472,
+      "step": 1327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.438244809340568e-06,
+      "loss": 1.5903,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.43328063268808e-06,
+      "loss": 1.7109,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.428314916254203e-06,
+      "loss": 1.6419,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.423347665373633e-06,
+      "loss": 1.5299,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.418378885382716e-06,
+      "loss": 1.6278,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41340858161944e-06,
+      "loss": 1.6165,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.408436759423431e-06,
+      "loss": 1.6244,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.403463424135943e-06,
+      "loss": 1.5657,
+      "step": 1335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.398488581099859e-06,
+      "loss": 1.5987,
+      "step": 1336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.393512235659681e-06,
+      "loss": 1.6861,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.388534393161525e-06,
+      "loss": 1.5066,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.383555058953115e-06,
+      "loss": 1.5115,
+      "step": 1339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.378574238383776e-06,
+      "loss": 1.5975,
+      "step": 1340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373591936804433e-06,
+      "loss": 1.7443,
+      "step": 1341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3686081595676e-06,
+      "loss": 1.6394,
+      "step": 1342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3636229120273766e-06,
+      "loss": 1.6366,
+      "step": 1343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3586361995394415e-06,
+      "loss": 1.6649,
+      "step": 1344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353648027461048e-06,
+      "loss": 1.5739,
+      "step": 1345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.348658401151018e-06,
+      "loss": 1.5142,
+      "step": 1346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.343667325969736e-06,
+      "loss": 1.6288,
+      "step": 1347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3386748072791395e-06,
+      "loss": 1.6977,
+      "step": 1348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.33368085044272e-06,
+      "loss": 1.7115,
+      "step": 1349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.328685460825512e-06,
+      "loss": 1.6475,
+      "step": 1350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.323688643794094e-06,
+      "loss": 1.5872,
+      "step": 1351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.318690404716572e-06,
+      "loss": 1.6245,
+      "step": 1352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.313690748962582e-06,
+      "loss": 1.6441,
+      "step": 1353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3086896819032814e-06,
+      "loss": 1.6055,
+      "step": 1354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.303687208911348e-06,
+      "loss": 1.6708,
+      "step": 1355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.298683335360962e-06,
+      "loss": 1.6661,
+      "step": 1356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.293678066627816e-06,
+      "loss": 1.5892,
+      "step": 1357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.288671408089098e-06,
+      "loss": 1.6714,
+      "step": 1358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.283663365123486e-06,
+      "loss": 1.654,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.278653943111152e-06,
+      "loss": 1.6084,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.273643147433743e-06,
+      "loss": 1.7169,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.268630983474388e-06,
+      "loss": 1.6326,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263617456617681e-06,
+      "loss": 1.6741,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.258602572249683e-06,
+      "loss": 1.6509,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2535863357579105e-06,
+      "loss": 1.6532,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.248568752531337e-06,
+      "loss": 1.6172,
+      "step": 1366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.243549827960378e-06,
+      "loss": 1.645,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.238529567436892e-06,
+      "loss": 1.6143,
+      "step": 1368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.233507976354174e-06,
+      "loss": 1.6152,
+      "step": 1369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.228485060106948e-06,
+      "loss": 1.5984,
+      "step": 1370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.223460824091358e-06,
+      "loss": 1.552,
+      "step": 1371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218435273704973e-06,
+      "loss": 1.5953,
+      "step": 1372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.213408414346765e-06,
+      "loss": 1.5982,
+      "step": 1373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.208380251417122e-06,
+      "loss": 1.6504,
+      "step": 1374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.203350790317825e-06,
+      "loss": 1.6363,
+      "step": 1375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.198320036452051e-06,
+      "loss": 1.7759,
+      "step": 1376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.193287995224371e-06,
+      "loss": 1.6313,
+      "step": 1377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.18825467204073e-06,
+      "loss": 1.5729,
+      "step": 1378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.183220072308459e-06,
+      "loss": 1.6097,
+      "step": 1379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.178184201436256e-06,
+      "loss": 1.6813,
+      "step": 1380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.173147064834183e-06,
+      "loss": 1.74,
+      "step": 1381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.168108667913666e-06,
+      "loss": 1.6896,
+      "step": 1382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.163069016087483e-06,
+      "loss": 1.6206,
+      "step": 1383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.158028114769758e-06,
+      "loss": 1.6046,
+      "step": 1384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.152985969375962e-06,
+      "loss": 1.6695,
+      "step": 1385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147942585322898e-06,
+      "loss": 1.6217,
+      "step": 1386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.142897968028704e-06,
+      "loss": 1.5238,
+      "step": 1387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.137852122912839e-06,
+      "loss": 1.6641,
+      "step": 1388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1328050553960804e-06,
+      "loss": 1.4706,
+      "step": 1389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1277567709005245e-06,
+      "loss": 1.6126,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.122707274849572e-06,
+      "loss": 1.6466,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.117656572667921e-06,
+      "loss": 1.6346,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.112604669781572e-06,
+      "loss": 1.4587,
+      "step": 1393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.107551571617813e-06,
+      "loss": 1.5424,
+      "step": 1394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1024972836052135e-06,
+      "loss": 1.7232,
+      "step": 1395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0974418111736235e-06,
+      "loss": 1.6612,
+      "step": 1396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.092385159754165e-06,
+      "loss": 1.6401,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0873273347792275e-06,
+      "loss": 1.5736,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0822683416824625e-06,
+      "loss": 1.5142,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.077208185898772e-06,
+      "loss": 1.6317,
+      "step": 1400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.07214687286431e-06,
+      "loss": 1.5747,
+      "step": 1401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.067084408016475e-06,
+      "loss": 1.6256,
+      "step": 1402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0620207967939e-06,
+      "loss": 1.6289,
+      "step": 1403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0569560446364495e-06,
+      "loss": 1.7056,
+      "step": 1404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.051890156985217e-06,
+      "loss": 1.7572,
+      "step": 1405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.046823139282515e-06,
+      "loss": 1.6029,
+      "step": 1406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.041754996971866e-06,
+      "loss": 1.5283,
+      "step": 1407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.036685735498004e-06,
+      "loss": 1.551,
+      "step": 1408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.031615360306867e-06,
+      "loss": 1.6254,
+      "step": 1409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.026543876845586e-06,
+      "loss": 1.6395,
+      "step": 1410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.021471290562484e-06,
+      "loss": 1.6371,
+      "step": 1411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.016397606907069e-06,
+      "loss": 1.6367,
+      "step": 1412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.011322831330028e-06,
+      "loss": 1.6307,
+      "step": 1413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0062469692832205e-06,
+      "loss": 1.663,
+      "step": 1414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.001170026219673e-06,
+      "loss": 1.6108,
+      "step": 1415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.996092007593572e-06,
+      "loss": 1.4994,
+      "step": 1416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9910129188602665e-06,
+      "loss": 1.6029,
+      "step": 1417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985932765476246e-06,
+      "loss": 1.6122,
+      "step": 1418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9808515528991486e-06,
+      "loss": 1.6107,
+      "step": 1419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.975769286587747e-06,
+      "loss": 1.6848,
+      "step": 1420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.970685972001953e-06,
+      "loss": 1.6856,
+      "step": 1421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.965601614602798e-06,
+      "loss": 1.6513,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.960516219852433e-06,
+      "loss": 1.6461,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.955429793214129e-06,
+      "loss": 1.6018,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.950342340152261e-06,
+      "loss": 1.723,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945253866132308e-06,
+      "loss": 1.5449,
+      "step": 1426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.940164376620847e-06,
+      "loss": 1.7331,
+      "step": 1427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.935073877085546e-06,
+      "loss": 1.5953,
+      "step": 1428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9299823729951544e-06,
+      "loss": 1.5347,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9248898698195054e-06,
+      "loss": 1.7125,
+      "step": 1430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.919796373029504e-06,
+      "loss": 1.5826,
+      "step": 1431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.914701888097121e-06,
+      "loss": 1.7135,
+      "step": 1432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.90960642049539e-06,
+      "loss": 1.7333,
+      "step": 1433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.904509975698399e-06,
+      "loss": 1.6674,
+      "step": 1434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8994125591812914e-06,
+      "loss": 1.6441,
+      "step": 1435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.894314176420247e-06,
+      "loss": 1.5428,
+      "step": 1436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.889214832892489e-06,
+      "loss": 1.5544,
+      "step": 1437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8841145340762665e-06,
+      "loss": 1.7111,
+      "step": 1438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.879013285450863e-06,
+      "loss": 1.7001,
+      "step": 1439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.873911092496577e-06,
+      "loss": 1.7108,
+      "step": 1440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8688079606947226e-06,
+      "loss": 1.6121,
+      "step": 1441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8637038955276225e-06,
+      "loss": 1.5452,
+      "step": 1442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.858598902478604e-06,
+      "loss": 1.6094,
+      "step": 1443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.853492987031989e-06,
+      "loss": 1.6334,
+      "step": 1444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8483861546730915e-06,
+      "loss": 1.6698,
+      "step": 1445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.843278410888208e-06,
+      "loss": 1.6204,
+      "step": 1446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.838169761164616e-06,
+      "loss": 1.5325,
+      "step": 1447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.83306021099057e-06,
+      "loss": 1.6136,
+      "step": 1448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.827949765855285e-06,
+      "loss": 1.5398,
+      "step": 1449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.822838431248943e-06,
+      "loss": 1.6749,
+      "step": 1450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.817726212662678e-06,
+      "loss": 1.759,
+      "step": 1451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812613115588575e-06,
+      "loss": 1.6234,
+      "step": 1452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807499145519663e-06,
+      "loss": 1.6444,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802384307949909e-06,
+      "loss": 1.5473,
+      "step": 1454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79726860837421e-06,
+      "loss": 1.7367,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.792152052288391e-06,
+      "loss": 1.6082,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.787034645189199e-06,
+      "loss": 1.7639,
+      "step": 1457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7819163925742915e-06,
+      "loss": 1.7142,
+      "step": 1458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.776797299942236e-06,
+      "loss": 1.5856,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.771677372792502e-06,
+      "loss": 1.494,
+      "step": 1460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.766556616625456e-06,
+      "loss": 1.7586,
+      "step": 1461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7614350369423555e-06,
+      "loss": 1.6054,
+      "step": 1462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7563126392453415e-06,
+      "loss": 1.489,
+      "step": 1463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.751189429037435e-06,
+      "loss": 1.6141,
+      "step": 1464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.746065411822528e-06,
+      "loss": 1.6481,
+      "step": 1465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.740940593105383e-06,
+      "loss": 1.6469,
+      "step": 1466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.73581497839162e-06,
+      "loss": 1.5017,
+      "step": 1467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.730688573187715e-06,
+      "loss": 1.6611,
+      "step": 1468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.725561383000994e-06,
+      "loss": 1.6666,
+      "step": 1469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.720433413339627e-06,
+      "loss": 1.6768,
+      "step": 1470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.71530466971262e-06,
+      "loss": 1.687,
+      "step": 1471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.710175157629812e-06,
+      "loss": 1.6419,
+      "step": 1472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.705044882601862e-06,
+      "loss": 1.6083,
+      "step": 1473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.69991385014026e-06,
+      "loss": 1.5221,
+      "step": 1474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.694782065757298e-06,
+      "loss": 1.5932,
+      "step": 1475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.689649534966083e-06,
+      "loss": 1.5266,
+      "step": 1476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.684516263280519e-06,
+      "loss": 1.6304,
+      "step": 1477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.679382256215311e-06,
+      "loss": 1.6096,
+      "step": 1478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.674247519285951e-06,
+      "loss": 1.6558,
+      "step": 1479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6691120580087126e-06,
+      "loss": 1.7407,
+      "step": 1480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6639758779006535e-06,
+      "loss": 1.6101,
+      "step": 1481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6588389844796e-06,
+      "loss": 1.671,
+      "step": 1482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.653701383264147e-06,
+      "loss": 1.6388,
+      "step": 1483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.648563079773646e-06,
+      "loss": 1.6518,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6434240795282045e-06,
+      "loss": 1.589,
+      "step": 1485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.63828438804868e-06,
+      "loss": 1.6131,
+      "step": 1486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6331440108566735e-06,
+      "loss": 1.601,
+      "step": 1487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.628002953474521e-06,
+      "loss": 1.5902,
+      "step": 1488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.622861221425286e-06,
+      "loss": 1.6581,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.617718820232762e-06,
+      "loss": 1.5739,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.612575755421459e-06,
+      "loss": 1.6145,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.607432032516601e-06,
+      "loss": 1.617,
+      "step": 1492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.602287657044116e-06,
+      "loss": 1.6052,
+      "step": 1493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.597142634530639e-06,
+      "loss": 1.6169,
+      "step": 1494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5919969705034914e-06,
+      "loss": 1.5961,
+      "step": 1495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.586850670490694e-06,
+      "loss": 1.644,
+      "step": 1496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.581703740020943e-06,
+      "loss": 1.7451,
+      "step": 1497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.576556184623615e-06,
+      "loss": 1.635,
+      "step": 1498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.571408009828757e-06,
+      "loss": 1.6687,
+      "step": 1499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.56625922116708e-06,
+      "loss": 1.5866,
+      "step": 1500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.561109824169962e-06,
+      "loss": 1.6749,
+      "step": 1501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.555959824369426e-06,
+      "loss": 1.6525,
+      "step": 1502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.550809227298144e-06,
+      "loss": 1.571,
+      "step": 1503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.545658038489433e-06,
+      "loss": 1.457,
+      "step": 1504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.540506263477243e-06,
+      "loss": 1.6858,
+      "step": 1505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.535353907796155e-06,
+      "loss": 1.6155,
+      "step": 1506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.530200976981375e-06,
+      "loss": 1.6362,
+      "step": 1507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.525047476568722e-06,
+      "loss": 1.6856,
+      "step": 1508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.519893412094631e-06,
+      "loss": 1.7182,
+      "step": 1509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.514738789096146e-06,
+      "loss": 1.6351,
+      "step": 1510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509583613110904e-06,
+      "loss": 1.518,
+      "step": 1511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.504427889677141e-06,
+      "loss": 1.5643,
+      "step": 1512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.499271624333676e-06,
+      "loss": 1.7602,
+      "step": 1513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.494114822619918e-06,
+      "loss": 1.8227,
+      "step": 1514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.488957490075846e-06,
+      "loss": 1.625,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.483799632242012e-06,
+      "loss": 1.7718,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.478641254659528e-06,
+      "loss": 1.6997,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.473482362870073e-06,
+      "loss": 1.4906,
+      "step": 1518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.468322962415871e-06,
+      "loss": 1.5711,
+      "step": 1519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.463163058839694e-06,
+      "loss": 1.5728,
+      "step": 1520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4580026576848565e-06,
+      "loss": 1.5419,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.452841764495203e-06,
+      "loss": 1.5387,
+      "step": 1522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4476803848151146e-06,
+      "loss": 1.5215,
+      "step": 1523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.442518524189489e-06,
+      "loss": 1.6186,
+      "step": 1524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4373561881637405e-06,
+      "loss": 1.6575,
+      "step": 1525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.432193382283794e-06,
+      "loss": 1.6372,
+      "step": 1526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4270301120960856e-06,
+      "loss": 1.5705,
+      "step": 1527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.421866383147541e-06,
+      "loss": 1.5832,
+      "step": 1528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.416702200985585e-06,
+      "loss": 1.64,
+      "step": 1529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.411537571158127e-06,
+      "loss": 1.7349,
+      "step": 1530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.406372499213557e-06,
+      "loss": 1.5862,
+      "step": 1531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.401206990700741e-06,
+      "loss": 1.6456,
+      "step": 1532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.396041051169016e-06,
+      "loss": 1.6165,
+      "step": 1533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.390874686168176e-06,
+      "loss": 1.7215,
+      "step": 1534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.385707901248478e-06,
+      "loss": 1.6185,
+      "step": 1535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.380540701960627e-06,
+      "loss": 1.5714,
+      "step": 1536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.375373093855774e-06,
+      "loss": 1.5664,
+      "step": 1537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.37020508248551e-06,
+      "loss": 1.5597,
+      "step": 1538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.365036673401857e-06,
+      "loss": 1.5907,
+      "step": 1539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.359867872157267e-06,
+      "loss": 1.607,
+      "step": 1540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.354698684304613e-06,
+      "loss": 1.7057,
+      "step": 1541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3495291153971806e-06,
+      "loss": 1.6277,
+      "step": 1542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.344359170988668e-06,
+      "loss": 1.7101,
+      "step": 1543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.339188856633173e-06,
+      "loss": 1.5366,
+      "step": 1544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3340181778851954e-06,
+      "loss": 1.6644,
+      "step": 1545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.328847140299624e-06,
+      "loss": 1.6909,
+      "step": 1546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.323675749431732e-06,
+      "loss": 1.7526,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.318504010837175e-06,
+      "loss": 1.6431,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.313331930071981e-06,
+      "loss": 1.6767,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.308159512692544e-06,
+      "loss": 1.6249,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.302986764255621e-06,
+      "loss": 1.6462,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.297813690318325e-06,
+      "loss": 1.4418,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.292640296438116e-06,
+      "loss": 1.6632,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.287466588172804e-06,
+      "loss": 1.5867,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2822925710805305e-06,
+      "loss": 1.6153,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.27711825071977e-06,
+      "loss": 1.5618,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2719436326493255e-06,
+      "loss": 1.5635,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.266768722428318e-06,
+      "loss": 1.6317,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.261593525616181e-06,
+      "loss": 1.5137,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.256418047772659e-06,
+      "loss": 1.5842,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.251242294457796e-06,
+      "loss": 1.6047,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2460662712319335e-06,
+      "loss": 1.6298,
+      "step": 1562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.240889983655701e-06,
+      "loss": 1.5827,
+      "step": 1563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.235713437290012e-06,
+      "loss": 1.6179,
+      "step": 1564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.230536637696062e-06,
+      "loss": 1.473,
+      "step": 1565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.225359590435312e-06,
+      "loss": 1.7038,
+      "step": 1566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.220182301069499e-06,
+      "loss": 1.565,
+      "step": 1567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.215004775160608e-06,
+      "loss": 1.5084,
+      "step": 1568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.209827018270886e-06,
+      "loss": 1.6452,
+      "step": 1569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.204649035962825e-06,
+      "loss": 1.5042,
+      "step": 1570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.199470833799164e-06,
+      "loss": 1.6742,
+      "step": 1571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1942924173428725e-06,
+      "loss": 1.5258,
+      "step": 1572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.18911379215715e-06,
+      "loss": 1.6551,
+      "step": 1573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1839349638054245e-06,
+      "loss": 1.566,
+      "step": 1574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.178755937851341e-06,
+      "loss": 1.6213,
+      "step": 1575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.173576719858755e-06,
+      "loss": 1.6674,
+      "step": 1576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.168397315391729e-06,
+      "loss": 1.5104,
+      "step": 1577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1632177300145255e-06,
+      "loss": 1.6689,
+      "step": 1578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1580379692916025e-06,
+      "loss": 1.6544,
+      "step": 1579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.152858038787608e-06,
+      "loss": 1.5364,
+      "step": 1580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.147677944067368e-06,
+      "loss": 1.63,
+      "step": 1581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.142497690695888e-06,
+      "loss": 1.5559,
+      "step": 1582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.137317284238344e-06,
+      "loss": 1.5943,
+      "step": 1583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1321367302600726e-06,
+      "loss": 1.6118,
+      "step": 1584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.126956034326573e-06,
+      "loss": 1.6705,
+      "step": 1585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.121775202003499e-06,
+      "loss": 1.4435,
+      "step": 1586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116594238856645e-06,
+      "loss": 1.5701,
+      "step": 1587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.111413150451948e-06,
+      "loss": 1.5551,
+      "step": 1588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1062319423554815e-06,
+      "loss": 1.5566,
+      "step": 1589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.101050620133447e-06,
+      "loss": 1.5716,
+      "step": 1590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.095869189352166e-06,
+      "loss": 1.5348,
+      "step": 1591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.090687655578078e-06,
+      "loss": 1.5324,
+      "step": 1592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0855060243777366e-06,
+      "loss": 1.5105,
+      "step": 1593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.080324301317795e-06,
+      "loss": 1.5581,
+      "step": 1594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0751424919650085e-06,
+      "loss": 1.6013,
+      "step": 1595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.069960601886224e-06,
+      "loss": 1.5631,
+      "step": 1596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.064778636648371e-06,
+      "loss": 1.6226,
+      "step": 1597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.05959660181847e-06,
+      "loss": 1.4688,
+      "step": 1598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.054414502963605e-06,
+      "loss": 1.6655,
+      "step": 1599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.049232345650936e-06,
+      "loss": 1.7646,
+      "step": 1600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.044050135447682e-06,
+      "loss": 1.5875,
+      "step": 1601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.038867877921124e-06,
+      "loss": 1.5746,
+      "step": 1602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.033685578638586e-06,
+      "loss": 1.7348,
+      "step": 1603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.028503243167443e-06,
+      "loss": 1.5959,
+      "step": 1604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.023320877075107e-06,
+      "loss": 1.6289,
+      "step": 1605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0181384859290215e-06,
+      "loss": 1.6216,
+      "step": 1606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.01295607529666e-06,
+      "loss": 1.5344,
+      "step": 1607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.007773650745514e-06,
+      "loss": 1.6169,
+      "step": 1608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0025912178430925e-06,
+      "loss": 1.5097,
+      "step": 1609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.997408782156909e-06,
+      "loss": 1.5628,
+      "step": 1610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9922263492544885e-06,
+      "loss": 1.5709,
+      "step": 1611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.987043924703342e-06,
+      "loss": 1.5554,
+      "step": 1612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.981861514070979e-06,
+      "loss": 1.572,
+      "step": 1613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.976679122924896e-06,
+      "loss": 1.6061,
+      "step": 1614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.971496756832557e-06,
+      "loss": 1.4766,
+      "step": 1615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.966314421361416e-06,
+      "loss": 1.5671,
+      "step": 1616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9611321220788775e-06,
+      "loss": 1.5225,
+      "step": 1617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.955949864552318e-06,
+      "loss": 1.5671,
+      "step": 1618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.950767654349067e-06,
+      "loss": 1.5397,
+      "step": 1619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.945585497036396e-06,
+      "loss": 1.5932,
+      "step": 1620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.940403398181531e-06,
+      "loss": 1.6826,
+      "step": 1621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.935221363351631e-06,
+      "loss": 1.5826,
+      "step": 1622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930039398113779e-06,
+      "loss": 1.5677,
+      "step": 1623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.924857508034994e-06,
+      "loss": 1.6111,
+      "step": 1624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.919675698682206e-06,
+      "loss": 1.6712,
+      "step": 1625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.914493975622263e-06,
+      "loss": 1.5757,
+      "step": 1626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.909312344421923e-06,
+      "loss": 1.5586,
+      "step": 1627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.904130810647836e-06,
+      "loss": 1.6973,
+      "step": 1628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.898949379866556e-06,
+      "loss": 1.5797,
+      "step": 1629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.893768057644519e-06,
+      "loss": 1.5567,
+      "step": 1630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.888586849548053e-06,
+      "loss": 1.5127,
+      "step": 1631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883405761143357e-06,
+      "loss": 1.6622,
+      "step": 1632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.878224797996502e-06,
+      "loss": 1.6046,
+      "step": 1633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.873043965673427e-06,
+      "loss": 1.6554,
+      "step": 1634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.86786326973993e-06,
+      "loss": 1.6002,
+      "step": 1635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.862682715761658e-06,
+      "loss": 1.5121,
+      "step": 1636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.857502309304114e-06,
+      "loss": 1.6734,
+      "step": 1637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.852322055932633e-06,
+      "loss": 1.7853,
+      "step": 1638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8471419612123925e-06,
+      "loss": 1.6923,
+      "step": 1639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.841962030708398e-06,
+      "loss": 1.737,
+      "step": 1640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.836782269985475e-06,
+      "loss": 1.6493,
+      "step": 1641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.831602684608274e-06,
+      "loss": 1.6086,
+      "step": 1642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.826423280141247e-06,
+      "loss": 1.4956,
+      "step": 1643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.82124406214866e-06,
+      "loss": 1.7713,
+      "step": 1644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816065036194576e-06,
+      "loss": 1.6411,
+      "step": 1645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.810886207842852e-06,
+      "loss": 1.7202,
+      "step": 1646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.80570758265713e-06,
+      "loss": 1.6786,
+      "step": 1647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.800529166200837e-06,
+      "loss": 1.6285,
+      "step": 1648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.795350964037174e-06,
+      "loss": 1.5347,
+      "step": 1649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.790172981729116e-06,
+      "loss": 1.6919,
+      "step": 1650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.784995224839394e-06,
+      "loss": 1.6351,
+      "step": 1651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.779817698930502e-06,
+      "loss": 1.6019,
+      "step": 1652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.774640409564688e-06,
+      "loss": 1.6592,
+      "step": 1653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.76946336230394e-06,
+      "loss": 1.5226,
+      "step": 1654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.76428656270999e-06,
+      "loss": 1.5258,
+      "step": 1655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.759110016344302e-06,
+      "loss": 1.489,
+      "step": 1656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.753933728768069e-06,
+      "loss": 1.5365,
+      "step": 1657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.748757705542205e-06,
+      "loss": 1.7199,
+      "step": 1658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.743581952227342e-06,
+      "loss": 1.5399,
+      "step": 1659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73840647438382e-06,
+      "loss": 1.6449,
+      "step": 1660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.733231277571683e-06,
+      "loss": 1.7287,
+      "step": 1661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7280563673506745e-06,
+      "loss": 1.6035,
+      "step": 1662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.722881749280232e-06,
+      "loss": 1.5938,
+      "step": 1663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.717707428919471e-06,
+      "loss": 1.5948,
+      "step": 1664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.712533411827197e-06,
+      "loss": 1.5309,
+      "step": 1665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.707359703561885e-06,
+      "loss": 1.6476,
+      "step": 1666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.702186309681677e-06,
+      "loss": 1.6027,
+      "step": 1667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.697013235744382e-06,
+      "loss": 1.5592,
+      "step": 1668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6918404873074574e-06,
+      "loss": 1.5584,
+      "step": 1669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.68666806992802e-06,
+      "loss": 1.7658,
+      "step": 1670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.681495989162826e-06,
+      "loss": 1.736,
+      "step": 1671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.676324250568269e-06,
+      "loss": 1.697,
+      "step": 1672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.671152859700377e-06,
+      "loss": 1.7912,
+      "step": 1673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665981822114805e-06,
+      "loss": 1.603,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660811143366828e-06,
+      "loss": 1.5781,
+      "step": 1675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.655640829011335e-06,
+      "loss": 1.7629,
+      "step": 1676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.65047088460282e-06,
+      "loss": 1.5897,
+      "step": 1677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.645301315695387e-06,
+      "loss": 1.5968,
+      "step": 1678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6401321278427334e-06,
+      "loss": 1.5849,
+      "step": 1679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.634963326598143e-06,
+      "loss": 1.7093,
+      "step": 1680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.629794917514492e-06,
+      "loss": 1.5546,
+      "step": 1681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.624626906144227e-06,
+      "loss": 1.6329,
+      "step": 1682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.619459298039373e-06,
+      "loss": 1.5871,
+      "step": 1683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614292098751524e-06,
+      "loss": 1.5954,
+      "step": 1684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609125313831826e-06,
+      "loss": 1.5656,
+      "step": 1685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.603958948830985e-06,
+      "loss": 1.6488,
+      "step": 1686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5987930092992596e-06,
+      "loss": 1.5221,
+      "step": 1687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.593627500786444e-06,
+      "loss": 1.5111,
+      "step": 1688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.588462428841875e-06,
+      "loss": 1.6804,
+      "step": 1689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5832977990144165e-06,
+      "loss": 1.6067,
+      "step": 1690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.578133616852462e-06,
+      "loss": 1.6115,
+      "step": 1691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.572969887903916e-06,
+      "loss": 1.5962,
+      "step": 1692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5678066177162065e-06,
+      "loss": 1.5565,
+      "step": 1693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.562643811836263e-06,
+      "loss": 1.5933,
+      "step": 1694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.557481475810512e-06,
+      "loss": 1.5718,
+      "step": 1695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5523196151848846e-06,
+      "loss": 1.7144,
+      "step": 1696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.547158235504797e-06,
+      "loss": 1.6659,
+      "step": 1697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.541997342315145e-06,
+      "loss": 1.5754,
+      "step": 1698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.536836941160308e-06,
+      "loss": 1.6736,
+      "step": 1699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5316770375841315e-06,
+      "loss": 1.6177,
+      "step": 1700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.526517637129927e-06,
+      "loss": 1.5024,
+      "step": 1701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5213587453404736e-06,
+      "loss": 1.5601,
+      "step": 1702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5162003677579905e-06,
+      "loss": 1.5883,
+      "step": 1703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.511042509924157e-06,
+      "loss": 1.6204,
+      "step": 1704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.505885177380083e-06,
+      "loss": 1.6859,
+      "step": 1705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5007283756663245e-06,
+      "loss": 1.6075,
+      "step": 1706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.495572110322862e-06,
+      "loss": 1.627,
+      "step": 1707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490416386889097e-06,
+      "loss": 1.5763,
+      "step": 1708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.485261210903854e-06,
+      "loss": 1.6142,
+      "step": 1709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.48010658790537e-06,
+      "loss": 1.6023,
+      "step": 1710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.47495252343128e-06,
+      "loss": 1.4913,
+      "step": 1711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.469799023018628e-06,
+      "loss": 1.6467,
+      "step": 1712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464646092203846e-06,
+      "loss": 1.4391,
+      "step": 1713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459493736522759e-06,
+      "loss": 1.6167,
+      "step": 1714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4543419615105685e-06,
+      "loss": 1.6435,
+      "step": 1715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.449190772701857e-06,
+      "loss": 1.5258,
+      "step": 1716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.444040175630577e-06,
+      "loss": 1.6531,
+      "step": 1717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.438890175830039e-06,
+      "loss": 1.483,
+      "step": 1718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.433740778832919e-06,
+      "loss": 1.5687,
+      "step": 1719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.428591990171246e-06,
+      "loss": 1.5745,
+      "step": 1720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.423443815376387e-06,
+      "loss": 1.6006,
+      "step": 1721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.41829625997906e-06,
+      "loss": 1.6792,
+      "step": 1722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.413149329509307e-06,
+      "loss": 1.6136,
+      "step": 1723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4080030294965085e-06,
+      "loss": 1.5359,
+      "step": 1724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.402857365469364e-06,
+      "loss": 1.6626,
+      "step": 1725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.397712342955885e-06,
+      "loss": 1.6809,
+      "step": 1726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.392567967483401e-06,
+      "loss": 1.6246,
+      "step": 1727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.387424244578543e-06,
+      "loss": 1.5669,
+      "step": 1728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.38228117976724e-06,
+      "loss": 1.5692,
+      "step": 1729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.377138778574716e-06,
+      "loss": 1.6987,
+      "step": 1730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.371997046525481e-06,
+      "loss": 1.562,
+      "step": 1731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.366855989143326e-06,
+      "loss": 1.6175,
+      "step": 1732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3617156119513206e-06,
+      "loss": 1.6436,
+      "step": 1733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.356575920471796e-06,
+      "loss": 1.6565,
+      "step": 1734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.351436920226357e-06,
+      "loss": 1.5219,
+      "step": 1735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.346298616735855e-06,
+      "loss": 1.5414,
+      "step": 1736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3411610155204e-06,
+      "loss": 1.6281,
+      "step": 1737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.336024122099348e-06,
+      "loss": 1.6292,
+      "step": 1738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.330887941991288e-06,
+      "loss": 1.552,
+      "step": 1739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.325752480714052e-06,
+      "loss": 1.6188,
+      "step": 1740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.320617743784691e-06,
+      "loss": 1.5672,
+      "step": 1741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.315483736719482e-06,
+      "loss": 1.6388,
+      "step": 1742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.310350465033919e-06,
+      "loss": 1.6848,
+      "step": 1743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.305217934242703e-06,
+      "loss": 1.6893,
+      "step": 1744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30008614985974e-06,
+      "loss": 1.5102,
+      "step": 1745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.294955117398139e-06,
+      "loss": 1.5425,
+      "step": 1746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.28982484237019e-06,
+      "loss": 1.6554,
+      "step": 1747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.284695330287383e-06,
+      "loss": 1.8029,
+      "step": 1748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.279566586660375e-06,
+      "loss": 1.6112,
+      "step": 1749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.274438616999007e-06,
+      "loss": 1.4813,
+      "step": 1750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.269311426812287e-06,
+      "loss": 1.6408,
+      "step": 1751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.264185021608382e-06,
+      "loss": 1.6621,
+      "step": 1752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.259059406894619e-06,
+      "loss": 1.5699,
+      "step": 1753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.253934588177473e-06,
+      "loss": 1.5855,
+      "step": 1754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.248810570962567e-06,
+      "loss": 1.5881,
+      "step": 1755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.24368736075466e-06,
+      "loss": 1.6435,
+      "step": 1756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.238564963057646e-06,
+      "loss": 1.7154,
+      "step": 1757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.233443383374545e-06,
+      "loss": 1.6265,
+      "step": 1758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.228322627207499e-06,
+      "loss": 1.551,
+      "step": 1759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.223202700057765e-06,
+      "loss": 1.5598,
+      "step": 1760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.21808360742571e-06,
+      "loss": 1.5965,
+      "step": 1761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.212965354810802e-06,
+      "loss": 1.6902,
+      "step": 1762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.207847947711609e-06,
+      "loss": 1.5247,
+      "step": 1763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.202731391625793e-06,
+      "loss": 1.6822,
+      "step": 1764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1976156920500935e-06,
+      "loss": 1.5588,
+      "step": 1765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.19250085448034e-06,
+      "loss": 1.6594,
+      "step": 1766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.187386884411426e-06,
+      "loss": 1.6278,
+      "step": 1767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.182273787337323e-06,
+      "loss": 1.6248,
+      "step": 1768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.177161568751058e-06,
+      "loss": 1.5875,
+      "step": 1769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.172050234144716e-06,
+      "loss": 1.5521,
+      "step": 1770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.16693978900943e-06,
+      "loss": 1.6565,
+      "step": 1771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.161830238835386e-06,
+      "loss": 1.7628,
+      "step": 1772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.156721589111794e-06,
+      "loss": 1.5501,
+      "step": 1773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.151613845326912e-06,
+      "loss": 1.5984,
+      "step": 1774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.146507012968013e-06,
+      "loss": 1.5022,
+      "step": 1775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.141401097521396e-06,
+      "loss": 1.6083,
+      "step": 1776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.136296104472378e-06,
+      "loss": 1.6057,
+      "step": 1777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.131192039305278e-06,
+      "loss": 1.5331,
+      "step": 1778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1260889075034254e-06,
+      "loss": 1.5339,
+      "step": 1779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120986714549139e-06,
+      "loss": 1.6626,
+      "step": 1780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.115885465923734e-06,
+      "loss": 1.6331,
+      "step": 1781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.110785167107514e-06,
+      "loss": 1.6099,
+      "step": 1782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1056858235797545e-06,
+      "loss": 1.5823,
+      "step": 1783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.100587440818709e-06,
+      "loss": 1.6134,
+      "step": 1784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0954900243016016e-06,
+      "loss": 1.7329,
+      "step": 1785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.090393579504612e-06,
+      "loss": 1.6914,
+      "step": 1786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.085298111902882e-06,
+      "loss": 1.6907,
+      "step": 1787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.080203626970498e-06,
+      "loss": 1.5868,
+      "step": 1788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0751101301804945e-06,
+      "loss": 1.6455,
+      "step": 1789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070017627004847e-06,
+      "loss": 1.7548,
+      "step": 1790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0649261229144554e-06,
+      "loss": 1.6386,
+      "step": 1791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059835623379155e-06,
+      "loss": 1.6485,
+      "step": 1792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.054746133867693e-06,
+      "loss": 1.6071,
+      "step": 1793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0496576598477396e-06,
+      "loss": 1.5924,
+      "step": 1794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044570206785874e-06,
+      "loss": 1.5621,
+      "step": 1795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.039483780147568e-06,
+      "loss": 1.4225,
+      "step": 1796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0343983853972045e-06,
+      "loss": 1.5963,
+      "step": 1797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029314027998049e-06,
+      "loss": 1.592,
+      "step": 1798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.024230713412253e-06,
+      "loss": 1.6015,
+      "step": 1799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.019148447100855e-06,
+      "loss": 1.6167,
+      "step": 1800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.014067234523756e-06,
+      "loss": 1.6035,
+      "step": 1801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.008987081139734e-06,
+      "loss": 1.5809,
+      "step": 1802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0039079924064285e-06,
+      "loss": 1.6017,
+      "step": 1803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.998829973780329e-06,
+      "loss": 1.5416,
+      "step": 1804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.993753030716783e-06,
+      "loss": 1.7119,
+      "step": 1805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.988677168669974e-06,
+      "loss": 1.6866,
+      "step": 1806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.983602393092931e-06,
+      "loss": 1.6728,
+      "step": 1807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.978528709437518e-06,
+      "loss": 1.6598,
+      "step": 1808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.973456123154415e-06,
+      "loss": 1.6777,
+      "step": 1809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9683846396931345e-06,
+      "loss": 1.5899,
+      "step": 1810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9633142645019965e-06,
+      "loss": 1.5487,
+      "step": 1811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.958245003028136e-06,
+      "loss": 1.6043,
+      "step": 1812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.953176860717488e-06,
+      "loss": 1.7218,
+      "step": 1813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.948109843014784e-06,
+      "loss": 1.5941,
+      "step": 1814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9430439553635504e-06,
+      "loss": 1.573,
+      "step": 1815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.937979203206103e-06,
+      "loss": 1.5141,
+      "step": 1816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.932915591983526e-06,
+      "loss": 1.6532,
+      "step": 1817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.927853127135692e-06,
+      "loss": 1.6495,
+      "step": 1818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.92279181410123e-06,
+      "loss": 1.528,
+      "step": 1819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.917731658317538e-06,
+      "loss": 1.5165,
+      "step": 1820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.912672665220773e-06,
+      "loss": 1.5503,
+      "step": 1821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.907614840245836e-06,
+      "loss": 1.5702,
+      "step": 1822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.90255818882638e-06,
+      "loss": 1.6097,
+      "step": 1823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.897502716394789e-06,
+      "loss": 1.6616,
+      "step": 1824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.892448428382189e-06,
+      "loss": 1.7052,
+      "step": 1825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.887395330218429e-06,
+      "loss": 1.5712,
+      "step": 1826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8823434273320794e-06,
+      "loss": 1.6201,
+      "step": 1827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.877292725150429e-06,
+      "loss": 1.6539,
+      "step": 1828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.872243229099476e-06,
+      "loss": 1.706,
+      "step": 1829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.86719494460392e-06,
+      "loss": 1.7413,
+      "step": 1830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8621478770871645e-06,
+      "loss": 1.6375,
+      "step": 1831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.857102031971298e-06,
+      "loss": 1.6882,
+      "step": 1832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.852057414677102e-06,
+      "loss": 1.5244,
+      "step": 1833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.84701403062404e-06,
+      "loss": 1.7413,
+      "step": 1834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.841971885230243e-06,
+      "loss": 1.5745,
+      "step": 1835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.83693098391252e-06,
+      "loss": 1.5547,
+      "step": 1836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8318913320863355e-06,
+      "loss": 1.6303,
+      "step": 1837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.826852935165818e-06,
+      "loss": 1.6925,
+      "step": 1838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8218157985637465e-06,
+      "loss": 1.471,
+      "step": 1839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.816779927691542e-06,
+      "loss": 1.5,
+      "step": 1840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.811745327959271e-06,
+      "loss": 1.6576,
+      "step": 1841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8067120047756313e-06,
+      "loss": 1.816,
+      "step": 1842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.801679963547949e-06,
+      "loss": 1.7396,
+      "step": 1843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7966492096821773e-06,
+      "loss": 1.606,
+      "step": 1844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7916197485828793e-06,
+      "loss": 1.5486,
+      "step": 1845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.786591585653235e-06,
+      "loss": 1.684,
+      "step": 1846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7815647262950293e-06,
+      "loss": 1.6652,
+      "step": 1847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7765391759086424e-06,
+      "loss": 1.4994,
+      "step": 1848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.771514939893055e-06,
+      "loss": 1.5662,
+      "step": 1849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.766492023645827e-06,
+      "loss": 1.713,
+      "step": 1850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.761470432563109e-06,
+      "loss": 1.5187,
+      "step": 1851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7564501720396242e-06,
+      "loss": 1.6051,
+      "step": 1852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7514312474686643e-06,
+      "loss": 1.5011,
+      "step": 1853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.74641366424209e-06,
+      "loss": 1.6804,
+      "step": 1854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7413974277503183e-06,
+      "loss": 1.717,
+      "step": 1855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7363825433823187e-06,
+      "loss": 1.6763,
+      "step": 1856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7313690165256134e-06,
+      "loss": 1.5893,
+      "step": 1857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7263568525662574e-06,
+      "loss": 1.5842,
+      "step": 1858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7213460568888493e-06,
+      "loss": 1.6206,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.716336634876516e-06,
+      "loss": 1.5352,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.711328591910904e-06,
+      "loss": 1.4686,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7063219333721857e-06,
+      "loss": 1.619,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7013166646390384e-06,
+      "loss": 1.5273,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6963127910886526e-06,
+      "loss": 1.5965,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.691310318096719e-06,
+      "loss": 1.6488,
+      "step": 1865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6863092510374198e-06,
+      "loss": 1.5214,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.68130959528343e-06,
+      "loss": 1.5955,
+      "step": 1867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6763113562059077e-06,
+      "loss": 1.5717,
+      "step": 1868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6713145391744877e-06,
+      "loss": 1.5016,
+      "step": 1869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6663191495572827e-06,
+      "loss": 1.6339,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.661325192720862e-06,
+      "loss": 1.5696,
+      "step": 1871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6563326740302664e-06,
+      "loss": 1.6826,
+      "step": 1872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6513415988489824e-06,
+      "loss": 1.5456,
+      "step": 1873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6463519725389516e-06,
+      "loss": 1.696,
+      "step": 1874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.64136380046056e-06,
+      "loss": 1.4761,
+      "step": 1875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6363770879726247e-06,
+      "loss": 1.5954,
+      "step": 1876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6313918404324e-06,
+      "loss": 1.6659,
+      "step": 1877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6264080631955683e-06,
+      "loss": 1.6617,
+      "step": 1878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.621425761616224e-06,
+      "loss": 1.6608,
+      "step": 1879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.616444941046887e-06,
+      "loss": 1.5406,
+      "step": 1880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6114656068384767e-06,
+      "loss": 1.6362,
+      "step": 1881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6064877643403194e-06,
+      "loss": 1.6452,
+      "step": 1882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.601511418900143e-06,
+      "loss": 1.6278,
+      "step": 1883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5965365758640587e-06,
+      "loss": 1.5713,
+      "step": 1884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.591563240576572e-06,
+      "loss": 1.6103,
+      "step": 1885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5865914183805606e-06,
+      "loss": 1.5735,
+      "step": 1886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.581621114617284e-06,
+      "loss": 1.5629,
+      "step": 1887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5766523346263682e-06,
+      "loss": 1.6063,
+      "step": 1888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.571685083745798e-06,
+      "loss": 1.618,
+      "step": 1889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.56671936731192e-06,
+      "loss": 1.5918,
+      "step": 1890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.561755190659434e-06,
+      "loss": 1.6292,
+      "step": 1891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.556792559121377e-06,
+      "loss": 1.708,
+      "step": 1892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5518314780291384e-06,
+      "loss": 1.6912,
+      "step": 1893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5468719527124294e-06,
+      "loss": 1.6753,
+      "step": 1894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.541913988499299e-06,
+      "loss": 1.5589,
+      "step": 1895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5369575907161167e-06,
+      "loss": 1.4678,
+      "step": 1896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5320027646875643e-06,
+      "loss": 1.5447,
+      "step": 1897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5270495157366434e-06,
+      "loss": 1.621,
+      "step": 1898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5220978491846534e-06,
+      "loss": 1.4979,
+      "step": 1899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.517147770351199e-06,
+      "loss": 1.5884,
+      "step": 1900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5121992845541797e-06,
+      "loss": 1.6247,
+      "step": 1901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.507252397109777e-06,
+      "loss": 1.5697,
+      "step": 1902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5023071133324627e-06,
+      "loss": 1.8289,
+      "step": 1903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497363438534984e-06,
+      "loss": 1.6349,
+      "step": 1904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4924213780283545e-06,
+      "loss": 1.5937,
+      "step": 1905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4874809371218608e-06,
+      "loss": 1.5656,
+      "step": 1906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4825421211230437e-06,
+      "loss": 1.6167,
+      "step": 1907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4776049353377016e-06,
+      "loss": 1.6333,
+      "step": 1908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4726693850698824e-06,
+      "loss": 1.5932,
+      "step": 1909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.467735475621873e-06,
+      "loss": 1.5463,
+      "step": 1910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4628032122942024e-06,
+      "loss": 1.6699,
+      "step": 1911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4578726003856245e-06,
+      "loss": 1.6041,
+      "step": 1912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4529436451931263e-06,
+      "loss": 1.5525,
+      "step": 1913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.448016352011914e-06,
+      "loss": 1.4998,
+      "step": 1914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4430907261354e-06,
+      "loss": 1.6314,
+      "step": 1915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.438166772855218e-06,
+      "loss": 1.5597,
+      "step": 1916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4332444974611946e-06,
+      "loss": 1.5352,
+      "step": 1917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.428323905241358e-06,
+      "loss": 1.5097,
+      "step": 1918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4234050014819308e-06,
+      "loss": 1.6978,
+      "step": 1919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4184877914673155e-06,
+      "loss": 1.6134,
+      "step": 1920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4135722804801004e-06,
+      "loss": 1.6086,
+      "step": 1921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4086584738010455e-06,
+      "loss": 1.6049,
+      "step": 1922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4037463767090807e-06,
+      "loss": 1.6671,
+      "step": 1923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3988359944812997e-06,
+      "loss": 1.6438,
+      "step": 1924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3939273323929533e-06,
+      "loss": 1.7279,
+      "step": 1925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3890203957174437e-06,
+      "loss": 1.6345,
+      "step": 1926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3841151897263234e-06,
+      "loss": 1.6688,
+      "step": 1927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.379211719689278e-06,
+      "loss": 1.5354,
+      "step": 1928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3743099908741385e-06,
+      "loss": 1.6646,
+      "step": 1929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3694100085468535e-06,
+      "loss": 1.6199,
+      "step": 1930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.364511777971504e-06,
+      "loss": 1.5971,
+      "step": 1931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3596153044102897e-06,
+      "loss": 1.5833,
+      "step": 1932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.354720593123514e-06,
+      "loss": 1.6364,
+      "step": 1933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.349827649369596e-06,
+      "loss": 1.6727,
+      "step": 1934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3449364784050515e-06,
+      "loss": 1.6109,
+      "step": 1935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3400470854844925e-06,
+      "loss": 1.6626,
+      "step": 1936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3351594758606222e-06,
+      "loss": 1.5556,
+      "step": 1937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3302736547842263e-06,
+      "loss": 1.5948,
+      "step": 1938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3253896275041677e-06,
+      "loss": 1.5957,
+      "step": 1939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3205073992673885e-06,
+      "loss": 1.6002,
+      "step": 1940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3156269753188895e-06,
+      "loss": 1.6065,
+      "step": 1941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.310748360901741e-06,
+      "loss": 1.6548,
+      "step": 1942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3058715612570623e-06,
+      "loss": 1.4916,
+      "step": 1943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.300996581624028e-06,
+      "loss": 1.6735,
+      "step": 1944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2961234272398578e-06,
+      "loss": 1.5659,
+      "step": 1945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.291252103339806e-06,
+      "loss": 1.5369,
+      "step": 1946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2863826151571654e-06,
+      "loss": 1.5896,
+      "step": 1947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2815149679232507e-06,
+      "loss": 1.5789,
+      "step": 1948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.276649166867406e-06,
+      "loss": 1.5187,
+      "step": 1949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.271785217216987e-06,
+      "loss": 1.6855,
+      "step": 1950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.266923124197363e-06,
+      "loss": 1.6045,
+      "step": 1951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2620628930319065e-06,
+      "loss": 1.5274,
+      "step": 1952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.257204528941993e-06,
+      "loss": 1.6645,
+      "step": 1953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2523480371469863e-06,
+      "loss": 1.6585,
+      "step": 1954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2474934228642475e-06,
+      "loss": 1.6366,
+      "step": 1955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.242640691309111e-06,
+      "loss": 1.54,
+      "step": 1956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2377898476948964e-06,
+      "loss": 1.5405,
+      "step": 1957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2329408972328934e-06,
+      "loss": 1.5428,
+      "step": 1958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2280938451323524e-06,
+      "loss": 1.5209,
+      "step": 1959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.223248696600493e-06,
+      "loss": 1.4972,
+      "step": 1960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2184054568424817e-06,
+      "loss": 1.5334,
+      "step": 1961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2135641310614383e-06,
+      "loss": 1.5712,
+      "step": 1962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.20872472445843e-06,
+      "loss": 1.6028,
+      "step": 1963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.203887242232455e-06,
+      "loss": 1.7275,
+      "step": 1964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1990516895804467e-06,
+      "loss": 1.6883,
+      "step": 1965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1942180716972698e-06,
+      "loss": 1.5775,
+      "step": 1966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.189386393775703e-06,
+      "loss": 1.601,
+      "step": 1967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1845566610064487e-06,
+      "loss": 1.6297,
+      "step": 1968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.179728878578112e-06,
+      "loss": 1.5826,
+      "step": 1969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1749030516772084e-06,
+      "loss": 1.644,
+      "step": 1970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.170079185488153e-06,
+      "loss": 1.6836,
+      "step": 1971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.165257285193248e-06,
+      "loss": 1.5981,
+      "step": 1972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1604373559726915e-06,
+      "loss": 1.5119,
+      "step": 1973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1556194030045563e-06,
+      "loss": 1.7028,
+      "step": 1974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1508034314647994e-06,
+      "loss": 1.7548,
+      "step": 1975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1459894465272467e-06,
+      "loss": 1.6322,
+      "step": 1976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1411774533635854e-06,
+      "loss": 1.552,
+      "step": 1977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.136367457143369e-06,
+      "loss": 1.716,
+      "step": 1978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1315594630340052e-06,
+      "loss": 1.5174,
+      "step": 1979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1267534762007435e-06,
+      "loss": 1.8026,
+      "step": 1980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1219495018066888e-06,
+      "loss": 1.5285,
+      "step": 1981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1171475450127717e-06,
+      "loss": 1.6621,
+      "step": 1982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.112347610977764e-06,
+      "loss": 1.6535,
+      "step": 1983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1075497048582635e-06,
+      "loss": 1.5387,
+      "step": 1984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.102753831808685e-06,
+      "loss": 1.4667,
+      "step": 1985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.097959996981263e-06,
+      "loss": 1.6476,
+      "step": 1986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.093168205526038e-06,
+      "loss": 1.5313,
+      "step": 1987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0883784625908618e-06,
+      "loss": 1.6606,
+      "step": 1988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.083590773321383e-06,
+      "loss": 1.6261,
+      "step": 1989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0788051428610377e-06,
+      "loss": 1.6614,
+      "step": 1990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0740215763510617e-06,
+      "loss": 1.7104,
+      "step": 1991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.069240078930461e-06,
+      "loss": 1.6244,
+      "step": 1992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0644606557360303e-06,
+      "loss": 1.7387,
+      "step": 1993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0596833119023283e-06,
+      "loss": 1.5726,
+      "step": 1994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.054908052561681e-06,
+      "loss": 1.5628,
+      "step": 1995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0501348828441767e-06,
+      "loss": 1.5561,
+      "step": 1996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0453638078776614e-06,
+      "loss": 1.5743,
+      "step": 1997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0405948327877233e-06,
+      "loss": 1.5954,
+      "step": 1998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0358279626977034e-06,
+      "loss": 1.5797,
+      "step": 1999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0310632027286717e-06,
+      "loss": 1.563,
+      "step": 2000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.026300557999439e-06,
+      "loss": 1.5647,
+      "step": 2001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.021540033626544e-06,
+      "loss": 1.6769,
+      "step": 2002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0167816347242396e-06,
+      "loss": 1.7214,
+      "step": 2003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.012025366404504e-06,
+      "loss": 1.5546,
+      "step": 2004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.00727123377702e-06,
+      "loss": 1.623,
+      "step": 2005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.002519241949181e-06,
+      "loss": 1.6212,
+      "step": 2006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.997769396026078e-06,
+      "loss": 1.5254,
+      "step": 2007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9930217011104957e-06,
+      "loss": 1.6718,
+      "step": 2008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.98827616230291e-06,
+      "loss": 1.6451,
+      "step": 2009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9835327847014816e-06,
+      "loss": 1.6795,
+      "step": 2010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9787915734020446e-06,
+      "loss": 1.5315,
+      "step": 2011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9740525334981105e-06,
+      "loss": 1.5786,
+      "step": 2012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9693156700808556e-06,
+      "loss": 1.6679,
+      "step": 2013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9645809882391187e-06,
+      "loss": 1.5518,
+      "step": 2014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.959848493059396e-06,
+      "loss": 1.5297,
+      "step": 2015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9551181896258317e-06,
+      "loss": 1.6144,
+      "step": 2016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9503900830202202e-06,
+      "loss": 1.5976,
+      "step": 2017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9456641783219897e-06,
+      "loss": 1.6358,
+      "step": 2018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9409404806082077e-06,
+      "loss": 1.6008,
+      "step": 2019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.936218994953568e-06,
+      "loss": 1.6312,
+      "step": 2020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.93149972643039e-06,
+      "loss": 1.6939,
+      "step": 2021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9267826801086103e-06,
+      "loss": 1.6342,
+      "step": 2022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9220678610557773e-06,
+      "loss": 1.6929,
+      "step": 2023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9173552743370454e-06,
+      "loss": 1.5871,
+      "step": 2024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.912644925015179e-06,
+      "loss": 1.7147,
+      "step": 2025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9079368181505263e-06,
+      "loss": 1.6029,
+      "step": 2026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9032309588010372e-06,
+      "loss": 1.58,
+      "step": 2027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8985273520222414e-06,
+      "loss": 1.7803,
+      "step": 2028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.893826002867247e-06,
+      "loss": 1.5528,
+      "step": 2029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.889126916386744e-06,
+      "loss": 1.6266,
+      "step": 2030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.884430097628984e-06,
+      "loss": 1.6479,
+      "step": 2031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.879735551639787e-06,
+      "loss": 1.4231,
+      "step": 2032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8750432834625312e-06,
+      "loss": 1.6511,
+      "step": 2033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8703532981381437e-06,
+      "loss": 1.5304,
+      "step": 2034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8656656007051055e-06,
+      "loss": 1.5061,
+      "step": 2035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.860980196199432e-06,
+      "loss": 1.5662,
+      "step": 2036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8562970896546815e-06,
+      "loss": 1.5188,
+      "step": 2037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8516162861019437e-06,
+      "loss": 1.6005,
+      "step": 2038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.846937790569828e-06,
+      "loss": 1.6323,
+      "step": 2039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.84226160808447e-06,
+      "loss": 1.6899,
+      "step": 2040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.837587743669521e-06,
+      "loss": 1.6201,
+      "step": 2041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8329162023461355e-06,
+      "loss": 1.6309,
+      "step": 2042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.82824698913298e-06,
+      "loss": 1.6008,
+      "step": 2043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.823580109046212e-06,
+      "loss": 1.6241,
+      "step": 2044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8189155670994913e-06,
+      "loss": 1.6329,
+      "step": 2045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.814253368303961e-06,
+      "loss": 1.6706,
+      "step": 2046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.809593517668243e-06,
+      "loss": 1.4833,
+      "step": 2047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.804936020198447e-06,
+      "loss": 1.632,
+      "step": 2048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.800280880898143e-06,
+      "loss": 1.5664,
+      "step": 2049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.795628104768376e-06,
+      "loss": 1.5544,
+      "step": 2050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.79097769680765e-06,
+      "loss": 1.5538,
+      "step": 2051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7863296620119217e-06,
+      "loss": 1.5682,
+      "step": 2052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.781684005374604e-06,
+      "loss": 1.5677,
+      "step": 2053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.777040731886549e-06,
+      "loss": 1.5134,
+      "step": 2054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7723998465360537e-06,
+      "loss": 1.633,
+      "step": 2055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7677613543088432e-06,
+      "loss": 1.6687,
+      "step": 2056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7631252601880816e-06,
+      "loss": 1.5746,
+      "step": 2057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7584915691543444e-06,
+      "loss": 1.695,
+      "step": 2058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.753860286185637e-06,
+      "loss": 1.6233,
+      "step": 2059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7492314162573687e-06,
+      "loss": 1.6913,
+      "step": 2060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.744604964342364e-06,
+      "loss": 1.5921,
+      "step": 2061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7399809354108415e-06,
+      "loss": 1.6041,
+      "step": 2062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.735359334430424e-06,
+      "loss": 1.569,
+      "step": 2063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7307401663661247e-06,
+      "loss": 1.525,
+      "step": 2064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7261234361803383e-06,
+      "loss": 1.6906,
+      "step": 2065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.721509148832847e-06,
+      "loss": 1.5436,
+      "step": 2066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7168973092808025e-06,
+      "loss": 1.5389,
+      "step": 2067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7122879224787315e-06,
+      "loss": 1.5701,
+      "step": 2068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7076809933785254e-06,
+      "loss": 1.7062,
+      "step": 2069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.70307652692943e-06,
+      "loss": 1.472,
+      "step": 2070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6984745280780524e-06,
+      "loss": 1.5524,
+      "step": 2071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6938750017683457e-06,
+      "loss": 1.6388,
+      "step": 2072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6892779529416045e-06,
+      "loss": 1.4941,
+      "step": 2073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6846833865364674e-06,
+      "loss": 1.5936,
+      "step": 2074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6800913074888984e-06,
+      "loss": 1.5031,
+      "step": 2075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6755017207321964e-06,
+      "loss": 1.6461,
+      "step": 2076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6709146311969813e-06,
+      "loss": 1.4936,
+      "step": 2077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.666330043811185e-06,
+      "loss": 1.5957,
+      "step": 2078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.66174796350006e-06,
+      "loss": 1.5631,
+      "step": 2079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.657168395186157e-06,
+      "loss": 1.6786,
+      "step": 2080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6525913437893346e-06,
+      "loss": 1.5617,
+      "step": 2081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.648016814226742e-06,
+      "loss": 1.4614,
+      "step": 2082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6434448114128252e-06,
+      "loss": 1.5636,
+      "step": 2083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6388753402593083e-06,
+      "loss": 1.607,
+      "step": 2084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6343084056752032e-06,
+      "loss": 1.5749,
+      "step": 2085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6297440125667904e-06,
+      "loss": 1.6701,
+      "step": 2086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6251821658376265e-06,
+      "loss": 1.6162,
+      "step": 2087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.620622870388524e-06,
+      "loss": 1.7937,
+      "step": 2088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.616066131117563e-06,
+      "loss": 1.5598,
+      "step": 2089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6115119529200748e-06,
+      "loss": 1.5,
+      "step": 2090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6069603406886347e-06,
+      "loss": 1.4747,
+      "step": 2091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.60241129931307e-06,
+      "loss": 1.5971,
+      "step": 2092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.597864833680436e-06,
+      "loss": 1.6085,
+      "step": 2093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.593320948675029e-06,
+      "loss": 1.6379,
+      "step": 2094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.588779649178371e-06,
+      "loss": 1.6096,
+      "step": 2095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5842409400692026e-06,
+      "loss": 1.5529,
+      "step": 2096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.579704826223488e-06,
+      "loss": 1.6569,
+      "step": 2097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.575171312514395e-06,
+      "loss": 1.6568,
+      "step": 2098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.570640403812306e-06,
+      "loss": 1.6347,
+      "step": 2099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5661121049848026e-06,
+      "loss": 1.5687,
+      "step": 2100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5615864208966573e-06,
+      "loss": 1.6021,
+      "step": 2101
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.55706335640984e-06,
+      "loss": 1.5642,
+      "step": 2102
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.552542916383507e-06,
+      "loss": 1.7503,
+      "step": 2103
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5480251056739874e-06,
+      "loss": 1.5004,
+      "step": 2104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.543509929134794e-06,
+      "loss": 1.6034,
+      "step": 2105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5389973916166037e-06,
+      "loss": 1.6437,
+      "step": 2106
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.534487497967262e-06,
+      "loss": 1.6638,
+      "step": 2107
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.529980253031774e-06,
+      "loss": 1.6919,
+      "step": 2108
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5254756616522953e-06,
+      "loss": 1.6123,
+      "step": 2109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5209737286681367e-06,
+      "loss": 1.5654,
+      "step": 2110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5164744589157488e-06,
+      "loss": 1.6592,
+      "step": 2111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5119778572287195e-06,
+      "loss": 1.7181,
+      "step": 2112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5074839284377774e-06,
+      "loss": 1.5571,
+      "step": 2113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5029926773707713e-06,
+      "loss": 1.6283,
+      "step": 2114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.49850410885268e-06,
+      "loss": 1.725,
+      "step": 2115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4940182277055987e-06,
+      "loss": 1.691,
+      "step": 2116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4895350387487304e-06,
+      "loss": 1.4791,
+      "step": 2117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.485054546798395e-06,
+      "loss": 1.7566,
+      "step": 2118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4805767566680057e-06,
+      "loss": 1.5053,
+      "step": 2119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4761016731680792e-06,
+      "loss": 1.5782,
+      "step": 2120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4716293011062248e-06,
+      "loss": 1.6324,
+      "step": 2121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.467159645287133e-06,
+      "loss": 1.6691,
+      "step": 2122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4626927105125834e-06,
+      "loss": 1.6339,
+      "step": 2123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4582285015814263e-06,
+      "loss": 1.5699,
+      "step": 2124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4537670232895866e-06,
+      "loss": 1.7653,
+      "step": 2125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4493082804300585e-06,
+      "loss": 1.5138,
+      "step": 2126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4448522777928903e-06,
+      "loss": 1.6146,
+      "step": 2127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4403990201651915e-06,
+      "loss": 1.4915,
+      "step": 2128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.435948512331125e-06,
+      "loss": 1.5849,
+      "step": 2129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4315007590718913e-06,
+      "loss": 1.5395,
+      "step": 2130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.427055765165741e-06,
+      "loss": 1.5716,
+      "step": 2131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4226135353879516e-06,
+      "loss": 1.5153,
+      "step": 2132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4181740745108377e-06,
+      "loss": 1.5694,
+      "step": 2133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.413737387303739e-06,
+      "loss": 1.6034,
+      "step": 2134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4093034785330087e-06,
+      "loss": 1.722,
+      "step": 2135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4048723529620246e-06,
+      "loss": 1.5806,
+      "step": 2136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4004440153511642e-06,
+      "loss": 1.5454,
+      "step": 2137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.396018470457821e-06,
+      "loss": 1.6908,
+      "step": 2138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3915957230363783e-06,
+      "loss": 1.6415,
+      "step": 2139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3871757778382216e-06,
+      "loss": 1.5411,
+      "step": 2140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3827586396117207e-06,
+      "loss": 1.5847,
+      "step": 2141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.378344313102231e-06,
+      "loss": 1.5364,
+      "step": 2142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.373932803052089e-06,
+      "loss": 1.7208,
+      "step": 2143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.369524114200607e-06,
+      "loss": 1.6667,
+      "step": 2144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3651182512840604e-06,
+      "loss": 1.6104,
+      "step": 2145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.360715219035694e-06,
+      "loss": 1.5494,
+      "step": 2146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.356315022185712e-06,
+      "loss": 1.5639,
+      "step": 2147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3519176654612657e-06,
+      "loss": 1.5873,
+      "step": 2148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3475231535864653e-06,
+      "loss": 1.5505,
+      "step": 2149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3431314912823543e-06,
+      "loss": 1.6494,
+      "step": 2150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.338742683266923e-06,
+      "loss": 1.5272,
+      "step": 2151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3343567342550933e-06,
+      "loss": 1.487,
+      "step": 2152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.329973648958712e-06,
+      "loss": 1.5729,
+      "step": 2153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3255934320865555e-06,
+      "loss": 1.5554,
+      "step": 2154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3212160883443107e-06,
+      "loss": 1.5707,
+      "step": 2155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.316841622434586e-06,
+      "loss": 1.5535,
+      "step": 2156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3124700390568945e-06,
+      "loss": 1.4876,
+      "step": 2157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.30810134290765e-06,
+      "loss": 1.6304,
+      "step": 2158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3037355386801683e-06,
+      "loss": 1.5956,
+      "step": 2159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2993726310646603e-06,
+      "loss": 1.6917,
+      "step": 2160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2950126247482178e-06,
+      "loss": 1.4995,
+      "step": 2161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2906555244148233e-06,
+      "loss": 1.5996,
+      "step": 2162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2863013347453305e-06,
+      "loss": 1.7063,
+      "step": 2163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2819500604174733e-06,
+      "loss": 1.7298,
+      "step": 2164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.277601706105847e-06,
+      "loss": 1.6094,
+      "step": 2165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2732562764819157e-06,
+      "loss": 1.5409,
+      "step": 2166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2689137762139952e-06,
+      "loss": 1.6754,
+      "step": 2167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.264574209967262e-06,
+      "loss": 1.7117,
+      "step": 2168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.260237582403732e-06,
+      "loss": 1.5524,
+      "step": 2169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2559038981822724e-06,
+      "loss": 1.6792,
+      "step": 2170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2515731619585814e-06,
+      "loss": 1.665,
+      "step": 2171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.247245378385195e-06,
+      "loss": 1.74,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.242920552111473e-06,
+      "loss": 1.6653,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.238598687783603e-06,
+      "loss": 1.5854,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.234279790044588e-06,
+      "loss": 1.6422,
+      "step": 2175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.229963863534241e-06,
+      "loss": 1.5584,
+      "step": 2176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.225650912889188e-06,
+      "loss": 1.6532,
+      "step": 2177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.221340942742858e-06,
+      "loss": 1.5635,
+      "step": 2178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2170339577254714e-06,
+      "loss": 1.5618,
+      "step": 2179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.212729962464051e-06,
+      "loss": 1.6646,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.208428961582399e-06,
+      "loss": 1.5992,
+      "step": 2181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2041309597011057e-06,
+      "loss": 1.6488,
+      "step": 2182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1998359614375412e-06,
+      "loss": 1.6933,
+      "step": 2183
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1955439714058422e-06,
+      "loss": 1.694,
+      "step": 2184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.191254994216922e-06,
+      "loss": 1.6662,
+      "step": 2185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.186969034478448e-06,
+      "loss": 1.5631,
+      "step": 2186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.182686096794852e-06,
+      "loss": 1.67,
+      "step": 2187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1784061857673217e-06,
+      "loss": 1.6255,
+      "step": 2188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.174129305993784e-06,
+      "loss": 1.5811,
+      "step": 2189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1698554620689178e-06,
+      "loss": 1.6702,
+      "step": 2190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.165584658584138e-06,
+      "loss": 1.5774,
+      "step": 2191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.16131690012759e-06,
+      "loss": 1.6555,
+      "step": 2192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.157052191284154e-06,
+      "loss": 1.6704,
+      "step": 2193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1527905366354292e-06,
+      "loss": 1.7302,
+      "step": 2194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1485319407597315e-06,
+      "loss": 1.5809,
+      "step": 2195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1442764082321e-06,
+      "loss": 1.6147,
+      "step": 2196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.140023943624272e-06,
+      "loss": 1.7422,
+      "step": 2197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.135774551504698e-06,
+      "loss": 1.6347,
+      "step": 2198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1315282364385197e-06,
+      "loss": 1.6039,
+      "step": 2199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1272850029875802e-06,
+      "loss": 1.5069,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1230448557104087e-06,
+      "loss": 1.6785,
+      "step": 2201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1188077991622174e-06,
+      "loss": 1.5772,
+      "step": 2202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1145738378949004e-06,
+      "loss": 1.648,
+      "step": 2203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.110342976457029e-06,
+      "loss": 1.599,
+      "step": 2204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1061152193938355e-06,
+      "loss": 1.7011,
+      "step": 2205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1018905712472285e-06,
+      "loss": 1.5988,
+      "step": 2206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0976690365557673e-06,
+      "loss": 1.523,
+      "step": 2207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.093450619854671e-06,
+      "loss": 1.6776,
+      "step": 2208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0892353256758107e-06,
+      "loss": 1.6115,
+      "step": 2209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0850231585476965e-06,
+      "loss": 1.6764,
+      "step": 2210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0808141229954876e-06,
+      "loss": 1.6836,
+      "step": 2211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0766082235409695e-06,
+      "loss": 1.7275,
+      "step": 2212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.072405464702566e-06,
+      "loss": 1.5174,
+      "step": 2213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.068205850995326e-06,
+      "loss": 1.5669,
+      "step": 2214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.064009386930915e-06,
+      "loss": 1.7016,
+      "step": 2215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0598160770176208e-06,
+      "loss": 1.5774,
+      "step": 2216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0556259257603355e-06,
+      "loss": 1.5522,
+      "step": 2217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0514389376605646e-06,
+      "loss": 1.6272,
+      "step": 2218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0472551172164152e-06,
+      "loss": 1.4924,
+      "step": 2219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0430744689225833e-06,
+      "loss": 1.7189,
+      "step": 2220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0388969972703688e-06,
+      "loss": 1.6264,
+      "step": 2221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0347227067476478e-06,
+      "loss": 1.668,
+      "step": 2222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.030551601838887e-06,
+      "loss": 1.4476,
+      "step": 2223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0263836870251277e-06,
+      "loss": 1.5715,
+      "step": 2224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0222189667839805e-06,
+      "loss": 1.5078,
+      "step": 2225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.01805744558963e-06,
+      "loss": 1.5552,
+      "step": 2226
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.013899127912824e-06,
+      "loss": 1.7656,
+      "step": 2227
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.009744018220863e-06,
+      "loss": 1.6145,
+      "step": 2228
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.005592120977606e-06,
+      "loss": 1.5538,
+      "step": 2229
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0014434406434584e-06,
+      "loss": 1.6134,
+      "step": 2230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9972979816753717e-06,
+      "loss": 1.6804,
+      "step": 2231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9931557485268365e-06,
+      "loss": 1.5979,
+      "step": 2232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9890167456478748e-06,
+      "loss": 1.572,
+      "step": 2233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.984880977485041e-06,
+      "loss": 1.6239,
+      "step": 2234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.980748448481416e-06,
+      "loss": 1.4955,
+      "step": 2235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9766191630765964e-06,
+      "loss": 1.6635,
+      "step": 2236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9724931257066988e-06,
+      "loss": 1.5594,
+      "step": 2237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9683703408043447e-06,
+      "loss": 1.7438,
+      "step": 2238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9642508127986676e-06,
+      "loss": 1.5651,
+      "step": 2239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9601345461153005e-06,
+      "loss": 1.6352,
+      "step": 2240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9560215451763685e-06,
+      "loss": 1.6038,
+      "step": 2241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.951911814400495e-06,
+      "loss": 1.5489,
+      "step": 2242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9478053582027826e-06,
+      "loss": 1.5338,
+      "step": 2243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9437021809948232e-06,
+      "loss": 1.5691,
+      "step": 2244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9396022871846836e-06,
+      "loss": 1.5162,
+      "step": 2245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.935505681176899e-06,
+      "loss": 1.5069,
+      "step": 2246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9314123673724805e-06,
+      "loss": 1.7365,
+      "step": 2247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9273223501688943e-06,
+      "loss": 1.571,
+      "step": 2248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9232356339600717e-06,
+      "loss": 1.665,
+      "step": 2249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.919152223136391e-06,
+      "loss": 1.6969,
+      "step": 2250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9150721220846884e-06,
+      "loss": 1.4743,
+      "step": 2251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.910995335188234e-06,
+      "loss": 1.6338,
+      "step": 2252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.906921866826747e-06,
+      "loss": 1.5861,
+      "step": 2253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9028517213763737e-06,
+      "loss": 1.5122,
+      "step": 2254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8987849032096973e-06,
+      "loss": 1.7421,
+      "step": 2255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.89472141669572e-06,
+      "loss": 1.6815,
+      "step": 2256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8906612661998698e-06,
+      "loss": 1.4872,
+      "step": 2257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8866044560839902e-06,
+      "loss": 1.6099,
+      "step": 2258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8825509907063328e-06,
+      "loss": 1.7033,
+      "step": 2259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8785008744215606e-06,
+      "loss": 1.5059,
+      "step": 2260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.874454111580733e-06,
+      "loss": 1.5844,
+      "step": 2261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8704107065313116e-06,
+      "loss": 1.5065,
+      "step": 2262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8663706636171503e-06,
+      "loss": 1.6844,
+      "step": 2263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8623339871784869e-06,
+      "loss": 1.5492,
+      "step": 2264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8583006815519473e-06,
+      "loss": 1.5815,
+      "step": 2265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8542707510705355e-06,
+      "loss": 1.5846,
+      "step": 2266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8502442000636246e-06,
+      "loss": 1.6744,
+      "step": 2267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.846221032856965e-06,
+      "loss": 1.6188,
+      "step": 2268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8422012537726646e-06,
+      "loss": 1.6304,
+      "step": 2269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8381848671291953e-06,
+      "loss": 1.6508,
+      "step": 2270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8341718772413852e-06,
+      "loss": 1.7354,
+      "step": 2271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8301622884204096e-06,
+      "loss": 1.621,
+      "step": 2272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8261561049737946e-06,
+      "loss": 1.6612,
+      "step": 2273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8221533312054024e-06,
+      "loss": 1.5463,
+      "step": 2274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.818153971415439e-06,
+      "loss": 1.5638,
+      "step": 2275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8141580299004342e-06,
+      "loss": 1.6305,
+      "step": 2276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8101655109532552e-06,
+      "loss": 1.5503,
+      "step": 2277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8061764188630831e-06,
+      "loss": 1.7257,
+      "step": 2278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8021907579154257e-06,
+      "loss": 1.5063,
+      "step": 2279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7982085323920973e-06,
+      "loss": 1.5481,
+      "step": 2280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7942297465712282e-06,
+      "loss": 1.6086,
+      "step": 2281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7902544047272468e-06,
+      "loss": 1.5333,
+      "step": 2282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7862825111308873e-06,
+      "loss": 1.5672,
+      "step": 2283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7823140700491786e-06,
+      "loss": 1.5426,
+      "step": 2284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7783490857454354e-06,
+      "loss": 1.6521,
+      "step": 2285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7743875624792662e-06,
+      "loss": 1.6993,
+      "step": 2286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770429504506554e-06,
+      "loss": 1.6648,
+      "step": 2287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7664749160794642e-06,
+      "loss": 1.5614,
+      "step": 2288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7625238014464358e-06,
+      "loss": 1.5405,
+      "step": 2289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7585761648521688e-06,
+      "loss": 1.7136,
+      "step": 2290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7546320105376346e-06,
+      "loss": 1.5184,
+      "step": 2291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.750691342740058e-06,
+      "loss": 1.6044,
+      "step": 2292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.746754165692921e-06,
+      "loss": 1.6213,
+      "step": 2293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.742820483625957e-06,
+      "loss": 1.5408,
+      "step": 2294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7388903007651398e-06,
+      "loss": 1.5881,
+      "step": 2295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7349636213326876e-06,
+      "loss": 1.5233,
+      "step": 2296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7310404495470557e-06,
+      "loss": 1.509,
+      "step": 2297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.727120789622927e-06,
+      "loss": 1.6249,
+      "step": 2298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7232046457712164e-06,
+      "loss": 1.5483,
+      "step": 2299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7192920221990566e-06,
+      "loss": 1.3939,
+      "step": 2300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7153829231098018e-06,
+      "loss": 1.5863,
+      "step": 2301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7114773527030215e-06,
+      "loss": 1.4046,
+      "step": 2302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7075753151744885e-06,
+      "loss": 1.6445,
+      "step": 2303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7036768147161853e-06,
+      "loss": 1.608,
+      "step": 2304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6997818555162915e-06,
+      "loss": 1.5709,
+      "step": 2305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6958904417591853e-06,
+      "loss": 1.7657,
+      "step": 2306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6920025776254334e-06,
+      "loss": 1.5101,
+      "step": 2307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6881182672917879e-06,
+      "loss": 1.6193,
+      "step": 2308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6842375149311868e-06,
+      "loss": 1.6569,
+      "step": 2309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.680360324712746e-06,
+      "loss": 1.5992,
+      "step": 2310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6764867008017493e-06,
+      "loss": 1.6337,
+      "step": 2311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.672616647359655e-06,
+      "loss": 1.5431,
+      "step": 2312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.668750168544081e-06,
+      "loss": 1.5778,
+      "step": 2313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.664887268508808e-06,
+      "loss": 1.5901,
+      "step": 2314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6610279514037725e-06,
+      "loss": 1.6373,
+      "step": 2315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.657172221375058e-06,
+      "loss": 1.4957,
+      "step": 2316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6533200825648993e-06,
+      "loss": 1.6442,
+      "step": 2317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6494715391116671e-06,
+      "loss": 1.6952,
+      "step": 2318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6456265951498763e-06,
+      "loss": 1.652,
+      "step": 2319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.641785254810172e-06,
+      "loss": 1.6832,
+      "step": 2320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6379475222193248e-06,
+      "loss": 1.6395,
+      "step": 2321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6341134015002352e-06,
+      "loss": 1.5771,
+      "step": 2322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6302828967719175e-06,
+      "loss": 1.5164,
+      "step": 2323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.626456012149506e-06,
+      "loss": 1.5889,
+      "step": 2324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6226327517442453e-06,
+      "loss": 1.6396,
+      "step": 2325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6188131196634827e-06,
+      "loss": 1.736,
+      "step": 2326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6149971200106723e-06,
+      "loss": 1.5023,
+      "step": 2327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6111847568853645e-06,
+      "loss": 1.5475,
+      "step": 2328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6073760343831996e-06,
+      "loss": 1.6762,
+      "step": 2329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.603570956595913e-06,
+      "loss": 1.5549,
+      "step": 2330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5997695276113168e-06,
+      "loss": 1.4462,
+      "step": 2331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.595971751513311e-06,
+      "loss": 1.4966,
+      "step": 2332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5921776323818655e-06,
+      "loss": 1.6105,
+      "step": 2333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5883871742930257e-06,
+      "loss": 1.6357,
+      "step": 2334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5846003813188993e-06,
+      "loss": 1.4789,
+      "step": 2335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5808172575276615e-06,
+      "loss": 1.7178,
+      "step": 2336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5770378069835412e-06,
+      "loss": 1.5347,
+      "step": 2337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5732620337468258e-06,
+      "loss": 1.6293,
+      "step": 2338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5694899418738462e-06,
+      "loss": 1.5598,
+      "step": 2339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5657215354169841e-06,
+      "loss": 1.5746,
+      "step": 2340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.561956818424661e-06,
+      "loss": 1.6466,
+      "step": 2341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5581957949413295e-06,
+      "loss": 1.4919,
+      "step": 2342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.554438469007482e-06,
+      "loss": 1.606,
+      "step": 2343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5506848446596317e-06,
+      "loss": 1.4773,
+      "step": 2344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.546934925930319e-06,
+      "loss": 1.4985,
+      "step": 2345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5431887168481051e-06,
+      "loss": 1.607,
+      "step": 2346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5394462214375593e-06,
+      "loss": 1.4736,
+      "step": 2347
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5357074437192688e-06,
+      "loss": 1.5695,
+      "step": 2348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5319723877098202e-06,
+      "loss": 1.6946,
+      "step": 2349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5282410574218072e-06,
+      "loss": 1.6238,
+      "step": 2350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5245134568638197e-06,
+      "loss": 1.6152,
+      "step": 2351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5207895900404363e-06,
+      "loss": 1.5869,
+      "step": 2352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5170694609522306e-06,
+      "loss": 1.6752,
+      "step": 2353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5133530735957586e-06,
+      "loss": 1.5203,
+      "step": 2354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5096404319635533e-06,
+      "loss": 1.5673,
+      "step": 2355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.50593154004413e-06,
+      "loss": 1.664,
+      "step": 2356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.502226401821968e-06,
+      "loss": 1.675,
+      "step": 2357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.498525021277521e-06,
+      "loss": 1.6049,
+      "step": 2358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4948274023872005e-06,
+      "loss": 1.6631,
+      "step": 2359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4911335491233818e-06,
+      "loss": 1.6079,
+      "step": 2360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.487443465454389e-06,
+      "loss": 1.7145,
+      "step": 2361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.483757155344503e-06,
+      "loss": 1.5555,
+      "step": 2362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4800746227539437e-06,
+      "loss": 1.5893,
+      "step": 2363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4763958716388798e-06,
+      "loss": 1.7358,
+      "step": 2364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4727209059514114e-06,
+      "loss": 1.5669,
+      "step": 2365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4690497296395773e-06,
+      "loss": 1.608,
+      "step": 2366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.46538234664734e-06,
+      "loss": 1.5754,
+      "step": 2367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4617187609145906e-06,
+      "loss": 1.6161,
+      "step": 2368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4580589763771413e-06,
+      "loss": 1.6159,
+      "step": 2369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4544029969667167e-06,
+      "loss": 1.7332,
+      "step": 2370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4507508266109565e-06,
+      "loss": 1.6452,
+      "step": 2371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4471024692334101e-06,
+      "loss": 1.6325,
+      "step": 2372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4434579287535244e-06,
+      "loss": 1.6168,
+      "step": 2373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.439817209086653e-06,
+      "loss": 1.6185,
+      "step": 2374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4361803141440384e-06,
+      "loss": 1.5649,
+      "step": 2375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.432547247832819e-06,
+      "loss": 1.6146,
+      "step": 2376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4289180140560189e-06,
+      "loss": 1.5704,
+      "step": 2377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4252926167125413e-06,
+      "loss": 1.6511,
+      "step": 2378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.421671059697175e-06,
+      "loss": 1.5154,
+      "step": 2379
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.418053346900574e-06,
+      "loss": 1.6446,
+      "step": 2380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4144394822092712e-06,
+      "loss": 1.5399,
+      "step": 2381
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4108294695056606e-06,
+      "loss": 1.7164,
+      "step": 2382
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4072233126679985e-06,
+      "loss": 1.6045,
+      "step": 2383
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4036210155703989e-06,
+      "loss": 1.57,
+      "step": 2384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4000225820828317e-06,
+      "loss": 1.583,
+      "step": 2385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3964280160711119e-06,
+      "loss": 1.5326,
+      "step": 2386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3928373213969038e-06,
+      "loss": 1.5122,
+      "step": 2387
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.38925050191771e-06,
+      "loss": 1.6602,
+      "step": 2388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3856675614868687e-06,
+      "loss": 1.6222,
+      "step": 2389
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3820885039535564e-06,
+      "loss": 1.5475,
+      "step": 2390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.378513333162771e-06,
+      "loss": 1.6007,
+      "step": 2391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3749420529553414e-06,
+      "loss": 1.6799,
+      "step": 2392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3713746671679112e-06,
+      "loss": 1.7089,
+      "step": 2393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3678111796329446e-06,
+      "loss": 1.6075,
+      "step": 2394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3642515941787171e-06,
+      "loss": 1.6644,
+      "step": 2395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3606959146293086e-06,
+      "loss": 1.6329,
+      "step": 2396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3571441448046086e-06,
+      "loss": 1.6368,
+      "step": 2397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3535962885202997e-06,
+      "loss": 1.6637,
+      "step": 2398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.350052349587866e-06,
+      "loss": 1.6353,
+      "step": 2399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3465123318145817e-06,
+      "loss": 1.6352,
+      "step": 2400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.342976239003505e-06,
+      "loss": 1.5715,
+      "step": 2401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.339444074953482e-06,
+      "loss": 1.6759,
+      "step": 2402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.335915843459137e-06,
+      "loss": 1.5522,
+      "step": 2403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3323915483108662e-06,
+      "loss": 1.5199,
+      "step": 2404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3288711932948427e-06,
+      "loss": 1.6027,
+      "step": 2405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3253547821930002e-06,
+      "loss": 1.6824,
+      "step": 2406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3218423187830409e-06,
+      "loss": 1.5832,
+      "step": 2407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3183338068384243e-06,
+      "loss": 1.52,
+      "step": 2408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3148292501283627e-06,
+      "loss": 1.5823,
+      "step": 2409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3113286524178232e-06,
+      "loss": 1.5462,
+      "step": 2410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3078320174675141e-06,
+      "loss": 1.5337,
+      "step": 2411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3043393490338918e-06,
+      "loss": 1.5027,
+      "step": 2412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3008506508691516e-06,
+      "loss": 1.6562,
+      "step": 2413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2973659267212173e-06,
+      "loss": 1.6728,
+      "step": 2414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2938851803337516e-06,
+      "loss": 1.5061,
+      "step": 2415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.290408415446136e-06,
+      "loss": 1.6714,
+      "step": 2416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2869356357934815e-06,
+      "loss": 1.7041,
+      "step": 2417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2834668451066118e-06,
+      "loss": 1.4821,
+      "step": 2418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2800020471120717e-06,
+      "loss": 1.4711,
+      "step": 2419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.276541245532109e-06,
+      "loss": 1.6455,
+      "step": 2420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2730844440846862e-06,
+      "loss": 1.6003,
+      "step": 2421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2696316464834607e-06,
+      "loss": 1.4373,
+      "step": 2422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2661828564377948e-06,
+      "loss": 1.6041,
+      "step": 2423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2627380776527415e-06,
+      "loss": 1.6866,
+      "step": 2424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.259297313829046e-06,
+      "loss": 1.586,
+      "step": 2425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.255860568663142e-06,
+      "loss": 1.5997,
+      "step": 2426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2524278458471411e-06,
+      "loss": 1.6384,
+      "step": 2427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.248999149068838e-06,
+      "loss": 1.4732,
+      "step": 2428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2455744820117028e-06,
+      "loss": 1.565,
+      "step": 2429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2421538483548706e-06,
+      "loss": 1.4918,
+      "step": 2430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2387372517731505e-06,
+      "loss": 1.5698,
+      "step": 2431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2353246959370086e-06,
+      "loss": 1.6392,
+      "step": 2432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2319161845125744e-06,
+      "loss": 1.7168,
+      "step": 2433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.228511721161631e-06,
+      "loss": 1.5549,
+      "step": 2434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2251113095416113e-06,
+      "loss": 1.6955,
+      "step": 2435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2217149533055976e-06,
+      "loss": 1.6082,
+      "step": 2436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2183226561023132e-06,
+      "loss": 1.67,
+      "step": 2437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2149344215761216e-06,
+      "loss": 1.6297,
+      "step": 2438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2115502533670253e-06,
+      "loss": 1.6476,
+      "step": 2439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2081701551106506e-06,
+      "loss": 1.5208,
+      "step": 2440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.20479413043826e-06,
+      "loss": 1.6243,
+      "step": 2441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.201422182976732e-06,
+      "loss": 1.628,
+      "step": 2442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1980543163485726e-06,
+      "loss": 1.6237,
+      "step": 2443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1946905341718951e-06,
+      "loss": 1.7931,
+      "step": 2444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1913308400604339e-06,
+      "loss": 1.6846,
+      "step": 2445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1879752376235231e-06,
+      "loss": 1.6448,
+      "step": 2446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1846237304661095e-06,
+      "loss": 1.591,
+      "step": 2447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.181276322188732e-06,
+      "loss": 1.687,
+      "step": 2448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1779330163875325e-06,
+      "loss": 1.6717,
+      "step": 2449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1745938166542414e-06,
+      "loss": 1.564,
+      "step": 2450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1712587265761799e-06,
+      "loss": 1.5582,
+      "step": 2451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1679277497362563e-06,
+      "loss": 1.7939,
+      "step": 2452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1646008897129546e-06,
+      "loss": 1.5011,
+      "step": 2453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.161278150080341e-06,
+      "loss": 1.614,
+      "step": 2454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.157959534408052e-06,
+      "loss": 1.5776,
+      "step": 2455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1546450462612951e-06,
+      "loss": 1.6954,
+      "step": 2456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.151334689200845e-06,
+      "loss": 1.659,
+      "step": 2457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1480284667830343e-06,
+      "loss": 1.5621,
+      "step": 2458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1447263825597577e-06,
+      "loss": 1.6126,
+      "step": 2459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1414284400784643e-06,
+      "loss": 1.6307,
+      "step": 2460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1381346428821482e-06,
+      "loss": 1.6873,
+      "step": 2461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.134844994509358e-06,
+      "loss": 1.5413,
+      "step": 2462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1315594984941786e-06,
+      "loss": 1.5293,
+      "step": 2463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1282781583662372e-06,
+      "loss": 1.7324,
+      "step": 2464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1250009776506982e-06,
+      "loss": 1.6114,
+      "step": 2465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1217279598682518e-06,
+      "loss": 1.7008,
+      "step": 2466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.118459108535122e-06,
+      "loss": 1.5569,
+      "step": 2467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1151944271630517e-06,
+      "loss": 1.5414,
+      "step": 2468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1119339192593077e-06,
+      "loss": 1.5905,
+      "step": 2469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1086775883266725e-06,
+      "loss": 1.7105,
+      "step": 2470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1054254378634399e-06,
+      "loss": 1.6459,
+      "step": 2471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.102177471363412e-06,
+      "loss": 1.5179,
+      "step": 2472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0989336923158999e-06,
+      "loss": 1.564,
+      "step": 2473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0956941042057106e-06,
+      "loss": 1.6162,
+      "step": 2474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0924587105131546e-06,
+      "loss": 1.59,
+      "step": 2475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0892275147140307e-06,
+      "loss": 1.6676,
+      "step": 2476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.086000520279632e-06,
+      "loss": 1.6548,
+      "step": 2477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0827777306767384e-06,
+      "loss": 1.6498,
+      "step": 2478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0795591493676072e-06,
+      "loss": 1.6193,
+      "step": 2479
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0763447798099813e-06,
+      "loss": 1.654,
+      "step": 2480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0731346254570735e-06,
+      "loss": 1.6554,
+      "step": 2481
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0699286897575718e-06,
+      "loss": 1.634,
+      "step": 2482
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.066726976155632e-06,
+      "loss": 1.7828,
+      "step": 2483
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0635294880908702e-06,
+      "loss": 1.5812,
+      "step": 2484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0603362289983687e-06,
+      "loss": 1.5456,
+      "step": 2485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0571472023086604e-06,
+      "loss": 1.6422,
+      "step": 2486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.053962411447736e-06,
+      "loss": 1.4697,
+      "step": 2487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0507818598370355e-06,
+      "loss": 1.6882,
+      "step": 2488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0476055508934408e-06,
+      "loss": 1.5398,
+      "step": 2489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0444334880292794e-06,
+      "loss": 1.5794,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0412656746523182e-06,
+      "loss": 1.6534,
+      "step": 2491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0381021141657526e-06,
+      "loss": 1.5437,
+      "step": 2492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0349428099682173e-06,
+      "loss": 1.5763,
+      "step": 2493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0317877654537672e-06,
+      "loss": 1.5724,
+      "step": 2494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0286369840118859e-06,
+      "loss": 1.6792,
+      "step": 2495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.025490469027477e-06,
+      "loss": 1.6445,
+      "step": 2496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0223482238808557e-06,
+      "loss": 1.6286,
+      "step": 2497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0192102519477565e-06,
+      "loss": 1.4849,
+      "step": 2498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.016076556599318e-06,
+      "loss": 1.5428,
+      "step": 2499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0129471412020886e-06,
+      "loss": 1.6629,
+      "step": 2500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0098220091180145e-06,
+      "loss": 1.5312,
+      "step": 2501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.006701163704445e-06,
+      "loss": 1.5965,
+      "step": 2502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0035846083141193e-06,
+      "loss": 1.5276,
+      "step": 2503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0004723462951732e-06,
+      "loss": 1.562,
+      "step": 2504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.973643809911238e-07,
+      "loss": 1.5866,
+      "step": 2505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.942607157408784e-07,
+      "loss": 1.5909,
+      "step": 2506
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.911613538787196e-07,
+      "loss": 1.6012,
+      "step": 2507
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.880662987343103e-07,
+      "loss": 1.5427,
+      "step": 2508
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.849755536326866e-07,
+      "loss": 1.5717,
+      "step": 2509
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.818891218942511e-07,
+      "loss": 1.6301,
+      "step": 2510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.78807006834777e-07,
+      "loss": 1.6187,
+      "step": 2511
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.757292117653955e-07,
+      "loss": 1.601,
+      "step": 2512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.726557399925995e-07,
+      "loss": 1.5945,
+      "step": 2513
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.695865948182392e-07,
+      "loss": 1.6253,
+      "step": 2514
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.66521779539511e-07,
+      "loss": 1.6381,
+      "step": 2515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.63461297448966e-07,
+      "loss": 1.6137,
+      "step": 2516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.604051518344948e-07,
+      "loss": 1.6725,
+      "step": 2517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.57353345979332e-07,
+      "loss": 1.5973,
+      "step": 2518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.543058831620528e-07,
+      "loss": 1.5152,
+      "step": 2519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.512627666565588e-07,
+      "loss": 1.6919,
+      "step": 2520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.482239997320903e-07,
+      "loss": 1.5965,
+      "step": 2521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.451895856532117e-07,
+      "loss": 1.4616,
+      "step": 2522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.421595276798084e-07,
+      "loss": 1.6133,
+      "step": 2523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.39133829067092e-07,
+      "loss": 1.6133,
+      "step": 2524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.361124930655841e-07,
+      "loss": 1.6171,
+      "step": 2525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.330955229211259e-07,
+      "loss": 1.4944,
+      "step": 2526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.300829218748625e-07,
+      "loss": 1.5974,
+      "step": 2527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.270746931632501e-07,
+      "loss": 1.6532,
+      "step": 2528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.240708400180437e-07,
+      "loss": 1.5322,
+      "step": 2529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.210713656663023e-07,
+      "loss": 1.5412,
+      "step": 2530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.180762733303745e-07,
+      "loss": 1.6021,
+      "step": 2531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.150855662279079e-07,
+      "loss": 1.7312,
+      "step": 2532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.120992475718333e-07,
+      "loss": 1.589,
+      "step": 2533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.091173205703708e-07,
+      "loss": 1.5437,
+      "step": 2534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.061397884270217e-07,
+      "loss": 1.5931,
+      "step": 2535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.031666543405637e-07,
+      "loss": 1.6847,
+      "step": 2536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.001979215050544e-07,
+      "loss": 1.5446,
+      "step": 2537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.972335931098159e-07,
+      "loss": 1.7322,
+      "step": 2538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.942736723394458e-07,
+      "loss": 1.5692,
+      "step": 2539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.913181623738032e-07,
+      "loss": 1.6468,
+      "step": 2540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.883670663880078e-07,
+      "loss": 1.7293,
+      "step": 2541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.854203875524403e-07,
+      "loss": 1.5412,
+      "step": 2542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.824781290327317e-07,
+      "loss": 1.6671,
+      "step": 2543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.795402939897679e-07,
+      "loss": 1.6957,
+      "step": 2544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.766068855796833e-07,
+      "loss": 1.4307,
+      "step": 2545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.736779069538521e-07,
+      "loss": 1.6188,
+      "step": 2546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.707533612588948e-07,
+      "loss": 1.683,
+      "step": 2547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.67833251636665e-07,
+      "loss": 1.6611,
+      "step": 2548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.649175812242532e-07,
+      "loss": 1.6936,
+      "step": 2549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.62006353153983e-07,
+      "loss": 1.6478,
+      "step": 2550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.590995705533994e-07,
+      "loss": 1.5501,
+      "step": 2551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.561972365452775e-07,
+      "loss": 1.5887,
+      "step": 2552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.532993542476108e-07,
+      "loss": 1.5271,
+      "step": 2553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.504059267736097e-07,
+      "loss": 1.6199,
+      "step": 2554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.475169572316988e-07,
+      "loss": 1.6079,
+      "step": 2555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.446324487255164e-07,
+      "loss": 1.613,
+      "step": 2556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.417524043539038e-07,
+      "loss": 1.6381,
+      "step": 2557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.388768272109105e-07,
+      "loss": 1.5301,
+      "step": 2558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.36005720385783e-07,
+      "loss": 1.5789,
+      "step": 2559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.331390869629702e-07,
+      "loss": 1.499,
+      "step": 2560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.302769300221098e-07,
+      "loss": 1.5277,
+      "step": 2561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.274192526380337e-07,
+      "loss": 1.5783,
+      "step": 2562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.24566057880763e-07,
+      "loss": 1.5353,
+      "step": 2563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.217173488154972e-07,
+      "loss": 1.7415,
+      "step": 2564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.188731285026219e-07,
+      "loss": 1.6809,
+      "step": 2565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.160333999977004e-07,
+      "loss": 1.6429,
+      "step": 2566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.131981663514665e-07,
+      "loss": 1.6202,
+      "step": 2567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.103674306098291e-07,
+      "loss": 1.6552,
+      "step": 2568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.075411958138623e-07,
+      "loss": 1.7066,
+      "step": 2569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.047194649998063e-07,
+      "loss": 1.6009,
+      "step": 2570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.019022411990634e-07,
+      "loss": 1.5528,
+      "step": 2571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.99089527438191e-07,
+      "loss": 1.529,
+      "step": 2572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.962813267389052e-07,
+      "loss": 1.6891,
+      "step": 2573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.93477642118069e-07,
+      "loss": 1.6863,
+      "step": 2574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.906784765876985e-07,
+      "loss": 1.6193,
+      "step": 2575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.878838331549538e-07,
+      "loss": 1.6259,
+      "step": 2576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.850937148221332e-07,
+      "loss": 1.5036,
+      "step": 2577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.823081245866776e-07,
+      "loss": 1.5858,
+      "step": 2578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.795270654411635e-07,
+      "loss": 1.5539,
+      "step": 2579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.767505403732961e-07,
+      "loss": 1.6778,
+      "step": 2580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.739785523659144e-07,
+      "loss": 1.5718,
+      "step": 2581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.712111043969772e-07,
+      "loss": 1.5892,
+      "step": 2582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.684481994395726e-07,
+      "loss": 1.5812,
+      "step": 2583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.656898404619029e-07,
+      "loss": 1.68,
+      "step": 2584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.629360304272882e-07,
+      "loss": 1.7361,
+      "step": 2585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.601867722941642e-07,
+      "loss": 1.5554,
+      "step": 2586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.57442069016071e-07,
+      "loss": 1.7302,
+      "step": 2587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.547019235416609e-07,
+      "loss": 1.608,
+      "step": 2588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.519663388146886e-07,
+      "loss": 1.5969,
+      "step": 2589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.492353177740047e-07,
+      "loss": 1.5268,
+      "step": 2590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.465088633535639e-07,
+      "loss": 1.6184,
+      "step": 2591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.437869784824086e-07,
+      "loss": 1.6837,
+      "step": 2592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.410696660846761e-07,
+      "loss": 1.5481,
+      "step": 2593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.383569290795911e-07,
+      "loss": 1.5458,
+      "step": 2594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.356487703814602e-07,
+      "loss": 1.6538,
+      "step": 2595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.329451928996745e-07,
+      "loss": 1.4152,
+      "step": 2596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.302461995387033e-07,
+      "loss": 1.6387,
+      "step": 2597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.275517931980886e-07,
+      "loss": 1.6035,
+      "step": 2598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.24861976772448e-07,
+      "loss": 1.567,
+      "step": 2599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.22176753151464e-07,
+      "loss": 1.6947,
+      "step": 2600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.194961252198885e-07,
+      "loss": 1.5318,
+      "step": 2601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.168200958575361e-07,
+      "loss": 1.6329,
+      "step": 2602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.141486679392778e-07,
+      "loss": 1.6358,
+      "step": 2603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.114818443350463e-07,
+      "loss": 1.5244,
+      "step": 2604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.088196279098225e-07,
+      "loss": 1.6105,
+      "step": 2605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.061620215236415e-07,
+      "loss": 1.6615,
+      "step": 2606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.035090280315854e-07,
+      "loss": 1.5521,
+      "step": 2607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.008606502837784e-07,
+      "loss": 1.5421,
+      "step": 2608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.982168911253895e-07,
+      "loss": 1.6623,
+      "step": 2609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.955777533966212e-07,
+      "loss": 1.5921,
+      "step": 2610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.929432399327174e-07,
+      "loss": 1.5534,
+      "step": 2611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.903133535639467e-07,
+      "loss": 1.5474,
+      "step": 2612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.876880971156147e-07,
+      "loss": 1.5784,
+      "step": 2613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.850674734080454e-07,
+      "loss": 1.5743,
+      "step": 2614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.824514852565922e-07,
+      "loss": 1.5465,
+      "step": 2615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.798401354716233e-07,
+      "loss": 1.5384,
+      "step": 2616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.772334268585296e-07,
+      "loss": 1.5664,
+      "step": 2617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.746313622177097e-07,
+      "loss": 1.5399,
+      "step": 2618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.720339443445772e-07,
+      "loss": 1.596,
+      "step": 2619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.694411760295538e-07,
+      "loss": 1.5656,
+      "step": 2620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.66853060058063e-07,
+      "loss": 1.6511,
+      "step": 2621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.642695992105347e-07,
+      "loss": 1.6449,
+      "step": 2622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.61690796262392e-07,
+      "loss": 1.5287,
+      "step": 2623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.591166539840599e-07,
+      "loss": 1.5745,
+      "step": 2624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.565471751409541e-07,
+      "loss": 1.5344,
+      "step": 2625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.539823624934777e-07,
+      "loss": 1.5867,
+      "step": 2626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.514222187970248e-07,
+      "loss": 1.6609,
+      "step": 2627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.488667468019727e-07,
+      "loss": 1.5912,
+      "step": 2628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.46315949253678e-07,
+      "loss": 1.637,
+      "step": 2629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.437698288924777e-07,
+      "loss": 1.41,
+      "step": 2630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.412283884536818e-07,
+      "loss": 1.639,
+      "step": 2631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.38691630667575e-07,
+      "loss": 1.7195,
+      "step": 2632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.36159558259411e-07,
+      "loss": 1.5489,
+      "step": 2633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.336321739494072e-07,
+      "loss": 1.5746,
+      "step": 2634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.31109480452749e-07,
+      "loss": 1.5922,
+      "step": 2635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.285914804795784e-07,
+      "loss": 1.5849,
+      "step": 2636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.260781767349983e-07,
+      "loss": 1.613,
+      "step": 2637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.235695719190632e-07,
+      "loss": 1.6692,
+      "step": 2638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.210656687267835e-07,
+      "loss": 1.5461,
+      "step": 2639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.185664698481137e-07,
+      "loss": 1.637,
+      "step": 2640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.160719779679597e-07,
+      "loss": 1.5215,
+      "step": 2641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.135821957661658e-07,
+      "loss": 1.6651,
+      "step": 2642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.110971259175208e-07,
+      "loss": 1.4043,
+      "step": 2643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.086167710917479e-07,
+      "loss": 1.5082,
+      "step": 2644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.061411339535062e-07,
+      "loss": 1.58,
+      "step": 2645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.036702171623876e-07,
+      "loss": 1.5954,
+      "step": 2646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.012040233729105e-07,
+      "loss": 1.5389,
+      "step": 2647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.987425552345222e-07,
+      "loss": 1.5182,
+      "step": 2648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.962858153915896e-07,
+      "loss": 1.6259,
+      "step": 2649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.938338064834037e-07,
+      "loss": 1.6138,
+      "step": 2650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.913865311441714e-07,
+      "loss": 1.6256,
+      "step": 2651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.889439920030127e-07,
+      "loss": 1.5688,
+      "step": 2652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.865061916839615e-07,
+      "loss": 1.5522,
+      "step": 2653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.840731328059629e-07,
+      "loss": 1.5802,
+      "step": 2654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.816448179828616e-07,
+      "loss": 1.6851,
+      "step": 2655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.792212498234134e-07,
+      "loss": 1.5501,
+      "step": 2656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.768024309312681e-07,
+      "loss": 1.6024,
+      "step": 2657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.74388363904978e-07,
+      "loss": 1.4492,
+      "step": 2658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.719790513379891e-07,
+      "loss": 1.4868,
+      "step": 2659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.695744958186383e-07,
+      "loss": 1.5909,
+      "step": 2660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.671746999301542e-07,
+      "loss": 1.6309,
+      "step": 2661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.647796662506493e-07,
+      "loss": 1.5637,
+      "step": 2662
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.623893973531225e-07,
+      "loss": 1.5836,
+      "step": 2663
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.600038958054538e-07,
+      "loss": 1.582,
+      "step": 2664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.576231641703994e-07,
+      "loss": 1.6625,
+      "step": 2665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.552472050055946e-07,
+      "loss": 1.6273,
+      "step": 2666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.528760208635436e-07,
+      "loss": 1.6851,
+      "step": 2667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.505096142916233e-07,
+      "loss": 1.5299,
+      "step": 2668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.481479878320784e-07,
+      "loss": 1.6337,
+      "step": 2669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.457911440220154e-07,
+      "loss": 1.54,
+      "step": 2670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.434390853934063e-07,
+      "loss": 1.6599,
+      "step": 2671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.410918144730815e-07,
+      "loss": 1.6067,
+      "step": 2672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.387493337827254e-07,
+      "loss": 1.673,
+      "step": 2673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.364116458388802e-07,
+      "loss": 1.6194,
+      "step": 2674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.340787531529346e-07,
+      "loss": 1.5485,
+      "step": 2675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3175065823113e-07,
+      "loss": 1.5714,
+      "step": 2676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.294273635745517e-07,
+      "loss": 1.6813,
+      "step": 2677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.271088716791273e-07,
+      "loss": 1.6173,
+      "step": 2678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.24795185035627e-07,
+      "loss": 1.5613,
+      "step": 2679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.224863061296553e-07,
+      "loss": 1.6119,
+      "step": 2680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.201822374416549e-07,
+      "loss": 1.5847,
+      "step": 2681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.178829814469006e-07,
+      "loss": 1.5663,
+      "step": 2682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.155885406154937e-07,
+      "loss": 1.6592,
+      "step": 2683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.132989174123659e-07,
+      "loss": 1.6127,
+      "step": 2684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.110141142972735e-07,
+      "loss": 1.6727,
+      "step": 2685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.087341337247914e-07,
+      "loss": 1.6968,
+      "step": 2686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.064589781443163e-07,
+      "loss": 1.517,
+      "step": 2687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.041886500000603e-07,
+      "loss": 1.5028,
+      "step": 2688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.019231517310491e-07,
+      "loss": 1.4789,
+      "step": 2689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.996624857711219e-07,
+      "loss": 1.5752,
+      "step": 2690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.97406654548922e-07,
+      "loss": 1.612,
+      "step": 2691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.951556604879049e-07,
+      "loss": 1.6296,
+      "step": 2692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.929095060063227e-07,
+      "loss": 1.6866,
+      "step": 2693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.906681935172342e-07,
+      "loss": 1.6257,
+      "step": 2694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.88431725428492e-07,
+      "loss": 1.6032,
+      "step": 2695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.862001041427488e-07,
+      "loss": 1.5332,
+      "step": 2696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.839733320574457e-07,
+      "loss": 1.5489,
+      "step": 2697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.817514115648164e-07,
+      "loss": 1.6256,
+      "step": 2698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.795343450518825e-07,
+      "loss": 1.6214,
+      "step": 2699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.773221349004531e-07,
+      "loss": 1.6623,
+      "step": 2700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7511478348711447e-07,
+      "loss": 1.5663,
+      "step": 2701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.729122931832392e-07,
+      "loss": 1.6781,
+      "step": 2702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.707146663549744e-07,
+      "loss": 1.6699,
+      "step": 2703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.685219053632423e-07,
+      "loss": 1.6277,
+      "step": 2704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.663340125637389e-07,
+      "loss": 1.5396,
+      "step": 2705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6415099030692914e-07,
+      "loss": 1.6241,
+      "step": 2706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.619728409380453e-07,
+      "loss": 1.6524,
+      "step": 2707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.597995667970878e-07,
+      "loss": 1.5507,
+      "step": 2708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5763117021881467e-07,
+      "loss": 1.6517,
+      "step": 2709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5546765353274846e-07,
+      "loss": 1.73,
+      "step": 2710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5330901906316506e-07,
+      "loss": 1.6117,
+      "step": 2711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.511552691290988e-07,
+      "loss": 1.6602,
+      "step": 2712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.490064060443361e-07,
+      "loss": 1.505,
+      "step": 2713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.468624321174109e-07,
+      "loss": 1.619,
+      "step": 2714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4472334965160736e-07,
+      "loss": 1.6464,
+      "step": 2715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4258916094495394e-07,
+      "loss": 1.71,
+      "step": 2716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4045986829022e-07,
+      "loss": 1.5612,
+      "step": 2717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.38335473974919e-07,
+      "loss": 1.5003,
+      "step": 2718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.362159802812971e-07,
+      "loss": 1.5909,
+      "step": 2719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.341013894863405e-07,
+      "loss": 1.5871,
+      "step": 2720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3199170386176325e-07,
+      "loss": 1.6043,
+      "step": 2721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2988692567401515e-07,
+      "loss": 1.5989,
+      "step": 2722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2778705718426907e-07,
+      "loss": 1.5794,
+      "step": 2723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2569210064842716e-07,
+      "loss": 1.6052,
+      "step": 2724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.236020583171108e-07,
+      "loss": 1.6019,
+      "step": 2725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.215169324356666e-07,
+      "loss": 1.5305,
+      "step": 2726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.194367252441545e-07,
+      "loss": 1.5929,
+      "step": 2727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1736143897735394e-07,
+      "loss": 1.6245,
+      "step": 2728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.152910758647577e-07,
+      "loss": 1.6468,
+      "step": 2729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1322563813056606e-07,
+      "loss": 1.6491,
+      "step": 2730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.111651279936929e-07,
+      "loss": 1.7194,
+      "step": 2731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.091095476677531e-07,
+      "loss": 1.5901,
+      "step": 2732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.070588993610697e-07,
+      "loss": 1.5902,
+      "step": 2733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.050131852766659e-07,
+      "loss": 1.6057,
+      "step": 2734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.029724076122621e-07,
+      "loss": 1.7033,
+      "step": 2735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.009365685602795e-07,
+      "loss": 1.6929,
+      "step": 2736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.989056703078292e-07,
+      "loss": 1.6814,
+      "step": 2737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.968797150367171e-07,
+      "loss": 1.6154,
+      "step": 2738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.948587049234398e-07,
+      "loss": 1.6399,
+      "step": 2739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.928426421391773e-07,
+      "loss": 1.6154,
+      "step": 2740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9083152884979935e-07,
+      "loss": 1.5438,
+      "step": 2741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8882536721585486e-07,
+      "loss": 1.5059,
+      "step": 2742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.868241593925742e-07,
+      "loss": 1.5702,
+      "step": 2743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.848279075298678e-07,
+      "loss": 1.5766,
+      "step": 2744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.828366137723183e-07,
+      "loss": 1.6146,
+      "step": 2745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.80850280259184e-07,
+      "loss": 1.5185,
+      "step": 2746
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7886890912439633e-07,
+      "loss": 1.5463,
+      "step": 2747
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.768925024965503e-07,
+      "loss": 1.5708,
+      "step": 2748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.749210624989125e-07,
+      "loss": 1.5904,
+      "step": 2749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.729545912494115e-07,
+      "loss": 1.5723,
+      "step": 2750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7099309086063794e-07,
+      "loss": 1.5676,
+      "step": 2751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6903656343984293e-07,
+      "loss": 1.5238,
+      "step": 2752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.670850110889346e-07,
+      "loss": 1.6625,
+      "step": 2753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.651384359044774e-07,
+      "loss": 1.6632,
+      "step": 2754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.631968399776864e-07,
+      "loss": 1.6443,
+      "step": 2755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6126022539442975e-07,
+      "loss": 1.6814,
+      "step": 2756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.593285942352237e-07,
+      "loss": 1.6169,
+      "step": 2757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5740194857523e-07,
+      "loss": 1.5941,
+      "step": 2758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.554802904842547e-07,
+      "loss": 1.626,
+      "step": 2759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5356362202674687e-07,
+      "loss": 1.5555,
+      "step": 2760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.516519452617922e-07,
+      "loss": 1.6431,
+      "step": 2761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4974526224311744e-07,
+      "loss": 1.6273,
+      "step": 2762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.478435750190817e-07,
+      "loss": 1.739,
+      "step": 2763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.459468856326792e-07,
+      "loss": 1.613,
+      "step": 2764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4405519612153326e-07,
+      "loss": 1.6836,
+      "step": 2765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4216850851789663e-07,
+      "loss": 1.6776,
+      "step": 2766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.402868248486485e-07,
+      "loss": 1.5847,
+      "step": 2767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3841014713529184e-07,
+      "loss": 1.5867,
+      "step": 2768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3653847739395174e-07,
+      "loss": 1.598,
+      "step": 2769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.346718176353747e-07,
+      "loss": 1.6628,
+      "step": 2770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3281016986492165e-07,
+      "loss": 1.6799,
+      "step": 2771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3095353608257385e-07,
+      "loss": 1.5993,
+      "step": 2772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2910191828292083e-07,
+      "loss": 1.562,
+      "step": 2773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2725531845516744e-07,
+      "loss": 1.6116,
+      "step": 2774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.254137385831263e-07,
+      "loss": 1.612,
+      "step": 2775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2357718064521594e-07,
+      "loss": 1.589,
+      "step": 2776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.217456466144614e-07,
+      "loss": 1.6587,
+      "step": 2777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.199191384584893e-07,
+      "loss": 1.4991,
+      "step": 2778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.180976581395295e-07,
+      "loss": 1.4498,
+      "step": 2779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1628120761440616e-07,
+      "loss": 1.5478,
+      "step": 2780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.144697888345427e-07,
+      "loss": 1.5986,
+      "step": 2781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1266340374595693e-07,
+      "loss": 1.6321,
+      "step": 2782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.108620542892593e-07,
+      "loss": 1.5104,
+      "step": 2783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0906574239964795e-07,
+      "loss": 1.5906,
+      "step": 2784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.072744700069119e-07,
+      "loss": 1.7019,
+      "step": 2785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.054882390354241e-07,
+      "loss": 1.6075,
+      "step": 2786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0370705140414293e-07,
+      "loss": 1.612,
+      "step": 2787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.019309090266087e-07,
+      "loss": 1.5754,
+      "step": 2788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0015981381094073e-07,
+      "loss": 1.5136,
+      "step": 2789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9839376765983583e-07,
+      "loss": 1.4912,
+      "step": 2790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9663277247056923e-07,
+      "loss": 1.6618,
+      "step": 2791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9487683013498523e-07,
+      "loss": 1.5716,
+      "step": 2792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.93125942539505e-07,
+      "loss": 1.6651,
+      "step": 2793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.913801115651144e-07,
+      "loss": 1.5752,
+      "step": 2794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.896393390873714e-07,
+      "loss": 1.5222,
+      "step": 2795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8790362697639685e-07,
+      "loss": 1.6795,
+      "step": 2796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8617297709687577e-07,
+      "loss": 1.5624,
+      "step": 2797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8444739130805587e-07,
+      "loss": 1.5697,
+      "step": 2798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.827268714637421e-07,
+      "loss": 1.6859,
+      "step": 2799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.810114194122998e-07,
+      "loss": 1.5717,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.793010369966487e-07,
+      "loss": 1.5807,
+      "step": 2801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7759572605426057e-07,
+      "loss": 1.6695,
+      "step": 2802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7589548841716274e-07,
+      "loss": 1.6219,
+      "step": 2803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7420032591192856e-07,
+      "loss": 1.5958,
+      "step": 2804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7251024035968134e-07,
+      "loss": 1.5094,
+      "step": 2805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7082523357608856e-07,
+      "loss": 1.5602,
+      "step": 2806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6914530737136346e-07,
+      "loss": 1.687,
+      "step": 2807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.674704635502584e-07,
+      "loss": 1.5092,
+      "step": 2808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.658007039120697e-07,
+      "loss": 1.601,
+      "step": 2809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.64136030250628e-07,
+      "loss": 1.595,
+      "step": 2810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6247644435430263e-07,
+      "loss": 1.6401,
+      "step": 2811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6082194800599424e-07,
+      "loss": 1.5242,
+      "step": 2812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.591725429831382e-07,
+      "loss": 1.6529,
+      "step": 2813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.57528231057701e-07,
+      "loss": 1.6604,
+      "step": 2814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.558890139961745e-07,
+      "loss": 1.5549,
+      "step": 2815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5425489355957956e-07,
+      "loss": 1.5913,
+      "step": 2816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526258715034602e-07,
+      "loss": 1.6116,
+      "step": 2817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.510019495778837e-07,
+      "loss": 1.6026,
+      "step": 2818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4938312952744016e-07,
+      "loss": 1.5647,
+      "step": 2819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.477694130912356e-07,
+      "loss": 1.7237,
+      "step": 2820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.461608020028944e-07,
+      "loss": 1.6108,
+      "step": 2821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.445572979905575e-07,
+      "loss": 1.531,
+      "step": 2822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4295890277687695e-07,
+      "loss": 1.6327,
+      "step": 2823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4136561807901916e-07,
+      "loss": 1.5677,
+      "step": 2824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.397774456086577e-07,
+      "loss": 1.575,
+      "step": 2825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3819438707197495e-07,
+      "loss": 1.6094,
+      "step": 2826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3661644416966057e-07,
+      "loss": 1.6349,
+      "step": 2827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3504361859690628e-07,
+      "loss": 1.7109,
+      "step": 2828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3347591204340881e-07,
+      "loss": 1.5845,
+      "step": 2829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3191332619336204e-07,
+      "loss": 1.483,
+      "step": 2830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3035586272546207e-07,
+      "loss": 1.6487,
+      "step": 2831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2880352331290102e-07,
+      "loss": 1.8013,
+      "step": 2832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2725630962336542e-07,
+      "loss": 1.6861,
+      "step": 2833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2571422331903458e-07,
+      "loss": 1.5538,
+      "step": 2834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2417726605658164e-07,
+      "loss": 1.5135,
+      "step": 2835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.226454394871669e-07,
+      "loss": 1.5706,
+      "step": 2836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2111874525644228e-07,
+      "loss": 1.6044,
+      "step": 2837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1959718500454196e-07,
+      "loss": 1.6862,
+      "step": 2838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1808076036608783e-07,
+      "loss": 1.5705,
+      "step": 2839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.165694729701834e-07,
+      "loss": 1.6215,
+      "step": 2840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1506332444041212e-07,
+      "loss": 1.6152,
+      "step": 2841
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1356231639483917e-07,
+      "loss": 1.5072,
+      "step": 2842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1206645044600404e-07,
+      "loss": 1.4915,
+      "step": 2843
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1057572820092576e-07,
+      "loss": 1.6446,
+      "step": 2844
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0909015126109488e-07,
+      "loss": 1.4997,
+      "step": 2845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0760972122247425e-07,
+      "loss": 1.5518,
+      "step": 2846
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.061344396754994e-07,
+      "loss": 1.6089,
+      "step": 2847
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.04664308205072e-07,
+      "loss": 1.5849,
+      "step": 2848
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0319932839056365e-07,
+      "loss": 1.6406,
+      "step": 2849
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0173950180581047e-07,
+      "loss": 1.6606,
+      "step": 2850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.002848300191118e-07,
+      "loss": 1.4968,
+      "step": 2851
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.988353145932298e-07,
+      "loss": 1.5929,
+      "step": 2852
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9739095708538714e-07,
+      "loss": 1.6391,
+      "step": 2853
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9595175904726481e-07,
+      "loss": 1.6122,
+      "step": 2854
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9451772202500163e-07,
+      "loss": 1.649,
+      "step": 2855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9308884755919132e-07,
+      "loss": 1.5271,
+      "step": 2856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9166513718488155e-07,
+      "loss": 1.5406,
+      "step": 2857
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.902465924315733e-07,
+      "loss": 1.7513,
+      "step": 2858
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8883321482321583e-07,
+      "loss": 1.721,
+      "step": 2859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8742500587820955e-07,
+      "loss": 1.5806,
+      "step": 2860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.86021967109401e-07,
+      "loss": 1.712,
+      "step": 2861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8462410002408228e-07,
+      "loss": 1.6086,
+      "step": 2862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8323140612399038e-07,
+      "loss": 1.6559,
+      "step": 2863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8184388690530242e-07,
+      "loss": 1.6614,
+      "step": 2864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.804615438586399e-07,
+      "loss": 1.5504,
+      "step": 2865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7908437846906158e-07,
+      "loss": 1.4809,
+      "step": 2866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7771239221606285e-07,
+      "loss": 1.6391,
+      "step": 2867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7634558657357748e-07,
+      "loss": 1.6484,
+      "step": 2868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7498396300997146e-07,
+      "loss": 1.6642,
+      "step": 2869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.736275229880441e-07,
+      "loss": 1.5346,
+      "step": 2870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7227626796502807e-07,
+      "loss": 1.6927,
+      "step": 2871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7093019939258327e-07,
+      "loss": 1.6467,
+      "step": 2872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6958931871679908e-07,
+      "loss": 1.6254,
+      "step": 2873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6825362737818985e-07,
+      "loss": 1.4597,
+      "step": 2874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6692312681169775e-07,
+      "loss": 1.6517,
+      "step": 2875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6559781844668666e-07,
+      "loss": 1.4801,
+      "step": 2876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6427770370694208e-07,
+      "loss": 1.5652,
+      "step": 2877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6296278401067122e-07,
+      "loss": 1.6772,
+      "step": 2878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6165306077049969e-07,
+      "loss": 1.605,
+      "step": 2879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.603485353934703e-07,
+      "loss": 1.4958,
+      "step": 2880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5904920928104196e-07,
+      "loss": 1.6033,
+      "step": 2881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.577550838290881e-07,
+      "loss": 1.5824,
+      "step": 2882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.564661604278944e-07,
+      "loss": 1.5907,
+      "step": 2883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5518244046215936e-07,
+      "loss": 1.6151,
+      "step": 2884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.539039253109892e-07,
+      "loss": 1.5982,
+      "step": 2885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.526306163479019e-07,
+      "loss": 1.6821,
+      "step": 2886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5136251494081822e-07,
+      "loss": 1.5711,
+      "step": 2887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5009962245206845e-07,
+      "loss": 1.5268,
+      "step": 2888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.488419402383834e-07,
+      "loss": 1.5613,
+      "step": 2889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4758946965089894e-07,
+      "loss": 1.624,
+      "step": 2890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4634221203515097e-07,
+      "loss": 1.5199,
+      "step": 2891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4510016873107657e-07,
+      "loss": 1.7219,
+      "step": 2892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4386334107300727e-07,
+      "loss": 1.554,
+      "step": 2893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4263173038967627e-07,
+      "loss": 1.6243,
+      "step": 2894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4140533800420853e-07,
+      "loss": 1.543,
+      "step": 2895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.401841652341246e-07,
+      "loss": 1.586,
+      "step": 2896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.389682133913378e-07,
+      "loss": 1.6733,
+      "step": 2897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3775748378215047e-07,
+      "loss": 1.61,
+      "step": 2898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3655197770725826e-07,
+      "loss": 1.58,
+      "step": 2899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3535169646174073e-07,
+      "loss": 1.5769,
+      "step": 2900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3415664133506812e-07,
+      "loss": 1.5782,
+      "step": 2901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3296681361109564e-07,
+      "loss": 1.5633,
+      "step": 2902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3178221456806028e-07,
+      "loss": 1.5553,
+      "step": 2903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3060284547858403e-07,
+      "loss": 1.4216,
+      "step": 2904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2942870760966952e-07,
+      "loss": 1.6404,
+      "step": 2905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.282598022226994e-07,
+      "loss": 1.5952,
+      "step": 2906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.270961305734364e-07,
+      "loss": 1.6807,
+      "step": 2907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2593769391201827e-07,
+      "loss": 1.4806,
+      "step": 2908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.247844934829606e-07,
+      "loss": 1.5965,
+      "step": 2909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2363653052515302e-07,
+      "loss": 1.5156,
+      "step": 2910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2249380627185781e-07,
+      "loss": 1.6219,
+      "step": 2911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2135632195071133e-07,
+      "loss": 1.5642,
+      "step": 2912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.202240787837178e-07,
+      "loss": 1.6253,
+      "step": 2913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1909707798725412e-07,
+      "loss": 1.5733,
+      "step": 2914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1797532077206187e-07,
+      "loss": 1.6421,
+      "step": 2915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1685880834325203e-07,
+      "loss": 1.7266,
+      "step": 2916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1574754190030014e-07,
+      "loss": 1.6369,
+      "step": 2917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1464152263704565e-07,
+      "loss": 1.6558,
+      "step": 2918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1354075174169088e-07,
+      "loss": 1.4598,
+      "step": 2919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.12445230396801e-07,
+      "loss": 1.5789,
+      "step": 2920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1135495977930011e-07,
+      "loss": 1.5659,
+      "step": 2921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1026994106047296e-07,
+      "loss": 1.6185,
+      "step": 2922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0919017540595933e-07,
+      "loss": 1.563,
+      "step": 2923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0811566397575912e-07,
+      "loss": 1.6398,
+      "step": 2924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0704640792422616e-07,
+      "loss": 1.6309,
+      "step": 2925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0598240840006658e-07,
+      "loss": 1.6252,
+      "step": 2926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0492366654634211e-07,
+      "loss": 1.7421,
+      "step": 2927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0387018350046519e-07,
+      "loss": 1.5985,
+      "step": 2928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0282196039419823e-07,
+      "loss": 1.6024,
+      "step": 2929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0177899835365323e-07,
+      "loss": 1.6765,
+      "step": 2930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0074129849928948e-07,
+      "loss": 1.6289,
+      "step": 2931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.970886194591467e-08,
+      "loss": 1.6395,
+      "step": 2932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.8681689802681e-08,
+      "loss": 1.5107,
+      "step": 2933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.765978317308522e-08,
+      "loss": 1.5484,
+      "step": 2934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.664314315496692e-08,
+      "loss": 1.5871,
+      "step": 2935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.5631770840508e-08,
+      "loss": 1.6065,
+      "step": 2936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.462566731623213e-08,
+      "loss": 1.5573,
+      "step": 2937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.36248336630019e-08,
+      "loss": 1.6743,
+      "step": 2938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.262927095601782e-08,
+      "loss": 1.4791,
+      "step": 2939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.163898026481876e-08,
+      "loss": 1.5497,
+      "step": 2940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.065396265327986e-08,
+      "loss": 1.6349,
+      "step": 2941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.967421917961072e-08,
+      "loss": 1.6702,
+      "step": 2942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.869975089635552e-08,
+      "loss": 1.6,
+      "step": 2943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.773055885039072e-08,
+      "loss": 1.4078,
+      "step": 2944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.676664408292457e-08,
+      "loss": 1.5457,
+      "step": 2945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.580800762949704e-08,
+      "loss": 1.5749,
+      "step": 2946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.485465051997488e-08,
+      "loss": 1.6132,
+      "step": 2947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.39065737785566e-08,
+      "loss": 1.5086,
+      "step": 2948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.296377842376524e-08,
+      "loss": 1.6338,
+      "step": 2949
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.202626546845172e-08,
+      "loss": 1.7592,
+      "step": 2950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.109403591979148e-08,
+      "loss": 1.5978,
+      "step": 2951
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.016709077928397e-08,
+      "loss": 1.5556,
+      "step": 2952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.924543104275095e-08,
+      "loss": 1.6359,
+      "step": 2953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.832905770033705e-08,
+      "loss": 1.6312,
+      "step": 2954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.7417971736507e-08,
+      "loss": 1.525,
+      "step": 2955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.651217413004674e-08,
+      "loss": 1.5973,
+      "step": 2956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.561166585405789e-08,
+      "loss": 1.6548,
+      "step": 2957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.47164478759621e-08,
+      "loss": 1.6528,
+      "step": 2958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.382652115749789e-08,
+      "loss": 1.534,
+      "step": 2959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.294188665471769e-08,
+      "loss": 1.4825,
+      "step": 2960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.206254531799018e-08,
+      "loss": 1.6389,
+      "step": 2961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.118849809199524e-08,
+      "loss": 1.5478,
+      "step": 2962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.031974591572732e-08,
+      "loss": 1.7217,
+      "step": 2963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.945628972249208e-08,
+      "loss": 1.5816,
+      "step": 2964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.859813043990526e-08,
+      "loss": 1.5829,
+      "step": 2965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.77452689898922e-08,
+      "loss": 1.7169,
+      "step": 2966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.689770628868609e-08,
+      "loss": 1.5747,
+      "step": 2967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.605544324682855e-08,
+      "loss": 1.5229,
+      "step": 2968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.521848076916859e-08,
+      "loss": 1.5773,
+      "step": 2969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.438681975485805e-08,
+      "loss": 1.4672,
+      "step": 2970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.356046109735614e-08,
+      "loss": 1.5697,
+      "step": 2971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.273940568442327e-08,
+      "loss": 1.6527,
+      "step": 2972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.192365439812553e-08,
+      "loss": 1.6154,
+      "step": 2973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.111320811482802e-08,
+      "loss": 1.6042,
+      "step": 2974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.030806770519815e-08,
+      "loss": 1.5872,
+      "step": 2975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9508234034202364e-08,
+      "loss": 1.7257,
+      "step": 2976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.871370796110665e-08,
+      "loss": 1.7367,
+      "step": 2977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7924490339474335e-08,
+      "loss": 1.6378,
+      "step": 2978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7140582017167764e-08,
+      "loss": 1.6073,
+      "step": 2979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.636198383634217e-08,
+      "loss": 1.5944,
+      "step": 2980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.558869663345123e-08,
+      "loss": 1.5663,
+      "step": 2981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.482072123924098e-08,
+      "loss": 1.6185,
+      "step": 2982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.405805847875256e-08,
+      "loss": 1.6883,
+      "step": 2983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.330070917131724e-08,
+      "loss": 1.6305,
+      "step": 2984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2548674130561974e-08,
+      "loss": 1.6303,
+      "step": 2985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1801954164399925e-08,
+      "loss": 1.4522,
+      "step": 2986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.106055007503774e-08,
+      "loss": 1.5563,
+      "step": 2987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0324462658969395e-08,
+      "loss": 1.5076,
+      "step": 2988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.959369270697789e-08,
+      "loss": 1.6076,
+      "step": 2989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.886824100413412e-08,
+      "loss": 1.6773,
+      "step": 2990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.814810832979411e-08,
+      "loss": 1.6819,
+      "step": 2991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.743329545760122e-08,
+      "loss": 1.7171,
+      "step": 2992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.67238031554812e-08,
+      "loss": 1.6966,
+      "step": 2993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6019632185647645e-08,
+      "loss": 1.4728,
+      "step": 2994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.532078330459433e-08,
+      "loss": 1.5905,
+      "step": 2995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4627257263098465e-08,
+      "loss": 1.5935,
+      "step": 2996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.393905480621907e-08,
+      "loss": 1.5365,
+      "step": 2997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3256176673295846e-08,
+      "loss": 1.5305,
+      "step": 2998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2578623597949174e-08,
+      "loss": 1.6298,
+      "step": 2999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1906396308077356e-08,
+      "loss": 1.5961,
+      "step": 3000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.123949552585826e-08,
+      "loss": 1.7157,
+      "step": 3001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0577921967747126e-08,
+      "loss": 1.5694,
+      "step": 3002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9921676344475966e-08,
+      "loss": 1.6027,
+      "step": 3003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.927075936105307e-08,
+      "loss": 1.648,
+      "step": 3004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8625171716762385e-08,
+      "loss": 1.5401,
+      "step": 3005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7984914105162474e-08,
+      "loss": 1.6722,
+      "step": 3006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7349987214084784e-08,
+      "loss": 1.5512,
+      "step": 3007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.672039172563646e-08,
+      "loss": 1.5535,
+      "step": 3008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.609612831619369e-08,
+      "loss": 1.622,
+      "step": 3009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.547719765640778e-08,
+      "loss": 1.5579,
+      "step": 3010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4863600411197404e-08,
+      "loss": 1.6422,
+      "step": 3011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.425533723975527e-08,
+      "loss": 1.6563,
+      "step": 3012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.365240879554144e-08,
+      "loss": 1.6367,
+      "step": 3013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3054815726285e-08,
+      "loss": 1.5992,
+      "step": 3014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2462558673983516e-08,
+      "loss": 1.6834,
+      "step": 3015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1875638274902476e-08,
+      "loss": 1.5542,
+      "step": 3016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.129405515957307e-08,
+      "loss": 1.7154,
+      "step": 3017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.071780995279439e-08,
+      "loss": 1.5382,
+      "step": 3018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.014690327362846e-08,
+      "loss": 1.6584,
+      "step": 3019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9581335735404672e-08,
+      "loss": 1.5841,
+      "step": 3020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9021107945714777e-08,
+      "loss": 1.7803,
+      "step": 3021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8466220506414565e-08,
+      "loss": 1.6609,
+      "step": 3022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.79166740136233e-08,
+      "loss": 1.4741,
+      "step": 3023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7372469057721506e-08,
+      "loss": 1.6577,
+      "step": 3024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6833606223351515e-08,
+      "loss": 1.5909,
+      "step": 3025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6300086089416366e-08,
+      "loss": 1.5308,
+      "step": 3026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.577190922908035e-08,
+      "loss": 1.6303,
+      "step": 3027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5249076209767353e-08,
+      "loss": 1.7103,
+      "step": 3028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.473158759315808e-08,
+      "loss": 1.6372,
+      "step": 3029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.421944393519504e-08,
+      "loss": 1.7447,
+      "step": 3030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3712645786075905e-08,
+      "loss": 1.685,
+      "step": 3031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3211193690257373e-08,
+      "loss": 1.6784,
+      "step": 3032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.271508818645185e-08,
+      "loss": 1.6198,
+      "step": 3033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.222432980762912e-08,
+      "loss": 1.549,
+      "step": 3034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1738919081012446e-08,
+      "loss": 1.5382,
+      "step": 3035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1258856528081906e-08,
+      "loss": 1.6794,
+      "step": 3036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0784142664571626e-08,
+      "loss": 1.6062,
+      "step": 3037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.031477800046866e-08,
+      "loss": 1.5514,
+      "step": 3038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9850763040014654e-08,
+      "loss": 1.5991,
+      "step": 3039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.939209828170363e-08,
+      "loss": 1.6843,
+      "step": 3040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8938784218281435e-08,
+      "loss": 1.6064,
+      "step": 3041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.849082133674518e-08,
+      "loss": 1.5597,
+      "step": 3042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.80482101183449e-08,
+      "loss": 1.6229,
+      "step": 3043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.761095103858024e-08,
+      "loss": 1.6098,
+      "step": 3044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.717904456720043e-08,
+      "loss": 1.5951,
+      "step": 3045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.675249116820543e-08,
+      "loss": 1.628,
+      "step": 3046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6331291299844233e-08,
+      "loss": 1.6298,
+      "step": 3047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5915445414613208e-08,
+      "loss": 1.6225,
+      "step": 3048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.550495395925944e-08,
+      "loss": 1.5953,
+      "step": 3049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5099817374774615e-08,
+      "loss": 1.5108,
+      "step": 3050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4700036096400028e-08,
+      "loss": 1.5613,
+      "step": 3051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4305610553623228e-08,
+      "loss": 1.5689,
+      "step": 3052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3916541170176934e-08,
+      "loss": 1.6215,
+      "step": 3053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3532828364041239e-08,
+      "loss": 1.5249,
+      "step": 3054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3154472547440289e-08,
+      "loss": 1.6395,
+      "step": 3055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2781474126845051e-08,
+      "loss": 1.6133,
+      "step": 3056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.241383350296832e-08,
+      "loss": 1.667,
+      "step": 3057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2051551070769719e-08,
+      "loss": 1.5943,
+      "step": 3058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1694627219450694e-08,
+      "loss": 1.6966,
+      "step": 3059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.134306233245619e-08,
+      "loss": 1.7124,
+      "step": 3060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0996856787475197e-08,
+      "loss": 1.6455,
+      "step": 3061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0656010956437979e-08,
+      "loss": 1.541,
+      "step": 3062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0320525205516629e-08,
+      "loss": 1.6038,
+      "step": 3063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.990399895125624e-09,
+      "loss": 1.5748,
+      "step": 3064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.665635379920157e-09,
+      "loss": 1.6786,
+      "step": 3065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.346232008797252e-09,
+      "loss": 1.6463,
+      "step": 3066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.032190124893536e-09,
+      "loss": 1.5548,
+      "step": 3067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.723510065585806e-09,
+      "loss": 1.6646,
+      "step": 3068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.42019216249046e-09,
+      "loss": 1.5409,
+      "step": 3069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.122236741464618e-09,
+      "loss": 1.5699,
+      "step": 3070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.82964412260223e-09,
+      "loss": 1.6786,
+      "step": 3071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.542414620237414e-09,
+      "loss": 1.6302,
+      "step": 3072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.260548542943335e-09,
+      "loss": 1.5449,
+      "step": 3073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.984046193528881e-09,
+      "loss": 1.6334,
+      "step": 3074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.712907869043661e-09,
+      "loss": 1.5267,
+      "step": 3075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.447133860771893e-09,
+      "loss": 1.5476,
+      "step": 3076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.186724454236847e-09,
+      "loss": 1.4951,
+      "step": 3077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9316799291969654e-09,
+      "loss": 1.6215,
+      "step": 3078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.682000559649181e-09,
+      "loss": 1.6867,
+      "step": 3079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.437686613823934e-09,
+      "loss": 1.5682,
+      "step": 3080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.198738354190158e-09,
+      "loss": 1.6384,
+      "step": 3081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9651560374514015e-09,
+      "loss": 1.4541,
+      "step": 3082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.736939914545824e-09,
+      "loss": 1.7202,
+      "step": 3083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.514090230647305e-09,
+      "loss": 1.5816,
+      "step": 3084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.296607225164895e-09,
+      "loss": 1.6952,
+      "step": 3085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.084491131741697e-09,
+      "loss": 1.5358,
+      "step": 3086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.877742178254873e-09,
+      "loss": 1.5787,
+      "step": 3087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6763605868167516e-09,
+      "loss": 1.5623,
+      "step": 3088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4803465737714983e-09,
+      "loss": 1.5523,
+      "step": 3089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.289700349698999e-09,
+      "loss": 1.4609,
+      "step": 3090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.104422119411532e-09,
+      "loss": 1.6859,
+      "step": 3091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9245120819543226e-09,
+      "loss": 1.6109,
+      "step": 3092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.749970430605542e-09,
+      "loss": 1.458,
+      "step": 3093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5807973528768626e-09,
+      "loss": 1.7146,
+      "step": 3094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.416993030511239e-09,
+      "loss": 1.5739,
+      "step": 3095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.258557639484571e-09,
+      "loss": 1.646,
+      "step": 3096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1054913500051512e-09,
+      "loss": 1.6871,
+      "step": 3097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.957794326513107e-09,
+      "loss": 1.6995,
+      "step": 3098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8154667276798488e-09,
+      "loss": 1.5433,
+      "step": 3099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6785087064086213e-09,
+      "loss": 1.5667,
+      "step": 3100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.546920409834507e-09,
+      "loss": 1.5896,
+      "step": 3101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4207019793238686e-09,
+      "loss": 1.7313,
+      "step": 3102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.299853550472685e-09,
+      "loss": 1.5189,
+      "step": 3103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1843752531104368e-09,
+      "loss": 1.5681,
+      "step": 3104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0742672112951103e-09,
+      "loss": 1.5904,
+      "step": 3105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.695295433170826e-10,
+      "loss": 1.5521,
+      "step": 3106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.701623616963472e-10,
+      "loss": 1.6201,
+      "step": 3107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.761657731836236e-10,
+      "loss": 1.6513,
+      "step": 3108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.87539878759802e-10,
+      "loss": 1.6488,
+      "step": 3109
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.042847736364987e-10,
+      "loss": 1.5999,
+      "step": 3110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.264005472549461e-10,
+      "loss": 1.5956,
+      "step": 3111
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.538872832865471e-10,
+      "loss": 1.7087,
+      "step": 3112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.867450596328759e-10,
+      "loss": 1.5474,
+      "step": 3113
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2497394842512244e-10,
+      "loss": 1.577,
+      "step": 3114
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.685740160240924e-10,
+      "loss": 1.7282,
+      "step": 3115
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1754532302076247e-10,
+      "loss": 1.546,
+      "step": 3116
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.718879242357252e-10,
+      "loss": 1.506,
+      "step": 3117
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.316018687191889e-10,
+      "loss": 1.5845,
+      "step": 3118
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.668719974986751e-11,
+      "loss": 1.4207,
+      "step": 3119
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.714395483720105e-11,
+      "loss": 1.5434,
+      "step": 3120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.297216571969021e-11,
+      "loss": 1.5558,
+      "step": 3121
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.417185836545155e-11,
+      "loss": 1.6485,
+      "step": 3122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0743052971107225e-11,
+      "loss": 1.4902,
+      "step": 3123
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6857639640054387e-12,
+      "loss": 1.6521,
+      "step": 3124
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.6939,
+      "step": 3125
+    },
+    {
+      "epoch": 1.0,
+      "step": 3125,
+      "total_flos": 5.465597220361339e+17,
+      "train_loss": 1.653589628982544,
+      "train_runtime": 81979.5198,
+      "train_samples_per_second": 1.22,
+      "train_steps_per_second": 0.038
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3125,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 5.465597220361339e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}